diff --git a/clang/include/clang/Basic/riscv_vector.td b/clang/include/clang/Basic/riscv_vector.td
--- a/clang/include/clang/Basic/riscv_vector.td
+++ b/clang/include/clang/Basic/riscv_vector.td
@@ -224,6 +224,11 @@
                                       [["vv", "w", "wwvv"],
                                        ["vf", "w", "wwev"]]>;
   }
+  multiclass RVVFloatingWidenTerBuiltinSetRoundingMode {
+    defm ""  : RVVOutOp1Op2BuiltinSet<NAME, "xf",
+                                      [["vv", "w", "wwvvu"],
+                                       ["vf", "w", "wwevu"]]>;
+  }
 }
 
 multiclass RVVFloatingBinBuiltinSet
@@ -2105,11 +2110,57 @@
   defm vfnmsub : RVVFloatingTerBuiltinSet;
 }
 
-// 14.7. Vector Widening Floating-Point Fused Multiply-Add Instructions
-defm vfwmacc  : RVVFloatingWidenTerBuiltinSet;
-defm vfwnmacc : RVVFloatingWidenTerBuiltinSet;
-defm vfwmsac  : RVVFloatingWidenTerBuiltinSet;
-defm vfwnmsac : RVVFloatingWidenTerBuiltinSet;
+let ManualCodegen = [{
+  {
+    // LLVM intrinsic
+    // Unmasked: (passthru, op0, op1, round_mode, vl)
+    // Masked:   (passthru, vector_in, vector_in/scalar_in, mask, frm, vl, policy)
+
+    SmallVector<llvm::Value*, 7> Operands;
+    bool HasRoundModeOp = IsMasked ? Ops.size() == 6 : Ops.size() == 5;
+
+    unsigned Offset = IsMasked ? 2 : 1;
+
+    Operands.push_back(Ops[IsMasked ? 1 : 0]); // passthrough
+
+    Operands.push_back(Ops[Offset]); // op0
+    Operands.push_back(Ops[Offset + 1]); // op1
+
+    if (IsMasked)
+      Operands.push_back(Ops[0]); // mask
+
+    if (HasRoundModeOp) {
+      Operands.push_back(Ops[Offset + 2]); // frm
+      Operands.push_back(Ops[Offset + 3]); // vl
+    } else {
+      Operands.push_back(ConstantInt::get(Ops[Offset + 2]->getType(), 7)); // frm
+      Operands.push_back(Ops[Offset + 2]); // vl
+    }
+
+    Operands.push_back(ConstantInt::get(Ops.back()->getType(), PolicyAttrs));
+
+    IntrinsicTypes = {ResultType, Ops[Offset]->getType(), Ops[Offset + 1]->getType(),
+                      Operands.back()->getType()};
+
+    llvm::Function *F = CGM.getIntrinsic(ID, IntrinsicTypes);
+
+    return Builder.CreateCall(F, Operands, "");
+  }
+}] in {
+  let HasFRMRoundModeOp = 1 in {
+    // 14.7. Vector Widening Floating-Point Fused Multiply-Add Instructions
+    defm vfwmacc  : RVVFloatingWidenTerBuiltinSetRoundingMode;
+    defm vfwnmacc : RVVFloatingWidenTerBuiltinSetRoundingMode;
+    defm vfwmsac  : RVVFloatingWidenTerBuiltinSetRoundingMode;
+    defm vfwnmsac : RVVFloatingWidenTerBuiltinSetRoundingMode;
+  }
+  // 14.7. Vector Widening Floating-Point Fused Multiply-Add Instructions
+  defm vfwmacc  : RVVFloatingWidenTerBuiltinSet;
+  defm vfwnmacc : RVVFloatingWidenTerBuiltinSet;
+  defm vfwmsac  : RVVFloatingWidenTerBuiltinSet;
+  defm vfwnmsac : RVVFloatingWidenTerBuiltinSet;
+}
+
 }
 
 // 14.8. Vector Floating-Point Square-Root Instruction
diff --git a/clang/lib/Sema/SemaChecking.cpp b/clang/lib/Sema/SemaChecking.cpp
--- a/clang/lib/Sema/SemaChecking.cpp
+++ b/clang/lib/Sema/SemaChecking.cpp
@@ -4865,6 +4865,14 @@
   case RISCVVector::BI__builtin_rvv_vfmsub_vf_rm:
   case RISCVVector::BI__builtin_rvv_vfnmsub_vv_rm:
   case RISCVVector::BI__builtin_rvv_vfnmsub_vf_rm:
+  case RISCVVector::BI__builtin_rvv_vfwmacc_vv_rm:
+  case RISCVVector::BI__builtin_rvv_vfwmacc_vf_rm:
+  case RISCVVector::BI__builtin_rvv_vfwnmacc_vv_rm:
+  case RISCVVector::BI__builtin_rvv_vfwnmacc_vf_rm:
+  case RISCVVector::BI__builtin_rvv_vfwmsac_vv_rm:
+  case RISCVVector::BI__builtin_rvv_vfwmsac_vf_rm:
+  case RISCVVector::BI__builtin_rvv_vfwnmsac_vv_rm:
+  case RISCVVector::BI__builtin_rvv_vfwnmsac_vf_rm:
   case RISCVVector::BI__builtin_rvv_vfmacc_vv_rm_tu:
   case RISCVVector::BI__builtin_rvv_vfmacc_vf_rm_tu:
   case RISCVVector::BI__builtin_rvv_vfnmacc_vv_rm_tu:
@@ -4881,6 +4889,14 @@
   case RISCVVector::BI__builtin_rvv_vfmsub_vf_rm_tu:
   case RISCVVector::BI__builtin_rvv_vfnmsub_vv_rm_tu:
   case RISCVVector::BI__builtin_rvv_vfnmsub_vf_rm_tu:
+  case RISCVVector::BI__builtin_rvv_vfwmacc_vv_rm_tu:
+  case RISCVVector::BI__builtin_rvv_vfwmacc_vf_rm_tu:
+  case RISCVVector::BI__builtin_rvv_vfwnmacc_vv_rm_tu:
+  case RISCVVector::BI__builtin_rvv_vfwnmacc_vf_rm_tu:
+  case RISCVVector::BI__builtin_rvv_vfwmsac_vv_rm_tu:
+  case RISCVVector::BI__builtin_rvv_vfwmsac_vf_rm_tu:
+  case RISCVVector::BI__builtin_rvv_vfwnmsac_vv_rm_tu:
+  case RISCVVector::BI__builtin_rvv_vfwnmsac_vf_rm_tu:
   case RISCVVector::BI__builtin_rvv_vfadd_vv_rm_tama:
   case RISCVVector::BI__builtin_rvv_vfadd_vf_rm_tama:
   case RISCVVector::BI__builtin_rvv_vfsub_vv_rm_tama:
@@ -4918,6 +4934,14 @@
   case RISCVVector::BI__builtin_rvv_vfmsub_vf_rm_tama:
   case RISCVVector::BI__builtin_rvv_vfnmsub_vv_rm_tama:
   case RISCVVector::BI__builtin_rvv_vfnmsub_vf_rm_tama:
+  case RISCVVector::BI__builtin_rvv_vfwmacc_vv_rm_tama:
+  case RISCVVector::BI__builtin_rvv_vfwmacc_vf_rm_tama:
+  case RISCVVector::BI__builtin_rvv_vfwnmacc_vv_rm_tama:
+  case RISCVVector::BI__builtin_rvv_vfwnmacc_vf_rm_tama:
+  case RISCVVector::BI__builtin_rvv_vfwmsac_vv_rm_tama:
+  case RISCVVector::BI__builtin_rvv_vfwmsac_vf_rm_tama:
+  case RISCVVector::BI__builtin_rvv_vfwnmsac_vv_rm_tama:
+  case RISCVVector::BI__builtin_rvv_vfwnmsac_vf_rm_tama:
   case RISCVVector::BI__builtin_rvv_vfadd_vv_rm_tum:
   case RISCVVector::BI__builtin_rvv_vfadd_vf_rm_tum:
   case RISCVVector::BI__builtin_rvv_vfsub_vv_rm_tum:
@@ -4954,6 +4978,14 @@
   case RISCVVector::BI__builtin_rvv_vfmsub_vf_rm_tum:
   case RISCVVector::BI__builtin_rvv_vfnmsub_vv_rm_tum:
   case RISCVVector::BI__builtin_rvv_vfnmsub_vf_rm_tum:
+  case RISCVVector::BI__builtin_rvv_vfwmacc_vv_rm_tum:
+  case RISCVVector::BI__builtin_rvv_vfwmacc_vf_rm_tum:
+  case RISCVVector::BI__builtin_rvv_vfwnmacc_vv_rm_tum:
+  case RISCVVector::BI__builtin_rvv_vfwnmacc_vf_rm_tum:
+  case RISCVVector::BI__builtin_rvv_vfwmsac_vv_rm_tum:
+  case RISCVVector::BI__builtin_rvv_vfwmsac_vf_rm_tum:
+  case RISCVVector::BI__builtin_rvv_vfwnmsac_vv_rm_tum:
+  case RISCVVector::BI__builtin_rvv_vfwnmsac_vf_rm_tum:
   case RISCVVector::BI__builtin_rvv_vfadd_vv_rm_tumu:
   case RISCVVector::BI__builtin_rvv_vfadd_vf_rm_tumu:
   case RISCVVector::BI__builtin_rvv_vfsub_vv_rm_tumu:
@@ -4990,6 +5022,14 @@
   case RISCVVector::BI__builtin_rvv_vfmsub_vf_rm_tumu:
   case RISCVVector::BI__builtin_rvv_vfnmsub_vv_rm_tumu:
   case RISCVVector::BI__builtin_rvv_vfnmsub_vf_rm_tumu:
+  case RISCVVector::BI__builtin_rvv_vfwmacc_vv_rm_tumu:
+  case RISCVVector::BI__builtin_rvv_vfwmacc_vf_rm_tumu:
+  case RISCVVector::BI__builtin_rvv_vfwnmacc_vv_rm_tumu:
+  case RISCVVector::BI__builtin_rvv_vfwnmacc_vf_rm_tumu:
+  case RISCVVector::BI__builtin_rvv_vfwmsac_vv_rm_tumu:
+  case RISCVVector::BI__builtin_rvv_vfwmsac_vf_rm_tumu:
+  case RISCVVector::BI__builtin_rvv_vfwnmsac_vv_rm_tumu:
+  case RISCVVector::BI__builtin_rvv_vfwnmsac_vf_rm_tumu:
   case RISCVVector::BI__builtin_rvv_vfadd_vv_rm_mu:
   case RISCVVector::BI__builtin_rvv_vfadd_vf_rm_mu:
   case RISCVVector::BI__builtin_rvv_vfsub_vv_rm_mu:
@@ -5026,6 +5066,14 @@
   case RISCVVector::BI__builtin_rvv_vfmsub_vf_rm_mu:
   case RISCVVector::BI__builtin_rvv_vfnmsub_vv_rm_mu:
   case RISCVVector::BI__builtin_rvv_vfnmsub_vf_rm_mu:
+  case RISCVVector::BI__builtin_rvv_vfwmacc_vv_rm_mu:
+  case RISCVVector::BI__builtin_rvv_vfwmacc_vf_rm_mu:
+  case RISCVVector::BI__builtin_rvv_vfwnmacc_vv_rm_mu:
+  case RISCVVector::BI__builtin_rvv_vfwnmacc_vf_rm_mu:
+  case RISCVVector::BI__builtin_rvv_vfwmsac_vv_rm_mu:
+  case RISCVVector::BI__builtin_rvv_vfwmsac_vf_rm_mu:
+  case RISCVVector::BI__builtin_rvv_vfwnmsac_vv_rm_mu:
+  case RISCVVector::BI__builtin_rvv_vfwnmsac_vf_rm_mu:
     return SemaBuiltinConstantArgRange(TheCall, 4, 0, 4);
   case RISCV::BI__builtin_riscv_ntl_load:
   case RISCV::BI__builtin_riscv_ntl_store:
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/vfwmacc.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/vfwmacc.c
--- a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/vfwmacc.c
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/vfwmacc.c
@@ -10,7 +10,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_f32mf2
 // CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmacc_vv_f32mf2(vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -20,7 +20,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_f32mf2
 // CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmacc_vf_f32mf2(vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -30,7 +30,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_f32m1
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmacc_vv_f32m1(vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -40,7 +40,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_f32m1
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmacc_vf_f32m1(vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -50,7 +50,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_f32m2
 // CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmacc_vv_f32m2(vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -60,7 +60,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_f32m2
 // CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmacc_vf_f32m2(vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -70,7 +70,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_f32m4
 // CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmacc_vv_f32m4(vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -80,7 +80,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_f32m4
 // CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmacc_vf_f32m4(vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -90,7 +90,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_f32m8
 // CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmacc_vv_f32m8(vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -100,7 +100,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_f32m8
 // CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmacc_vf_f32m8(vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -110,7 +110,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vv_f64m1
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmacc_vv_f64m1(vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -120,7 +120,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vf_f64m1
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmacc_vf_f64m1(vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -130,7 +130,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vv_f64m2
 // CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmacc_vv_f64m2(vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -140,7 +140,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vf_f64m2
 // CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmacc_vf_f64m2(vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -150,7 +150,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vv_f64m4
 // CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmacc_vv_f64m4(vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -160,7 +160,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vf_f64m4
 // CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmacc_vf_f64m4(vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -170,7 +170,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vv_f64m8
 // CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmacc_vv_f64m8(vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -180,7 +180,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vf_f64m8
 // CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmacc_vf_f64m8(vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
@@ -190,7 +190,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_f32mf2_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmacc_vv_f32mf2_m(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -200,7 +200,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_f32mf2_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmacc_vf_f32mf2_m(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -210,7 +210,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmacc_vv_f32m1_m(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -220,7 +220,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmacc_vf_f32m1_m(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -230,7 +230,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_f32m2_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmacc_vv_f32m2_m(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -240,7 +240,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_f32m2_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmacc_vf_f32m2_m(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -250,7 +250,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_f32m4_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmacc_vv_f32m4_m(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -260,7 +260,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_f32m4_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmacc_vf_f32m4_m(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -270,7 +270,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_f32m8_m
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmacc_vv_f32m8_m(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -280,7 +280,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_f32m8_m
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmacc_vf_f32m8_m(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -290,7 +290,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vv_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmacc_vv_f64m1_m(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -300,7 +300,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vf_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmacc_vf_f64m1_m(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -310,7 +310,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vv_f64m2_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmacc_vv_f64m2_m(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -320,7 +320,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vf_f64m2_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmacc_vf_f64m2_m(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -330,7 +330,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vv_f64m4_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmacc_vv_f64m4_m(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -340,7 +340,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vf_f64m4_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmacc_vf_f64m4_m(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -350,7 +350,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vv_f64m8_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmacc_vv_f64m8_m(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -360,10 +360,370 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vf_f64m8_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmacc_vf_f64m8_m(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
   return __riscv_vfwmacc_vf_f64m8_m(mask, vd, vs1, vs2, vl);
 }
 
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_f32mf2_rm
+// CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vv_f32mf2_rm(vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f32mf2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_f32mf2_rm
+// CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vf_f32mf2_rm(vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f32mf2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vv_f32m1_rm(vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f32m1_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vf_f32m1_rm(vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f32m1_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_f32m2_rm
+// CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vv_f32m2_rm(vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f32m2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_f32m2_rm
+// CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vf_f32m2_rm(vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f32m2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_f32m4_rm
+// CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vv_f32m4_rm(vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f32m4_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_f32m4_rm
+// CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vf_f32m4_rm(vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f32m4_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_f32m8_rm
+// CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vv_f32m8_rm(vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f32m8_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_f32m8_rm
+// CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vf_f32m8_rm(vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f32m8_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vv_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmacc_vv_f64m1_rm(vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f64m1_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vf_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmacc_vf_f64m1_rm(vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f64m1_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vv_f64m2_rm
+// CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmacc_vv_f64m2_rm(vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f64m2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vf_f64m2_rm
+// CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmacc_vf_f64m2_rm(vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f64m2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vv_f64m4_rm
+// CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmacc_vv_f64m4_rm(vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f64m4_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vf_f64m4_rm
+// CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmacc_vf_f64m4_rm(vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f64m4_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vv_f64m8_rm
+// CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmacc_vv_f64m8_rm(vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f64m8_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vf_f64m8_rm
+// CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmacc_vf_f64m8_rm(vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f64m8_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_f32mf2_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vv_f32mf2_rm_m(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f32mf2_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_f32mf2_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vf_f32mf2_rm_m(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f32mf2_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vv_f32m1_rm_m(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f32m1_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vf_f32m1_rm_m(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f32m1_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_f32m2_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vv_f32m2_rm_m(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f32m2_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_f32m2_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vf_f32m2_rm_m(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f32m2_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_f32m4_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vv_f32m4_rm_m(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f32m4_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_f32m4_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vf_f32m4_rm_m(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f32m4_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_f32m8_rm_m
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vv_f32m8_rm_m(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f32m8_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_f32m8_rm_m
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vf_f32m8_rm_m(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f32m8_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vv_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmacc_vv_f64m1_rm_m(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f64m1_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vf_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmacc_vf_f64m1_rm_m(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f64m1_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vv_f64m2_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmacc_vv_f64m2_rm_m(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f64m2_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vf_f64m2_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmacc_vf_f64m2_rm_m(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f64m2_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vv_f64m4_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmacc_vv_f64m4_rm_m(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f64m4_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vf_f64m4_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmacc_vf_f64m4_rm_m(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f64m4_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vv_f64m8_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmacc_vv_f64m8_rm_m(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f64m8_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vf_f64m8_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmacc_vf_f64m8_rm_m(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f64m8_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/vfwmsac.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/vfwmsac.c
--- a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/vfwmsac.c
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/vfwmsac.c
@@ -10,7 +10,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_f32mf2
 // CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmsac_vv_f32mf2(vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -20,7 +20,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_f32mf2
 // CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmsac_vf_f32mf2(vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -30,7 +30,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_f32m1
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmsac_vv_f32m1(vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -40,7 +40,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_f32m1
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmsac_vf_f32m1(vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -50,7 +50,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_f32m2
 // CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmsac_vv_f32m2(vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -60,7 +60,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_f32m2
 // CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmsac_vf_f32m2(vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -70,7 +70,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_f32m4
 // CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmsac_vv_f32m4(vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -80,7 +80,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_f32m4
 // CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmsac_vf_f32m4(vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -90,7 +90,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_f32m8
 // CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmsac_vv_f32m8(vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -100,7 +100,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_f32m8
 // CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmsac_vf_f32m8(vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -110,7 +110,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vv_f64m1
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmsac_vv_f64m1(vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -120,7 +120,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vf_f64m1
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmsac_vf_f64m1(vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -130,7 +130,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vv_f64m2
 // CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmsac_vv_f64m2(vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -140,7 +140,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vf_f64m2
 // CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmsac_vf_f64m2(vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -150,7 +150,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vv_f64m4
 // CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmsac_vv_f64m4(vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -160,7 +160,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vf_f64m4
 // CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmsac_vf_f64m4(vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -170,7 +170,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vv_f64m8
 // CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmsac_vv_f64m8(vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -180,7 +180,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vf_f64m8
 // CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmsac_vf_f64m8(vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
@@ -190,7 +190,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_f32mf2_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmsac_vv_f32mf2_m(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -200,7 +200,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_f32mf2_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmsac_vf_f32mf2_m(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -210,7 +210,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmsac_vv_f32m1_m(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -220,7 +220,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmsac_vf_f32m1_m(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -230,7 +230,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_f32m2_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmsac_vv_f32m2_m(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -240,7 +240,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_f32m2_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmsac_vf_f32m2_m(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -250,7 +250,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_f32m4_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmsac_vv_f32m4_m(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -260,7 +260,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_f32m4_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmsac_vf_f32m4_m(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -270,7 +270,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_f32m8_m
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmsac_vv_f32m8_m(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -280,7 +280,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_f32m8_m
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmsac_vf_f32m8_m(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -290,7 +290,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vv_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmsac_vv_f64m1_m(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -300,7 +300,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vf_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmsac_vf_f64m1_m(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -310,7 +310,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vv_f64m2_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmsac_vv_f64m2_m(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -320,7 +320,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vf_f64m2_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmsac_vf_f64m2_m(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -330,7 +330,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vv_f64m4_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmsac_vv_f64m4_m(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -340,7 +340,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vf_f64m4_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmsac_vf_f64m4_m(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -350,7 +350,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vv_f64m8_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmsac_vv_f64m8_m(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -360,10 +360,370 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vf_f64m8_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmsac_vf_f64m8_m(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
   return __riscv_vfwmsac_vf_f64m8_m(mask, vd, vs1, vs2, vl);
 }
 
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_f32mf2_rm
+// CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vv_f32mf2_rm(vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f32mf2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_f32mf2_rm
+// CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vf_f32mf2_rm(vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f32mf2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vv_f32m1_rm(vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f32m1_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vf_f32m1_rm(vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f32m1_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_f32m2_rm
+// CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vv_f32m2_rm(vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f32m2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_f32m2_rm
+// CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vf_f32m2_rm(vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f32m2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_f32m4_rm
+// CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vv_f32m4_rm(vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f32m4_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_f32m4_rm
+// CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vf_f32m4_rm(vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f32m4_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_f32m8_rm
+// CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vv_f32m8_rm(vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f32m8_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_f32m8_rm
+// CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vf_f32m8_rm(vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f32m8_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vv_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmsac_vv_f64m1_rm(vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f64m1_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vf_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmsac_vf_f64m1_rm(vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f64m1_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vv_f64m2_rm
+// CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmsac_vv_f64m2_rm(vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f64m2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vf_f64m2_rm
+// CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmsac_vf_f64m2_rm(vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f64m2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vv_f64m4_rm
+// CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmsac_vv_f64m4_rm(vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f64m4_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vf_f64m4_rm
+// CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmsac_vf_f64m4_rm(vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f64m4_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vv_f64m8_rm
+// CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmsac_vv_f64m8_rm(vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f64m8_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vf_f64m8_rm
+// CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmsac_vf_f64m8_rm(vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f64m8_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_f32mf2_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vv_f32mf2_rm_m(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f32mf2_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_f32mf2_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vf_f32mf2_rm_m(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f32mf2_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vv_f32m1_rm_m(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f32m1_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vf_f32m1_rm_m(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f32m1_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_f32m2_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vv_f32m2_rm_m(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f32m2_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_f32m2_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vf_f32m2_rm_m(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f32m2_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_f32m4_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vv_f32m4_rm_m(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f32m4_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_f32m4_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vf_f32m4_rm_m(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f32m4_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_f32m8_rm_m
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vv_f32m8_rm_m(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f32m8_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_f32m8_rm_m
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vf_f32m8_rm_m(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f32m8_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vv_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmsac_vv_f64m1_rm_m(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f64m1_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vf_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmsac_vf_f64m1_rm_m(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f64m1_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vv_f64m2_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmsac_vv_f64m2_rm_m(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f64m2_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vf_f64m2_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmsac_vf_f64m2_rm_m(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f64m2_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vv_f64m4_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmsac_vv_f64m4_rm_m(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f64m4_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vf_f64m4_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmsac_vf_f64m4_rm_m(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f64m4_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vv_f64m8_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmsac_vv_f64m8_rm_m(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f64m8_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vf_f64m8_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmsac_vf_f64m8_rm_m(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f64m8_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/vfwnmacc.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/vfwnmacc.c
--- a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/vfwnmacc.c
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/vfwnmacc.c
@@ -10,7 +10,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_f32mf2
 // CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmacc_vv_f32mf2(vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -20,7 +20,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_f32mf2
 // CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmacc_vf_f32mf2(vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -30,7 +30,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_f32m1
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmacc_vv_f32m1(vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -40,7 +40,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_f32m1
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmacc_vf_f32m1(vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -50,7 +50,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_f32m2
 // CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmacc_vv_f32m2(vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -60,7 +60,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_f32m2
 // CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmacc_vf_f32m2(vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -70,7 +70,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_f32m4
 // CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmacc_vv_f32m4(vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -80,7 +80,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_f32m4
 // CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmacc_vf_f32m4(vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -90,7 +90,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_f32m8
 // CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmacc_vv_f32m8(vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -100,7 +100,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_f32m8
 // CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmacc_vf_f32m8(vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -110,7 +110,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vv_f64m1
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmacc_vv_f64m1(vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -120,7 +120,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vf_f64m1
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmacc_vf_f64m1(vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -130,7 +130,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vv_f64m2
 // CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmacc_vv_f64m2(vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -140,7 +140,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vf_f64m2
 // CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmacc_vf_f64m2(vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -150,7 +150,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vv_f64m4
 // CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmacc_vv_f64m4(vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -160,7 +160,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vf_f64m4
 // CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmacc_vf_f64m4(vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -170,7 +170,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vv_f64m8
 // CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmacc_vv_f64m8(vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -180,7 +180,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vf_f64m8
 // CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmacc_vf_f64m8(vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
@@ -190,7 +190,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_f32mf2_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmacc_vv_f32mf2_m(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -200,7 +200,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_f32mf2_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmacc_vf_f32mf2_m(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -210,7 +210,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmacc_vv_f32m1_m(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -220,7 +220,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmacc_vf_f32m1_m(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -230,7 +230,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_f32m2_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmacc_vv_f32m2_m(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -240,7 +240,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_f32m2_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmacc_vf_f32m2_m(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -250,7 +250,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_f32m4_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmacc_vv_f32m4_m(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -260,7 +260,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_f32m4_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmacc_vf_f32m4_m(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -270,7 +270,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_f32m8_m
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmacc_vv_f32m8_m(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -280,7 +280,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_f32m8_m
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmacc_vf_f32m8_m(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -290,7 +290,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vv_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmacc_vv_f64m1_m(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -300,7 +300,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vf_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmacc_vf_f64m1_m(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -310,7 +310,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vv_f64m2_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmacc_vv_f64m2_m(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -320,7 +320,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vf_f64m2_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmacc_vf_f64m2_m(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -330,7 +330,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vv_f64m4_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmacc_vv_f64m4_m(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -340,7 +340,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vf_f64m4_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmacc_vf_f64m4_m(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -350,7 +350,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vv_f64m8_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmacc_vv_f64m8_m(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -360,10 +360,370 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vf_f64m8_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmacc_vf_f64m8_m(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
   return __riscv_vfwnmacc_vf_f64m8_m(mask, vd, vs1, vs2, vl);
 }
 
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_f32mf2_rm
+// CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vv_f32mf2_rm(vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f32mf2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_f32mf2_rm
+// CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vf_f32mf2_rm(vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f32mf2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vv_f32m1_rm(vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f32m1_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vf_f32m1_rm(vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f32m1_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_f32m2_rm
+// CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vv_f32m2_rm(vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f32m2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_f32m2_rm
+// CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vf_f32m2_rm(vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f32m2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_f32m4_rm
+// CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vv_f32m4_rm(vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f32m4_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_f32m4_rm
+// CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vf_f32m4_rm(vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f32m4_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_f32m8_rm
+// CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vv_f32m8_rm(vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f32m8_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_f32m8_rm
+// CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vf_f32m8_rm(vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f32m8_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vv_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmacc_vv_f64m1_rm(vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f64m1_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vf_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmacc_vf_f64m1_rm(vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f64m1_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vv_f64m2_rm
+// CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmacc_vv_f64m2_rm(vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f64m2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vf_f64m2_rm
+// CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmacc_vf_f64m2_rm(vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f64m2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vv_f64m4_rm
+// CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmacc_vv_f64m4_rm(vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f64m4_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vf_f64m4_rm
+// CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmacc_vf_f64m4_rm(vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f64m4_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vv_f64m8_rm
+// CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmacc_vv_f64m8_rm(vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f64m8_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vf_f64m8_rm
+// CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmacc_vf_f64m8_rm(vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f64m8_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_f32mf2_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vv_f32mf2_rm_m(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f32mf2_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_f32mf2_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vf_f32mf2_rm_m(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f32mf2_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vv_f32m1_rm_m(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f32m1_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vf_f32m1_rm_m(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f32m1_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_f32m2_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vv_f32m2_rm_m(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f32m2_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_f32m2_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vf_f32m2_rm_m(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f32m2_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_f32m4_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vv_f32m4_rm_m(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f32m4_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_f32m4_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vf_f32m4_rm_m(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f32m4_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_f32m8_rm_m
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vv_f32m8_rm_m(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f32m8_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_f32m8_rm_m
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vf_f32m8_rm_m(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f32m8_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vv_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmacc_vv_f64m1_rm_m(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f64m1_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vf_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmacc_vf_f64m1_rm_m(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f64m1_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vv_f64m2_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmacc_vv_f64m2_rm_m(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f64m2_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vf_f64m2_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmacc_vf_f64m2_rm_m(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f64m2_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vv_f64m4_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmacc_vv_f64m4_rm_m(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f64m4_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vf_f64m4_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmacc_vf_f64m4_rm_m(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f64m4_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vv_f64m8_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmacc_vv_f64m8_rm_m(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f64m8_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vf_f64m8_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmacc_vf_f64m8_rm_m(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f64m8_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/vfwnmsac.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/vfwnmsac.c
--- a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/vfwnmsac.c
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/vfwnmsac.c
@@ -10,7 +10,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_f32mf2
 // CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmsac_vv_f32mf2(vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -20,7 +20,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_f32mf2
 // CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmsac_vf_f32mf2(vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -30,7 +30,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_f32m1
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmsac_vv_f32m1(vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -40,7 +40,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_f32m1
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmsac_vf_f32m1(vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -50,7 +50,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_f32m2
 // CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmsac_vv_f32m2(vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -60,7 +60,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_f32m2
 // CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmsac_vf_f32m2(vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -70,7 +70,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_f32m4
 // CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmsac_vv_f32m4(vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -80,7 +80,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_f32m4
 // CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmsac_vf_f32m4(vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -90,7 +90,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_f32m8
 // CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmsac_vv_f32m8(vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -100,7 +100,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_f32m8
 // CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmsac_vf_f32m8(vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -110,7 +110,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vv_f64m1
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmsac_vv_f64m1(vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -120,7 +120,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vf_f64m1
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmsac_vf_f64m1(vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -130,7 +130,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vv_f64m2
 // CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmsac_vv_f64m2(vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -140,7 +140,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vf_f64m2
 // CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmsac_vf_f64m2(vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -150,7 +150,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vv_f64m4
 // CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmsac_vv_f64m4(vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -160,7 +160,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vf_f64m4
 // CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmsac_vf_f64m4(vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -170,7 +170,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vv_f64m8
 // CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmsac_vv_f64m8(vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -180,7 +180,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vf_f64m8
 // CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmsac_vf_f64m8(vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
@@ -190,7 +190,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_f32mf2_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmsac_vv_f32mf2_m(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -200,7 +200,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_f32mf2_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmsac_vf_f32mf2_m(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -210,7 +210,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmsac_vv_f32m1_m(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -220,7 +220,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmsac_vf_f32m1_m(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -230,7 +230,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_f32m2_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmsac_vv_f32m2_m(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -240,7 +240,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_f32m2_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmsac_vf_f32m2_m(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -250,7 +250,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_f32m4_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmsac_vv_f32m4_m(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -260,7 +260,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_f32m4_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmsac_vf_f32m4_m(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -270,7 +270,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_f32m8_m
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmsac_vv_f32m8_m(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -280,7 +280,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_f32m8_m
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmsac_vf_f32m8_m(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -290,7 +290,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vv_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmsac_vv_f64m1_m(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -300,7 +300,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vf_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmsac_vf_f64m1_m(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -310,7 +310,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vv_f64m2_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmsac_vv_f64m2_m(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -320,7 +320,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vf_f64m2_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmsac_vf_f64m2_m(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -330,7 +330,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vv_f64m4_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmsac_vv_f64m4_m(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -340,7 +340,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vf_f64m4_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmsac_vf_f64m4_m(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -350,7 +350,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vv_f64m8_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmsac_vv_f64m8_m(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -360,10 +360,370 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vf_f64m8_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmsac_vf_f64m8_m(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
   return __riscv_vfwnmsac_vf_f64m8_m(mask, vd, vs1, vs2, vl);
 }
 
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_f32mf2_rm
+// CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vv_f32mf2_rm(vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f32mf2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_f32mf2_rm
+// CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vf_f32mf2_rm(vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f32mf2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vv_f32m1_rm(vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f32m1_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vf_f32m1_rm(vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f32m1_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_f32m2_rm
+// CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vv_f32m2_rm(vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f32m2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_f32m2_rm
+// CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vf_f32m2_rm(vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f32m2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_f32m4_rm
+// CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vv_f32m4_rm(vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f32m4_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_f32m4_rm
+// CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vf_f32m4_rm(vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f32m4_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_f32m8_rm
+// CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vv_f32m8_rm(vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f32m8_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_f32m8_rm
+// CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vf_f32m8_rm(vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f32m8_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vv_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmsac_vv_f64m1_rm(vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f64m1_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vf_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmsac_vf_f64m1_rm(vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f64m1_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vv_f64m2_rm
+// CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmsac_vv_f64m2_rm(vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f64m2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vf_f64m2_rm
+// CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmsac_vf_f64m2_rm(vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f64m2_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vv_f64m4_rm
+// CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmsac_vv_f64m4_rm(vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f64m4_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vf_f64m4_rm
+// CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmsac_vf_f64m4_rm(vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f64m4_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vv_f64m8_rm
+// CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmsac_vv_f64m8_rm(vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f64m8_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vf_f64m8_rm
+// CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmsac_vf_f64m8_rm(vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f64m8_rm(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_f32mf2_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vv_f32mf2_rm_m(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f32mf2_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_f32mf2_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vf_f32mf2_rm_m(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f32mf2_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vv_f32m1_rm_m(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f32m1_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vf_f32m1_rm_m(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f32m1_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_f32m2_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vv_f32m2_rm_m(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f32m2_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_f32m2_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vf_f32m2_rm_m(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f32m2_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_f32m4_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vv_f32m4_rm_m(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f32m4_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_f32m4_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vf_f32m4_rm_m(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f32m4_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_f32m8_rm_m
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vv_f32m8_rm_m(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f32m8_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_f32m8_rm_m
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vf_f32m8_rm_m(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f32m8_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vv_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmsac_vv_f64m1_rm_m(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f64m1_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vf_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmsac_vf_f64m1_rm_m(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f64m1_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vv_f64m2_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmsac_vv_f64m2_rm_m(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f64m2_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vf_f64m2_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmsac_vf_f64m2_rm_m(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f64m2_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vv_f64m4_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmsac_vv_f64m4_rm_m(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f64m4_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vf_f64m4_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmsac_vf_f64m4_rm_m(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f64m4_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vv_f64m8_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmsac_vv_f64m8_rm_m(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f64m8_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vf_f64m8_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmsac_vf_f64m8_rm_m(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f64m8_rm_m(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/vfwmacc.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/vfwmacc.c
--- a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/vfwmacc.c
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/vfwmacc.c
@@ -10,7 +10,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_f32mf2
 // CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmacc_vv_f32mf2(vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -20,7 +20,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_f32mf2
 // CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmacc_vf_f32mf2(vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -30,7 +30,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_f32m1
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmacc_vv_f32m1(vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -40,7 +40,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_f32m1
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmacc_vf_f32m1(vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -50,7 +50,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_f32m2
 // CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmacc_vv_f32m2(vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -60,7 +60,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_f32m2
 // CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmacc_vf_f32m2(vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -70,7 +70,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_f32m4
 // CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmacc_vv_f32m4(vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -80,7 +80,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_f32m4
 // CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmacc_vf_f32m4(vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -90,7 +90,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_f32m8
 // CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmacc_vv_f32m8(vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -100,7 +100,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_f32m8
 // CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmacc_vf_f32m8(vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -110,7 +110,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vv_f64m1
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmacc_vv_f64m1(vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -120,7 +120,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vf_f64m1
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmacc_vf_f64m1(vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -130,7 +130,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vv_f64m2
 // CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmacc_vv_f64m2(vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -140,7 +140,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vf_f64m2
 // CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmacc_vf_f64m2(vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -150,7 +150,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vv_f64m4
 // CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmacc_vv_f64m4(vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -160,7 +160,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vf_f64m4
 // CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmacc_vf_f64m4(vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -170,7 +170,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vv_f64m8
 // CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmacc_vv_f64m8(vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -180,7 +180,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vf_f64m8
 // CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmacc_vf_f64m8(vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
@@ -190,7 +190,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_f32mf2_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmacc_vv_f32mf2_m(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -200,7 +200,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_f32mf2_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmacc_vf_f32mf2_m(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -210,7 +210,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmacc_vv_f32m1_m(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -220,7 +220,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmacc_vf_f32m1_m(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -230,7 +230,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_f32m2_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmacc_vv_f32m2_m(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -240,7 +240,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_f32m2_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmacc_vf_f32m2_m(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -250,7 +250,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_f32m4_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmacc_vv_f32m4_m(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -260,7 +260,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_f32m4_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmacc_vf_f32m4_m(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -270,7 +270,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_f32m8_m
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmacc_vv_f32m8_m(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -280,7 +280,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_f32m8_m
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmacc_vf_f32m8_m(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -290,7 +290,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vv_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmacc_vv_f64m1_m(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -300,7 +300,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vf_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmacc_vf_f64m1_m(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -310,7 +310,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vv_f64m2_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmacc_vv_f64m2_m(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -320,7 +320,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vf_f64m2_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmacc_vf_f64m2_m(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -330,7 +330,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vv_f64m4_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmacc_vv_f64m4_m(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -340,7 +340,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vf_f64m4_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmacc_vf_f64m4_m(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -350,7 +350,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vv_f64m8_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmacc_vv_f64m8_m(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -360,10 +360,370 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vf_f64m8_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmacc_vf_f64m8_m(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
   return __riscv_vfwmacc(mask, vd, vs1, vs2, vl);
 }
 
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_f32mf2_rm
+// CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vv_f32mf2_rm(vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_f32mf2_rm
+// CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vf_f32mf2_rm(vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vv_f32m1_rm(vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vf_f32m1_rm(vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_f32m2_rm
+// CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vv_f32m2_rm(vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_f32m2_rm
+// CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vf_f32m2_rm(vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_f32m4_rm
+// CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vv_f32m4_rm(vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_f32m4_rm
+// CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vf_f32m4_rm(vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_f32m8_rm
+// CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vv_f32m8_rm(vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_f32m8_rm
+// CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vf_f32m8_rm(vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vv_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmacc_vv_f64m1_rm(vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vf_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmacc_vf_f64m1_rm(vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vv_f64m2_rm
+// CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmacc_vv_f64m2_rm(vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vf_f64m2_rm
+// CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmacc_vf_f64m2_rm(vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vv_f64m4_rm
+// CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmacc_vv_f64m4_rm(vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vf_f64m4_rm
+// CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmacc_vf_f64m4_rm(vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vv_f64m8_rm
+// CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmacc_vv_f64m8_rm(vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vf_f64m8_rm
+// CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmacc_vf_f64m8_rm(vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_f32mf2_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vv_f32mf2_rm_m(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_f32mf2_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vf_f32mf2_rm_m(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vv_f32m1_rm_m(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vf_f32m1_rm_m(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_f32m2_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vv_f32m2_rm_m(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_f32m2_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vf_f32m2_rm_m(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_f32m4_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vv_f32m4_rm_m(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_f32m4_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vf_f32m4_rm_m(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_f32m8_rm_m
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vv_f32m8_rm_m(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_f32m8_rm_m
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vf_f32m8_rm_m(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vv_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmacc_vv_f64m1_rm_m(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vf_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmacc_vf_f64m1_rm_m(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vv_f64m2_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmacc_vv_f64m2_rm_m(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vf_f64m2_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmacc_vf_f64m2_rm_m(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vv_f64m4_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmacc_vv_f64m4_rm_m(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vf_f64m4_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmacc_vf_f64m4_rm_m(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vv_f64m8_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmacc_vv_f64m8_rm_m(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vf_f64m8_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmacc_vf_f64m8_rm_m(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/vfwmsac.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/vfwmsac.c
--- a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/vfwmsac.c
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/vfwmsac.c
@@ -10,7 +10,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_f32mf2
 // CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmsac_vv_f32mf2(vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -20,7 +20,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_f32mf2
 // CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmsac_vf_f32mf2(vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -30,7 +30,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_f32m1
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmsac_vv_f32m1(vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -40,7 +40,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_f32m1
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmsac_vf_f32m1(vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -50,7 +50,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_f32m2
 // CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmsac_vv_f32m2(vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -60,7 +60,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_f32m2
 // CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmsac_vf_f32m2(vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -70,7 +70,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_f32m4
 // CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmsac_vv_f32m4(vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -80,7 +80,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_f32m4
 // CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmsac_vf_f32m4(vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -90,7 +90,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_f32m8
 // CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmsac_vv_f32m8(vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -100,7 +100,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_f32m8
 // CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmsac_vf_f32m8(vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -110,7 +110,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vv_f64m1
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmsac_vv_f64m1(vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -120,7 +120,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vf_f64m1
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmsac_vf_f64m1(vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -130,7 +130,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vv_f64m2
 // CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmsac_vv_f64m2(vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -140,7 +140,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vf_f64m2
 // CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmsac_vf_f64m2(vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -150,7 +150,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vv_f64m4
 // CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmsac_vv_f64m4(vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -160,7 +160,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vf_f64m4
 // CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmsac_vf_f64m4(vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -170,7 +170,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vv_f64m8
 // CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmsac_vv_f64m8(vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -180,7 +180,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vf_f64m8
 // CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmsac_vf_f64m8(vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
@@ -190,7 +190,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_f32mf2_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmsac_vv_f32mf2_m(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -200,7 +200,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_f32mf2_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmsac_vf_f32mf2_m(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -210,7 +210,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmsac_vv_f32m1_m(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -220,7 +220,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmsac_vf_f32m1_m(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -230,7 +230,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_f32m2_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmsac_vv_f32m2_m(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -240,7 +240,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_f32m2_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmsac_vf_f32m2_m(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -250,7 +250,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_f32m4_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmsac_vv_f32m4_m(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -260,7 +260,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_f32m4_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmsac_vf_f32m4_m(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -270,7 +270,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_f32m8_m
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmsac_vv_f32m8_m(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -280,7 +280,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_f32m8_m
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmsac_vf_f32m8_m(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -290,7 +290,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vv_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmsac_vv_f64m1_m(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -300,7 +300,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vf_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmsac_vf_f64m1_m(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -310,7 +310,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vv_f64m2_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmsac_vv_f64m2_m(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -320,7 +320,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vf_f64m2_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmsac_vf_f64m2_m(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -330,7 +330,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vv_f64m4_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmsac_vv_f64m4_m(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -340,7 +340,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vf_f64m4_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmsac_vf_f64m4_m(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -350,7 +350,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vv_f64m8_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmsac_vv_f64m8_m(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -360,10 +360,370 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vf_f64m8_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmsac_vf_f64m8_m(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
   return __riscv_vfwmsac(mask, vd, vs1, vs2, vl);
 }
 
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_f32mf2_rm
+// CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vv_f32mf2_rm(vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_f32mf2_rm
+// CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vf_f32mf2_rm(vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vv_f32m1_rm(vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vf_f32m1_rm(vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_f32m2_rm
+// CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vv_f32m2_rm(vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_f32m2_rm
+// CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vf_f32m2_rm(vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_f32m4_rm
+// CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vv_f32m4_rm(vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_f32m4_rm
+// CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vf_f32m4_rm(vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_f32m8_rm
+// CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vv_f32m8_rm(vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_f32m8_rm
+// CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vf_f32m8_rm(vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vv_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmsac_vv_f64m1_rm(vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vf_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmsac_vf_f64m1_rm(vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vv_f64m2_rm
+// CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmsac_vv_f64m2_rm(vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vf_f64m2_rm
+// CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmsac_vf_f64m2_rm(vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vv_f64m4_rm
+// CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmsac_vv_f64m4_rm(vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vf_f64m4_rm
+// CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmsac_vf_f64m4_rm(vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vv_f64m8_rm
+// CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmsac_vv_f64m8_rm(vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vf_f64m8_rm
+// CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmsac_vf_f64m8_rm(vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_f32mf2_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vv_f32mf2_rm_m(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_f32mf2_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vf_f32mf2_rm_m(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vv_f32m1_rm_m(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vf_f32m1_rm_m(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_f32m2_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vv_f32m2_rm_m(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_f32m2_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vf_f32m2_rm_m(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_f32m4_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vv_f32m4_rm_m(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_f32m4_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vf_f32m4_rm_m(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_f32m8_rm_m
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vv_f32m8_rm_m(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_f32m8_rm_m
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vf_f32m8_rm_m(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vv_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmsac_vv_f64m1_rm_m(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vf_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmsac_vf_f64m1_rm_m(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vv_f64m2_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmsac_vv_f64m2_rm_m(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vf_f64m2_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmsac_vf_f64m2_rm_m(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vv_f64m4_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmsac_vv_f64m4_rm_m(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vf_f64m4_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmsac_vf_f64m4_rm_m(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vv_f64m8_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmsac_vv_f64m8_rm_m(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vf_f64m8_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmsac_vf_f64m8_rm_m(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/vfwnmacc.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/vfwnmacc.c
--- a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/vfwnmacc.c
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/vfwnmacc.c
@@ -10,7 +10,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_f32mf2
 // CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmacc_vv_f32mf2(vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -20,7 +20,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_f32mf2
 // CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmacc_vf_f32mf2(vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -30,7 +30,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_f32m1
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmacc_vv_f32m1(vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -40,7 +40,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_f32m1
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmacc_vf_f32m1(vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -50,7 +50,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_f32m2
 // CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmacc_vv_f32m2(vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -60,7 +60,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_f32m2
 // CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmacc_vf_f32m2(vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -70,7 +70,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_f32m4
 // CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmacc_vv_f32m4(vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -80,7 +80,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_f32m4
 // CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmacc_vf_f32m4(vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -90,7 +90,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_f32m8
 // CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmacc_vv_f32m8(vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -100,7 +100,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_f32m8
 // CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmacc_vf_f32m8(vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -110,7 +110,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vv_f64m1
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmacc_vv_f64m1(vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -120,7 +120,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vf_f64m1
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmacc_vf_f64m1(vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -130,7 +130,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vv_f64m2
 // CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmacc_vv_f64m2(vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -140,7 +140,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vf_f64m2
 // CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmacc_vf_f64m2(vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -150,7 +150,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vv_f64m4
 // CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmacc_vv_f64m4(vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -160,7 +160,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vf_f64m4
 // CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmacc_vf_f64m4(vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -170,7 +170,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vv_f64m8
 // CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmacc_vv_f64m8(vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -180,7 +180,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vf_f64m8
 // CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmacc_vf_f64m8(vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
@@ -190,7 +190,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_f32mf2_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmacc_vv_f32mf2_m(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -200,7 +200,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_f32mf2_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmacc_vf_f32mf2_m(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -210,7 +210,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmacc_vv_f32m1_m(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -220,7 +220,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmacc_vf_f32m1_m(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -230,7 +230,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_f32m2_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmacc_vv_f32m2_m(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -240,7 +240,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_f32m2_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmacc_vf_f32m2_m(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -250,7 +250,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_f32m4_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmacc_vv_f32m4_m(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -260,7 +260,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_f32m4_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmacc_vf_f32m4_m(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -270,7 +270,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_f32m8_m
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmacc_vv_f32m8_m(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -280,7 +280,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_f32m8_m
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmacc_vf_f32m8_m(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -290,7 +290,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vv_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmacc_vv_f64m1_m(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -300,7 +300,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vf_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmacc_vf_f64m1_m(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -310,7 +310,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vv_f64m2_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmacc_vv_f64m2_m(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -320,7 +320,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vf_f64m2_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmacc_vf_f64m2_m(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -330,7 +330,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vv_f64m4_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmacc_vv_f64m4_m(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -340,7 +340,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vf_f64m4_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmacc_vf_f64m4_m(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -350,7 +350,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vv_f64m8_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmacc_vv_f64m8_m(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -360,10 +360,370 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vf_f64m8_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmacc_vf_f64m8_m(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
   return __riscv_vfwnmacc(mask, vd, vs1, vs2, vl);
 }
 
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_f32mf2_rm
+// CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vv_f32mf2_rm(vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_f32mf2_rm
+// CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vf_f32mf2_rm(vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vv_f32m1_rm(vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vf_f32m1_rm(vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_f32m2_rm
+// CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vv_f32m2_rm(vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_f32m2_rm
+// CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vf_f32m2_rm(vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_f32m4_rm
+// CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vv_f32m4_rm(vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_f32m4_rm
+// CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vf_f32m4_rm(vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_f32m8_rm
+// CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vv_f32m8_rm(vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_f32m8_rm
+// CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vf_f32m8_rm(vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vv_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmacc_vv_f64m1_rm(vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vf_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmacc_vf_f64m1_rm(vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vv_f64m2_rm
+// CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmacc_vv_f64m2_rm(vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vf_f64m2_rm
+// CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmacc_vf_f64m2_rm(vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vv_f64m4_rm
+// CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmacc_vv_f64m4_rm(vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vf_f64m4_rm
+// CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmacc_vf_f64m4_rm(vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vv_f64m8_rm
+// CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmacc_vv_f64m8_rm(vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vf_f64m8_rm
+// CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmacc_vf_f64m8_rm(vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_f32mf2_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vv_f32mf2_rm_m(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_f32mf2_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vf_f32mf2_rm_m(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vv_f32m1_rm_m(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vf_f32m1_rm_m(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_f32m2_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vv_f32m2_rm_m(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_f32m2_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vf_f32m2_rm_m(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_f32m4_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vv_f32m4_rm_m(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_f32m4_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vf_f32m4_rm_m(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_f32m8_rm_m
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vv_f32m8_rm_m(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_f32m8_rm_m
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vf_f32m8_rm_m(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vv_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmacc_vv_f64m1_rm_m(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vf_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmacc_vf_f64m1_rm_m(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vv_f64m2_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmacc_vv_f64m2_rm_m(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vf_f64m2_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmacc_vf_f64m2_rm_m(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vv_f64m4_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmacc_vv_f64m4_rm_m(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vf_f64m4_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmacc_vf_f64m4_rm_m(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vv_f64m8_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmacc_vv_f64m8_rm_m(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vf_f64m8_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmacc_vf_f64m8_rm_m(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/vfwnmsac.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/vfwnmsac.c
--- a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/vfwnmsac.c
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/overloaded/vfwnmsac.c
@@ -10,7 +10,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_f32mf2
 // CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmsac_vv_f32mf2(vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -20,7 +20,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_f32mf2
 // CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmsac_vf_f32mf2(vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -30,7 +30,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_f32m1
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmsac_vv_f32m1(vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -40,7 +40,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_f32m1
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmsac_vf_f32m1(vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -50,7 +50,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_f32m2
 // CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmsac_vv_f32m2(vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -60,7 +60,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_f32m2
 // CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmsac_vf_f32m2(vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -70,7 +70,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_f32m4
 // CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmsac_vv_f32m4(vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -80,7 +80,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_f32m4
 // CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmsac_vf_f32m4(vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -90,7 +90,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_f32m8
 // CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmsac_vv_f32m8(vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -100,7 +100,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_f32m8
 // CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmsac_vf_f32m8(vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -110,7 +110,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vv_f64m1
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmsac_vv_f64m1(vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -120,7 +120,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vf_f64m1
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmsac_vf_f64m1(vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -130,7 +130,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vv_f64m2
 // CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmsac_vv_f64m2(vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -140,7 +140,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vf_f64m2
 // CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmsac_vf_f64m2(vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -150,7 +150,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vv_f64m4
 // CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmsac_vv_f64m4(vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -160,7 +160,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vf_f64m4
 // CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmsac_vf_f64m4(vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -170,7 +170,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vv_f64m8
 // CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmsac_vv_f64m8(vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -180,7 +180,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vf_f64m8
 // CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmsac_vf_f64m8(vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
@@ -190,7 +190,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_f32mf2_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmsac_vv_f32mf2_m(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -200,7 +200,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_f32mf2_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmsac_vf_f32mf2_m(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -210,7 +210,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmsac_vv_f32m1_m(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -220,7 +220,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_f32m1_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmsac_vf_f32m1_m(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -230,7 +230,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_f32m2_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmsac_vv_f32m2_m(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -240,7 +240,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_f32m2_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmsac_vf_f32m2_m(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -250,7 +250,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_f32m4_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmsac_vv_f32m4_m(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -260,7 +260,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_f32m4_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmsac_vf_f32m4_m(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -270,7 +270,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_f32m8_m
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmsac_vv_f32m8_m(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -280,7 +280,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_f32m8_m
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmsac_vf_f32m8_m(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -290,7 +290,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vv_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmsac_vv_f64m1_m(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -300,7 +300,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vf_f64m1_m
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmsac_vf_f64m1_m(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -310,7 +310,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vv_f64m2_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmsac_vv_f64m2_m(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -320,7 +320,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vf_f64m2_m
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmsac_vf_f64m2_m(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -330,7 +330,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vv_f64m4_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmsac_vv_f64m4_m(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -340,7 +340,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vf_f64m4_m
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmsac_vf_f64m4_m(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -350,7 +350,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vv_f64m8_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmsac_vv_f64m8_m(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -360,10 +360,370 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vf_f64m8_m
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 3)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmsac_vf_f64m8_m(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
   return __riscv_vfwnmsac(mask, vd, vs1, vs2, vl);
 }
 
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_f32mf2_rm
+// CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vv_f32mf2_rm(vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_f32mf2_rm
+// CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vf_f32mf2_rm(vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vv_f32m1_rm(vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_f32m1_rm
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vf_f32m1_rm(vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_f32m2_rm
+// CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vv_f32m2_rm(vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_f32m2_rm
+// CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vf_f32m2_rm(vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_f32m4_rm
+// CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vv_f32m4_rm(vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_f32m4_rm
+// CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vf_f32m4_rm(vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_f32m8_rm
+// CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vv_f32m8_rm(vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_f32m8_rm
+// CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vf_f32m8_rm(vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vv_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmsac_vv_f64m1_rm(vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vf_f64m1_rm
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmsac_vf_f64m1_rm(vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vv_f64m2_rm
+// CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmsac_vv_f64m2_rm(vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vf_f64m2_rm
+// CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmsac_vf_f64m2_rm(vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vv_f64m4_rm
+// CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmsac_vv_f64m4_rm(vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vf_f64m4_rm
+// CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmsac_vf_f64m4_rm(vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vv_f64m8_rm
+// CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmsac_vv_f64m8_rm(vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vf_f64m8_rm
+// CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmsac_vf_f64m8_rm(vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_f32mf2_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vv_f32mf2_rm_m(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_f32mf2_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vf_f32mf2_rm_m(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vv_f32m1_rm_m(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_f32m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vf_f32m1_rm_m(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_f32m2_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vv_f32m2_rm_m(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_f32m2_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vf_f32m2_rm_m(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_f32m4_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vv_f32m4_rm_m(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_f32m4_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vf_f32m4_rm_m(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_f32m8_rm_m
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vv_f32m8_rm_m(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_f32m8_rm_m
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vf_f32m8_rm_m(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vv_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmsac_vv_f64m1_rm_m(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vf_f64m1_rm_m
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmsac_vf_f64m1_rm_m(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vv_f64m2_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmsac_vv_f64m2_rm_m(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vf_f64m2_rm_m
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmsac_vf_f64m2_rm_m(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vv_f64m4_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmsac_vv_f64m4_rm_m(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vf_f64m4_rm_m
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmsac_vf_f64m4_rm_m(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vv_f64m8_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmsac_vv_f64m8_rm_m(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vf_f64m8_rm_m
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 3)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmsac_vf_f64m8_rm_m(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/vfwmacc.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/vfwmacc.c
--- a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/vfwmacc.c
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/vfwmacc.c
@@ -10,7 +10,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_f32mf2_tu
 // CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmacc_vv_f32mf2_tu(vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -20,7 +20,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_f32mf2_tu
 // CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmacc_vf_f32mf2_tu(vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -30,7 +30,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmacc_vv_f32m1_tu(vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -40,7 +40,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmacc_vf_f32m1_tu(vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -50,7 +50,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_f32m2_tu
 // CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmacc_vv_f32m2_tu(vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -60,7 +60,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_f32m2_tu
 // CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmacc_vf_f32m2_tu(vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -70,7 +70,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_f32m4_tu
 // CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmacc_vv_f32m4_tu(vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -80,7 +80,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_f32m4_tu
 // CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmacc_vf_f32m4_tu(vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -90,7 +90,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_f32m8_tu
 // CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmacc_vv_f32m8_tu(vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -100,7 +100,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_f32m8_tu
 // CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmacc_vf_f32m8_tu(vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -110,7 +110,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vv_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmacc_vv_f64m1_tu(vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -120,7 +120,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vf_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmacc_vf_f64m1_tu(vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -130,7 +130,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vv_f64m2_tu
 // CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmacc_vv_f64m2_tu(vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -140,7 +140,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vf_f64m2_tu
 // CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmacc_vf_f64m2_tu(vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -150,7 +150,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vv_f64m4_tu
 // CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmacc_vv_f64m4_tu(vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -160,7 +160,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vf_f64m4_tu
 // CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmacc_vf_f64m4_tu(vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -170,7 +170,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vv_f64m8_tu
 // CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmacc_vv_f64m8_tu(vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -180,7 +180,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vf_f64m8_tu
 // CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmacc_vf_f64m8_tu(vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
@@ -190,7 +190,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_f32mf2_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmacc_vv_f32mf2_tum(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -200,7 +200,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_f32mf2_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmacc_vf_f32mf2_tum(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -210,7 +210,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmacc_vv_f32m1_tum(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -220,7 +220,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmacc_vf_f32m1_tum(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -230,7 +230,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_f32m2_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmacc_vv_f32m2_tum(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -240,7 +240,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_f32m2_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmacc_vf_f32m2_tum(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -250,7 +250,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_f32m4_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmacc_vv_f32m4_tum(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -260,7 +260,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_f32m4_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmacc_vf_f32m4_tum(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -270,7 +270,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_f32m8_tum
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmacc_vv_f32m8_tum(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -280,7 +280,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_f32m8_tum
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmacc_vf_f32m8_tum(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -290,7 +290,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vv_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmacc_vv_f64m1_tum(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -300,7 +300,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vf_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmacc_vf_f64m1_tum(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -310,7 +310,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vv_f64m2_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmacc_vv_f64m2_tum(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -320,7 +320,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vf_f64m2_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmacc_vf_f64m2_tum(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -330,7 +330,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vv_f64m4_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmacc_vv_f64m4_tum(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -340,7 +340,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vf_f64m4_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmacc_vf_f64m4_tum(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -350,7 +350,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vv_f64m8_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmacc_vv_f64m8_tum(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -360,7 +360,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vf_f64m8_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmacc_vf_f64m8_tum(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
@@ -370,7 +370,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_f32mf2_tumu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmacc_vv_f32mf2_tumu(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -380,7 +380,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_f32mf2_tumu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmacc_vf_f32mf2_tumu(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -390,7 +390,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_f32m1_tumu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmacc_vv_f32m1_tumu(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -400,7 +400,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_f32m1_tumu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmacc_vf_f32m1_tumu(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -410,7 +410,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_f32m2_tumu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmacc_vv_f32m2_tumu(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -420,7 +420,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_f32m2_tumu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmacc_vf_f32m2_tumu(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -430,7 +430,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_f32m4_tumu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmacc_vv_f32m4_tumu(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -440,7 +440,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_f32m4_tumu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmacc_vf_f32m4_tumu(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -450,7 +450,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_f32m8_tumu
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmacc_vv_f32m8_tumu(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -460,7 +460,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_f32m8_tumu
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmacc_vf_f32m8_tumu(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -470,7 +470,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vv_f64m1_tumu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmacc_vv_f64m1_tumu(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -480,7 +480,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vf_f64m1_tumu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmacc_vf_f64m1_tumu(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -490,7 +490,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vv_f64m2_tumu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmacc_vv_f64m2_tumu(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -500,7 +500,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vf_f64m2_tumu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmacc_vf_f64m2_tumu(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -510,7 +510,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vv_f64m4_tumu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmacc_vv_f64m4_tumu(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -520,7 +520,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vf_f64m4_tumu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmacc_vf_f64m4_tumu(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -530,7 +530,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vv_f64m8_tumu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmacc_vv_f64m8_tumu(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -540,7 +540,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vf_f64m8_tumu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmacc_vf_f64m8_tumu(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
@@ -550,7 +550,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_f32mf2_mu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmacc_vv_f32mf2_mu(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -560,7 +560,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_f32mf2_mu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmacc_vf_f32mf2_mu(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -570,7 +570,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_f32m1_mu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmacc_vv_f32m1_mu(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -580,7 +580,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_f32m1_mu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmacc_vf_f32m1_mu(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -590,7 +590,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_f32m2_mu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmacc_vv_f32m2_mu(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -600,7 +600,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_f32m2_mu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmacc_vf_f32m2_mu(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -610,7 +610,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_f32m4_mu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmacc_vv_f32m4_mu(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -620,7 +620,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_f32m4_mu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmacc_vf_f32m4_mu(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -630,7 +630,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_f32m8_mu
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmacc_vv_f32m8_mu(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -640,7 +640,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_f32m8_mu
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmacc_vf_f32m8_mu(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -650,7 +650,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vv_f64m1_mu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmacc_vv_f64m1_mu(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -660,7 +660,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vf_f64m1_mu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmacc_vf_f64m1_mu(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -670,7 +670,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vv_f64m2_mu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmacc_vv_f64m2_mu(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -680,7 +680,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vf_f64m2_mu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmacc_vf_f64m2_mu(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -690,7 +690,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vv_f64m4_mu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmacc_vv_f64m4_mu(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -700,7 +700,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vf_f64m4_mu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmacc_vf_f64m4_mu(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -710,7 +710,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vv_f64m8_mu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmacc_vv_f64m8_mu(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -720,10 +720,730 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vf_f64m8_mu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmacc_vf_f64m8_mu(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
   return __riscv_vfwmacc_vf_f64m8_mu(mask, vd, vs1, vs2, vl);
 }
 
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_f32mf2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vv_f32mf2_rm_tu(vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f32mf2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_f32mf2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vf_f32mf2_rm_tu(vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f32mf2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vv_f32m1_rm_tu(vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f32m1_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vf_f32m1_rm_tu(vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f32m1_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_f32m2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vv_f32m2_rm_tu(vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f32m2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_f32m2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vf_f32m2_rm_tu(vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f32m2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_f32m4_rm_tu
+// CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vv_f32m4_rm_tu(vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f32m4_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_f32m4_rm_tu
+// CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vf_f32m4_rm_tu(vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f32m4_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_f32m8_rm_tu
+// CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vv_f32m8_rm_tu(vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f32m8_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_f32m8_rm_tu
+// CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vf_f32m8_rm_tu(vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f32m8_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vv_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmacc_vv_f64m1_rm_tu(vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f64m1_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vf_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmacc_vf_f64m1_rm_tu(vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f64m1_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vv_f64m2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmacc_vv_f64m2_rm_tu(vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f64m2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vf_f64m2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmacc_vf_f64m2_rm_tu(vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f64m2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vv_f64m4_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmacc_vv_f64m4_rm_tu(vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f64m4_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vf_f64m4_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmacc_vf_f64m4_rm_tu(vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f64m4_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vv_f64m8_rm_tu
+// CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmacc_vv_f64m8_rm_tu(vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f64m8_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vf_f64m8_rm_tu
+// CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmacc_vf_f64m8_rm_tu(vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f64m8_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_f32mf2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vv_f32mf2_rm_tum(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f32mf2_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_f32mf2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vf_f32mf2_rm_tum(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f32mf2_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vv_f32m1_rm_tum(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f32m1_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vf_f32m1_rm_tum(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f32m1_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_f32m2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vv_f32m2_rm_tum(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f32m2_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_f32m2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vf_f32m2_rm_tum(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f32m2_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_f32m4_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vv_f32m4_rm_tum(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f32m4_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_f32m4_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vf_f32m4_rm_tum(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f32m4_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_f32m8_rm_tum
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vv_f32m8_rm_tum(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f32m8_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_f32m8_rm_tum
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vf_f32m8_rm_tum(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f32m8_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vv_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmacc_vv_f64m1_rm_tum(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f64m1_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vf_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmacc_vf_f64m1_rm_tum(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f64m1_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vv_f64m2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmacc_vv_f64m2_rm_tum(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f64m2_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vf_f64m2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmacc_vf_f64m2_rm_tum(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f64m2_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vv_f64m4_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmacc_vv_f64m4_rm_tum(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f64m4_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vf_f64m4_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmacc_vf_f64m4_rm_tum(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f64m4_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vv_f64m8_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmacc_vv_f64m8_rm_tum(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f64m8_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vf_f64m8_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmacc_vf_f64m8_rm_tum(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f64m8_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_f32mf2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vv_f32mf2_rm_tumu(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f32mf2_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_f32mf2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vf_f32mf2_rm_tumu(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f32mf2_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_f32m1_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vv_f32m1_rm_tumu(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f32m1_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_f32m1_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vf_f32m1_rm_tumu(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f32m1_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_f32m2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vv_f32m2_rm_tumu(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f32m2_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_f32m2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vf_f32m2_rm_tumu(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f32m2_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_f32m4_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vv_f32m4_rm_tumu(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f32m4_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_f32m4_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vf_f32m4_rm_tumu(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f32m4_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_f32m8_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vv_f32m8_rm_tumu(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f32m8_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_f32m8_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vf_f32m8_rm_tumu(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f32m8_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vv_f64m1_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmacc_vv_f64m1_rm_tumu(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f64m1_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vf_f64m1_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmacc_vf_f64m1_rm_tumu(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f64m1_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vv_f64m2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmacc_vv_f64m2_rm_tumu(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f64m2_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vf_f64m2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmacc_vf_f64m2_rm_tumu(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f64m2_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vv_f64m4_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmacc_vv_f64m4_rm_tumu(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f64m4_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vf_f64m4_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmacc_vf_f64m4_rm_tumu(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f64m4_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vv_f64m8_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmacc_vv_f64m8_rm_tumu(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f64m8_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vf_f64m8_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmacc_vf_f64m8_rm_tumu(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f64m8_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_f32mf2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vv_f32mf2_rm_mu(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f32mf2_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_f32mf2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vf_f32mf2_rm_mu(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f32mf2_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_f32m1_rm_mu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vv_f32m1_rm_mu(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f32m1_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_f32m1_rm_mu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vf_f32m1_rm_mu(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f32m1_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_f32m2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vv_f32m2_rm_mu(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f32m2_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_f32m2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vf_f32m2_rm_mu(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f32m2_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_f32m4_rm_mu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vv_f32m4_rm_mu(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f32m4_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_f32m4_rm_mu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vf_f32m4_rm_mu(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f32m4_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_f32m8_rm_mu
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vv_f32m8_rm_mu(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f32m8_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_f32m8_rm_mu
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vf_f32m8_rm_mu(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f32m8_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vv_f64m1_rm_mu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmacc_vv_f64m1_rm_mu(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f64m1_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vf_f64m1_rm_mu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmacc_vf_f64m1_rm_mu(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f64m1_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vv_f64m2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmacc_vv_f64m2_rm_mu(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f64m2_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vf_f64m2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmacc_vf_f64m2_rm_mu(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f64m2_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vv_f64m4_rm_mu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmacc_vv_f64m4_rm_mu(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f64m4_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vf_f64m4_rm_mu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmacc_vf_f64m4_rm_mu(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f64m4_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vv_f64m8_rm_mu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmacc_vv_f64m8_rm_mu(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vv_f64m8_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vf_f64m8_rm_mu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmacc_vf_f64m8_rm_mu(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_vf_f64m8_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/vfwmsac.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/vfwmsac.c
--- a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/vfwmsac.c
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/vfwmsac.c
@@ -10,7 +10,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_f32mf2_tu
 // CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmsac_vv_f32mf2_tu(vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -20,7 +20,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_f32mf2_tu
 // CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmsac_vf_f32mf2_tu(vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -30,7 +30,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmsac_vv_f32m1_tu(vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -40,7 +40,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmsac_vf_f32m1_tu(vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -50,7 +50,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_f32m2_tu
 // CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmsac_vv_f32m2_tu(vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -60,7 +60,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_f32m2_tu
 // CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmsac_vf_f32m2_tu(vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -70,7 +70,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_f32m4_tu
 // CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmsac_vv_f32m4_tu(vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -80,7 +80,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_f32m4_tu
 // CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmsac_vf_f32m4_tu(vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -90,7 +90,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_f32m8_tu
 // CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmsac_vv_f32m8_tu(vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -100,7 +100,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_f32m8_tu
 // CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmsac_vf_f32m8_tu(vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -110,7 +110,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vv_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmsac_vv_f64m1_tu(vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -120,7 +120,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vf_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmsac_vf_f64m1_tu(vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -130,7 +130,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vv_f64m2_tu
 // CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmsac_vv_f64m2_tu(vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -140,7 +140,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vf_f64m2_tu
 // CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmsac_vf_f64m2_tu(vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -150,7 +150,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vv_f64m4_tu
 // CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmsac_vv_f64m4_tu(vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -160,7 +160,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vf_f64m4_tu
 // CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmsac_vf_f64m4_tu(vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -170,7 +170,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vv_f64m8_tu
 // CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmsac_vv_f64m8_tu(vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -180,7 +180,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vf_f64m8_tu
 // CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmsac_vf_f64m8_tu(vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
@@ -190,7 +190,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_f32mf2_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmsac_vv_f32mf2_tum(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -200,7 +200,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_f32mf2_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmsac_vf_f32mf2_tum(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -210,7 +210,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmsac_vv_f32m1_tum(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -220,7 +220,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmsac_vf_f32m1_tum(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -230,7 +230,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_f32m2_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmsac_vv_f32m2_tum(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -240,7 +240,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_f32m2_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmsac_vf_f32m2_tum(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -250,7 +250,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_f32m4_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmsac_vv_f32m4_tum(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -260,7 +260,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_f32m4_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmsac_vf_f32m4_tum(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -270,7 +270,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_f32m8_tum
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmsac_vv_f32m8_tum(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -280,7 +280,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_f32m8_tum
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmsac_vf_f32m8_tum(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -290,7 +290,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vv_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmsac_vv_f64m1_tum(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -300,7 +300,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vf_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmsac_vf_f64m1_tum(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -310,7 +310,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vv_f64m2_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmsac_vv_f64m2_tum(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -320,7 +320,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vf_f64m2_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmsac_vf_f64m2_tum(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -330,7 +330,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vv_f64m4_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmsac_vv_f64m4_tum(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -340,7 +340,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vf_f64m4_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmsac_vf_f64m4_tum(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -350,7 +350,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vv_f64m8_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmsac_vv_f64m8_tum(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -360,7 +360,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vf_f64m8_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmsac_vf_f64m8_tum(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
@@ -370,7 +370,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_f32mf2_tumu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmsac_vv_f32mf2_tumu(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -380,7 +380,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_f32mf2_tumu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmsac_vf_f32mf2_tumu(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -390,7 +390,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_f32m1_tumu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmsac_vv_f32m1_tumu(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -400,7 +400,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_f32m1_tumu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmsac_vf_f32m1_tumu(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -410,7 +410,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_f32m2_tumu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmsac_vv_f32m2_tumu(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -420,7 +420,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_f32m2_tumu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmsac_vf_f32m2_tumu(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -430,7 +430,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_f32m4_tumu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmsac_vv_f32m4_tumu(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -440,7 +440,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_f32m4_tumu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmsac_vf_f32m4_tumu(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -450,7 +450,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_f32m8_tumu
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmsac_vv_f32m8_tumu(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -460,7 +460,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_f32m8_tumu
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmsac_vf_f32m8_tumu(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -470,7 +470,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vv_f64m1_tumu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmsac_vv_f64m1_tumu(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -480,7 +480,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vf_f64m1_tumu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmsac_vf_f64m1_tumu(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -490,7 +490,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vv_f64m2_tumu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmsac_vv_f64m2_tumu(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -500,7 +500,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vf_f64m2_tumu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmsac_vf_f64m2_tumu(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -510,7 +510,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vv_f64m4_tumu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmsac_vv_f64m4_tumu(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -520,7 +520,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vf_f64m4_tumu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmsac_vf_f64m4_tumu(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -530,7 +530,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vv_f64m8_tumu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmsac_vv_f64m8_tumu(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -540,7 +540,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vf_f64m8_tumu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmsac_vf_f64m8_tumu(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
@@ -550,7 +550,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_f32mf2_mu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmsac_vv_f32mf2_mu(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -560,7 +560,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_f32mf2_mu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmsac_vf_f32mf2_mu(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -570,7 +570,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_f32m1_mu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmsac_vv_f32m1_mu(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -580,7 +580,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_f32m1_mu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmsac_vf_f32m1_mu(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -590,7 +590,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_f32m2_mu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmsac_vv_f32m2_mu(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -600,7 +600,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_f32m2_mu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmsac_vf_f32m2_mu(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -610,7 +610,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_f32m4_mu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmsac_vv_f32m4_mu(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -620,7 +620,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_f32m4_mu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmsac_vf_f32m4_mu(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -630,7 +630,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_f32m8_mu
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmsac_vv_f32m8_mu(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -640,7 +640,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_f32m8_mu
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmsac_vf_f32m8_mu(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -650,7 +650,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vv_f64m1_mu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmsac_vv_f64m1_mu(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -660,7 +660,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vf_f64m1_mu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmsac_vf_f64m1_mu(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -670,7 +670,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vv_f64m2_mu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmsac_vv_f64m2_mu(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -680,7 +680,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vf_f64m2_mu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmsac_vf_f64m2_mu(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -690,7 +690,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vv_f64m4_mu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmsac_vv_f64m4_mu(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -700,7 +700,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vf_f64m4_mu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmsac_vf_f64m4_mu(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -710,7 +710,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vv_f64m8_mu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmsac_vv_f64m8_mu(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -720,10 +720,730 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vf_f64m8_mu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmsac_vf_f64m8_mu(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
   return __riscv_vfwmsac_vf_f64m8_mu(mask, vd, vs1, vs2, vl);
 }
 
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_f32mf2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vv_f32mf2_rm_tu(vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f32mf2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_f32mf2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vf_f32mf2_rm_tu(vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f32mf2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vv_f32m1_rm_tu(vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f32m1_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vf_f32m1_rm_tu(vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f32m1_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_f32m2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vv_f32m2_rm_tu(vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f32m2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_f32m2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vf_f32m2_rm_tu(vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f32m2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_f32m4_rm_tu
+// CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vv_f32m4_rm_tu(vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f32m4_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_f32m4_rm_tu
+// CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vf_f32m4_rm_tu(vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f32m4_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_f32m8_rm_tu
+// CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vv_f32m8_rm_tu(vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f32m8_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_f32m8_rm_tu
+// CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vf_f32m8_rm_tu(vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f32m8_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vv_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmsac_vv_f64m1_rm_tu(vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f64m1_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vf_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmsac_vf_f64m1_rm_tu(vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f64m1_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vv_f64m2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmsac_vv_f64m2_rm_tu(vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f64m2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vf_f64m2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmsac_vf_f64m2_rm_tu(vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f64m2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vv_f64m4_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmsac_vv_f64m4_rm_tu(vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f64m4_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vf_f64m4_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmsac_vf_f64m4_rm_tu(vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f64m4_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vv_f64m8_rm_tu
+// CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmsac_vv_f64m8_rm_tu(vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f64m8_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vf_f64m8_rm_tu
+// CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmsac_vf_f64m8_rm_tu(vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f64m8_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_f32mf2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vv_f32mf2_rm_tum(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f32mf2_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_f32mf2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vf_f32mf2_rm_tum(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f32mf2_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vv_f32m1_rm_tum(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f32m1_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vf_f32m1_rm_tum(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f32m1_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_f32m2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vv_f32m2_rm_tum(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f32m2_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_f32m2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vf_f32m2_rm_tum(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f32m2_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_f32m4_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vv_f32m4_rm_tum(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f32m4_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_f32m4_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vf_f32m4_rm_tum(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f32m4_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_f32m8_rm_tum
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vv_f32m8_rm_tum(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f32m8_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_f32m8_rm_tum
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vf_f32m8_rm_tum(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f32m8_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vv_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmsac_vv_f64m1_rm_tum(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f64m1_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vf_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmsac_vf_f64m1_rm_tum(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f64m1_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vv_f64m2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmsac_vv_f64m2_rm_tum(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f64m2_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vf_f64m2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmsac_vf_f64m2_rm_tum(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f64m2_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vv_f64m4_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmsac_vv_f64m4_rm_tum(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f64m4_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vf_f64m4_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmsac_vf_f64m4_rm_tum(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f64m4_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vv_f64m8_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmsac_vv_f64m8_rm_tum(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f64m8_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vf_f64m8_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmsac_vf_f64m8_rm_tum(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f64m8_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_f32mf2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vv_f32mf2_rm_tumu(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f32mf2_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_f32mf2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vf_f32mf2_rm_tumu(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f32mf2_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_f32m1_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vv_f32m1_rm_tumu(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f32m1_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_f32m1_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vf_f32m1_rm_tumu(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f32m1_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_f32m2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vv_f32m2_rm_tumu(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f32m2_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_f32m2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vf_f32m2_rm_tumu(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f32m2_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_f32m4_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vv_f32m4_rm_tumu(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f32m4_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_f32m4_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vf_f32m4_rm_tumu(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f32m4_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_f32m8_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vv_f32m8_rm_tumu(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f32m8_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_f32m8_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vf_f32m8_rm_tumu(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f32m8_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vv_f64m1_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmsac_vv_f64m1_rm_tumu(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f64m1_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vf_f64m1_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmsac_vf_f64m1_rm_tumu(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f64m1_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vv_f64m2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmsac_vv_f64m2_rm_tumu(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f64m2_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vf_f64m2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmsac_vf_f64m2_rm_tumu(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f64m2_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vv_f64m4_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmsac_vv_f64m4_rm_tumu(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f64m4_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vf_f64m4_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmsac_vf_f64m4_rm_tumu(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f64m4_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vv_f64m8_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmsac_vv_f64m8_rm_tumu(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f64m8_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vf_f64m8_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmsac_vf_f64m8_rm_tumu(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f64m8_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_f32mf2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vv_f32mf2_rm_mu(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f32mf2_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_f32mf2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vf_f32mf2_rm_mu(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f32mf2_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_f32m1_rm_mu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vv_f32m1_rm_mu(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f32m1_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_f32m1_rm_mu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vf_f32m1_rm_mu(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f32m1_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_f32m2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vv_f32m2_rm_mu(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f32m2_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_f32m2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vf_f32m2_rm_mu(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f32m2_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_f32m4_rm_mu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vv_f32m4_rm_mu(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f32m4_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_f32m4_rm_mu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vf_f32m4_rm_mu(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f32m4_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_f32m8_rm_mu
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vv_f32m8_rm_mu(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f32m8_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_f32m8_rm_mu
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vf_f32m8_rm_mu(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f32m8_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vv_f64m1_rm_mu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmsac_vv_f64m1_rm_mu(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f64m1_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vf_f64m1_rm_mu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmsac_vf_f64m1_rm_mu(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f64m1_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vv_f64m2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmsac_vv_f64m2_rm_mu(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f64m2_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vf_f64m2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmsac_vf_f64m2_rm_mu(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f64m2_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vv_f64m4_rm_mu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmsac_vv_f64m4_rm_mu(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f64m4_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vf_f64m4_rm_mu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmsac_vf_f64m4_rm_mu(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f64m4_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vv_f64m8_rm_mu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmsac_vv_f64m8_rm_mu(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vv_f64m8_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vf_f64m8_rm_mu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmsac_vf_f64m8_rm_mu(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_vf_f64m8_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/vfwnmacc.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/vfwnmacc.c
--- a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/vfwnmacc.c
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/vfwnmacc.c
@@ -10,7 +10,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_f32mf2_tu
 // CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmacc_vv_f32mf2_tu(vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -20,7 +20,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_f32mf2_tu
 // CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmacc_vf_f32mf2_tu(vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -30,7 +30,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmacc_vv_f32m1_tu(vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -40,7 +40,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmacc_vf_f32m1_tu(vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -50,7 +50,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_f32m2_tu
 // CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmacc_vv_f32m2_tu(vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -60,7 +60,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_f32m2_tu
 // CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmacc_vf_f32m2_tu(vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -70,7 +70,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_f32m4_tu
 // CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmacc_vv_f32m4_tu(vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -80,7 +80,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_f32m4_tu
 // CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmacc_vf_f32m4_tu(vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -90,7 +90,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_f32m8_tu
 // CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmacc_vv_f32m8_tu(vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -100,7 +100,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_f32m8_tu
 // CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmacc_vf_f32m8_tu(vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -110,7 +110,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vv_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmacc_vv_f64m1_tu(vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -120,7 +120,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vf_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmacc_vf_f64m1_tu(vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -130,7 +130,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vv_f64m2_tu
 // CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmacc_vv_f64m2_tu(vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -140,7 +140,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vf_f64m2_tu
 // CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmacc_vf_f64m2_tu(vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -150,7 +150,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vv_f64m4_tu
 // CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmacc_vv_f64m4_tu(vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -160,7 +160,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vf_f64m4_tu
 // CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmacc_vf_f64m4_tu(vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -170,7 +170,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vv_f64m8_tu
 // CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmacc_vv_f64m8_tu(vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -180,7 +180,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vf_f64m8_tu
 // CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmacc_vf_f64m8_tu(vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
@@ -190,7 +190,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_f32mf2_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmacc_vv_f32mf2_tum(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -200,7 +200,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_f32mf2_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmacc_vf_f32mf2_tum(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -210,7 +210,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmacc_vv_f32m1_tum(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -220,7 +220,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmacc_vf_f32m1_tum(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -230,7 +230,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_f32m2_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmacc_vv_f32m2_tum(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -240,7 +240,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_f32m2_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmacc_vf_f32m2_tum(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -250,7 +250,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_f32m4_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmacc_vv_f32m4_tum(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -260,7 +260,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_f32m4_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmacc_vf_f32m4_tum(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -270,7 +270,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_f32m8_tum
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmacc_vv_f32m8_tum(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -280,7 +280,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_f32m8_tum
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmacc_vf_f32m8_tum(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -290,7 +290,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vv_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmacc_vv_f64m1_tum(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -300,7 +300,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vf_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmacc_vf_f64m1_tum(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -310,7 +310,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vv_f64m2_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmacc_vv_f64m2_tum(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -320,7 +320,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vf_f64m2_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmacc_vf_f64m2_tum(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -330,7 +330,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vv_f64m4_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmacc_vv_f64m4_tum(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -340,7 +340,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vf_f64m4_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmacc_vf_f64m4_tum(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -350,7 +350,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vv_f64m8_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmacc_vv_f64m8_tum(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -360,7 +360,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vf_f64m8_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmacc_vf_f64m8_tum(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
@@ -370,7 +370,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_f32mf2_tumu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmacc_vv_f32mf2_tumu(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -380,7 +380,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_f32mf2_tumu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmacc_vf_f32mf2_tumu(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -390,7 +390,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_f32m1_tumu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmacc_vv_f32m1_tumu(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -400,7 +400,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_f32m1_tumu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmacc_vf_f32m1_tumu(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -410,7 +410,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_f32m2_tumu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmacc_vv_f32m2_tumu(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -420,7 +420,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_f32m2_tumu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmacc_vf_f32m2_tumu(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -430,7 +430,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_f32m4_tumu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmacc_vv_f32m4_tumu(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -440,7 +440,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_f32m4_tumu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmacc_vf_f32m4_tumu(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -450,7 +450,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_f32m8_tumu
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmacc_vv_f32m8_tumu(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -460,7 +460,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_f32m8_tumu
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmacc_vf_f32m8_tumu(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -470,7 +470,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vv_f64m1_tumu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmacc_vv_f64m1_tumu(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -480,7 +480,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vf_f64m1_tumu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmacc_vf_f64m1_tumu(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -490,7 +490,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vv_f64m2_tumu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmacc_vv_f64m2_tumu(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -500,7 +500,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vf_f64m2_tumu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmacc_vf_f64m2_tumu(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -510,7 +510,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vv_f64m4_tumu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmacc_vv_f64m4_tumu(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -520,7 +520,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vf_f64m4_tumu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmacc_vf_f64m4_tumu(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -530,7 +530,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vv_f64m8_tumu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmacc_vv_f64m8_tumu(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -540,7 +540,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vf_f64m8_tumu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmacc_vf_f64m8_tumu(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
@@ -550,7 +550,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_f32mf2_mu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmacc_vv_f32mf2_mu(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -560,7 +560,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_f32mf2_mu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmacc_vf_f32mf2_mu(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -570,7 +570,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_f32m1_mu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmacc_vv_f32m1_mu(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -580,7 +580,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_f32m1_mu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmacc_vf_f32m1_mu(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -590,7 +590,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_f32m2_mu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmacc_vv_f32m2_mu(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -600,7 +600,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_f32m2_mu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmacc_vf_f32m2_mu(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -610,7 +610,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_f32m4_mu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmacc_vv_f32m4_mu(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -620,7 +620,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_f32m4_mu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmacc_vf_f32m4_mu(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -630,7 +630,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_f32m8_mu
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmacc_vv_f32m8_mu(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -640,7 +640,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_f32m8_mu
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmacc_vf_f32m8_mu(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -650,7 +650,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vv_f64m1_mu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmacc_vv_f64m1_mu(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -660,7 +660,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vf_f64m1_mu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmacc_vf_f64m1_mu(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -670,7 +670,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vv_f64m2_mu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmacc_vv_f64m2_mu(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -680,7 +680,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vf_f64m2_mu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmacc_vf_f64m2_mu(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -690,7 +690,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vv_f64m4_mu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmacc_vv_f64m4_mu(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -700,7 +700,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vf_f64m4_mu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmacc_vf_f64m4_mu(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -710,7 +710,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vv_f64m8_mu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmacc_vv_f64m8_mu(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -720,10 +720,730 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vf_f64m8_mu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmacc_vf_f64m8_mu(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
   return __riscv_vfwnmacc_vf_f64m8_mu(mask, vd, vs1, vs2, vl);
 }
 
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_f32mf2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vv_f32mf2_rm_tu(vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f32mf2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_f32mf2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vf_f32mf2_rm_tu(vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f32mf2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vv_f32m1_rm_tu(vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f32m1_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vf_f32m1_rm_tu(vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f32m1_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_f32m2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vv_f32m2_rm_tu(vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f32m2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_f32m2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vf_f32m2_rm_tu(vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f32m2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_f32m4_rm_tu
+// CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vv_f32m4_rm_tu(vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f32m4_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_f32m4_rm_tu
+// CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vf_f32m4_rm_tu(vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f32m4_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_f32m8_rm_tu
+// CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vv_f32m8_rm_tu(vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f32m8_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_f32m8_rm_tu
+// CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vf_f32m8_rm_tu(vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f32m8_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vv_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmacc_vv_f64m1_rm_tu(vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f64m1_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vf_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmacc_vf_f64m1_rm_tu(vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f64m1_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vv_f64m2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmacc_vv_f64m2_rm_tu(vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f64m2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vf_f64m2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmacc_vf_f64m2_rm_tu(vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f64m2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vv_f64m4_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmacc_vv_f64m4_rm_tu(vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f64m4_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vf_f64m4_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmacc_vf_f64m4_rm_tu(vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f64m4_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vv_f64m8_rm_tu
+// CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmacc_vv_f64m8_rm_tu(vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f64m8_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vf_f64m8_rm_tu
+// CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmacc_vf_f64m8_rm_tu(vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f64m8_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_f32mf2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vv_f32mf2_rm_tum(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f32mf2_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_f32mf2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vf_f32mf2_rm_tum(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f32mf2_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vv_f32m1_rm_tum(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f32m1_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vf_f32m1_rm_tum(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f32m1_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_f32m2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vv_f32m2_rm_tum(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f32m2_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_f32m2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vf_f32m2_rm_tum(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f32m2_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_f32m4_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vv_f32m4_rm_tum(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f32m4_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_f32m4_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vf_f32m4_rm_tum(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f32m4_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_f32m8_rm_tum
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vv_f32m8_rm_tum(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f32m8_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_f32m8_rm_tum
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vf_f32m8_rm_tum(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f32m8_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vv_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmacc_vv_f64m1_rm_tum(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f64m1_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vf_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmacc_vf_f64m1_rm_tum(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f64m1_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vv_f64m2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmacc_vv_f64m2_rm_tum(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f64m2_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vf_f64m2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmacc_vf_f64m2_rm_tum(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f64m2_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vv_f64m4_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmacc_vv_f64m4_rm_tum(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f64m4_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vf_f64m4_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmacc_vf_f64m4_rm_tum(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f64m4_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vv_f64m8_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmacc_vv_f64m8_rm_tum(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f64m8_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vf_f64m8_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmacc_vf_f64m8_rm_tum(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f64m8_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_f32mf2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vv_f32mf2_rm_tumu(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f32mf2_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_f32mf2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vf_f32mf2_rm_tumu(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f32mf2_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_f32m1_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vv_f32m1_rm_tumu(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f32m1_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_f32m1_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vf_f32m1_rm_tumu(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f32m1_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_f32m2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vv_f32m2_rm_tumu(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f32m2_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_f32m2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vf_f32m2_rm_tumu(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f32m2_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_f32m4_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vv_f32m4_rm_tumu(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f32m4_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_f32m4_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vf_f32m4_rm_tumu(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f32m4_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_f32m8_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vv_f32m8_rm_tumu(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f32m8_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_f32m8_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vf_f32m8_rm_tumu(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f32m8_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vv_f64m1_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmacc_vv_f64m1_rm_tumu(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f64m1_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vf_f64m1_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmacc_vf_f64m1_rm_tumu(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f64m1_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vv_f64m2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmacc_vv_f64m2_rm_tumu(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f64m2_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vf_f64m2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmacc_vf_f64m2_rm_tumu(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f64m2_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vv_f64m4_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmacc_vv_f64m4_rm_tumu(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f64m4_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vf_f64m4_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmacc_vf_f64m4_rm_tumu(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f64m4_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vv_f64m8_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmacc_vv_f64m8_rm_tumu(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f64m8_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vf_f64m8_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmacc_vf_f64m8_rm_tumu(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f64m8_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_f32mf2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vv_f32mf2_rm_mu(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f32mf2_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_f32mf2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vf_f32mf2_rm_mu(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f32mf2_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_f32m1_rm_mu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vv_f32m1_rm_mu(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f32m1_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_f32m1_rm_mu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vf_f32m1_rm_mu(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f32m1_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_f32m2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vv_f32m2_rm_mu(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f32m2_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_f32m2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vf_f32m2_rm_mu(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f32m2_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_f32m4_rm_mu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vv_f32m4_rm_mu(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f32m4_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_f32m4_rm_mu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vf_f32m4_rm_mu(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f32m4_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_f32m8_rm_mu
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vv_f32m8_rm_mu(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f32m8_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_f32m8_rm_mu
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vf_f32m8_rm_mu(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f32m8_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vv_f64m1_rm_mu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmacc_vv_f64m1_rm_mu(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f64m1_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vf_f64m1_rm_mu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmacc_vf_f64m1_rm_mu(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f64m1_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vv_f64m2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmacc_vv_f64m2_rm_mu(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f64m2_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vf_f64m2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmacc_vf_f64m2_rm_mu(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f64m2_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vv_f64m4_rm_mu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmacc_vv_f64m4_rm_mu(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f64m4_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vf_f64m4_rm_mu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmacc_vf_f64m4_rm_mu(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f64m4_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vv_f64m8_rm_mu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmacc_vv_f64m8_rm_mu(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vv_f64m8_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vf_f64m8_rm_mu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmacc_vf_f64m8_rm_mu(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_vf_f64m8_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/vfwnmsac.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/vfwnmsac.c
--- a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/vfwnmsac.c
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/non-overloaded/vfwnmsac.c
@@ -10,7 +10,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_f32mf2_tu
 // CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmsac_vv_f32mf2_tu(vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -20,7 +20,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_f32mf2_tu
 // CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmsac_vf_f32mf2_tu(vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -30,7 +30,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmsac_vv_f32m1_tu(vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -40,7 +40,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmsac_vf_f32m1_tu(vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -50,7 +50,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_f32m2_tu
 // CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmsac_vv_f32m2_tu(vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -60,7 +60,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_f32m2_tu
 // CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmsac_vf_f32m2_tu(vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -70,7 +70,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_f32m4_tu
 // CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmsac_vv_f32m4_tu(vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -80,7 +80,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_f32m4_tu
 // CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmsac_vf_f32m4_tu(vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -90,7 +90,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_f32m8_tu
 // CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmsac_vv_f32m8_tu(vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -100,7 +100,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_f32m8_tu
 // CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmsac_vf_f32m8_tu(vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -110,7 +110,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vv_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmsac_vv_f64m1_tu(vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -120,7 +120,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vf_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmsac_vf_f64m1_tu(vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -130,7 +130,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vv_f64m2_tu
 // CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmsac_vv_f64m2_tu(vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -140,7 +140,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vf_f64m2_tu
 // CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmsac_vf_f64m2_tu(vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -150,7 +150,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vv_f64m4_tu
 // CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmsac_vv_f64m4_tu(vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -160,7 +160,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vf_f64m4_tu
 // CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmsac_vf_f64m4_tu(vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -170,7 +170,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vv_f64m8_tu
 // CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmsac_vv_f64m8_tu(vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -180,7 +180,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vf_f64m8_tu
 // CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmsac_vf_f64m8_tu(vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
@@ -190,7 +190,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_f32mf2_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmsac_vv_f32mf2_tum(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -200,7 +200,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_f32mf2_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmsac_vf_f32mf2_tum(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -210,7 +210,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmsac_vv_f32m1_tum(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -220,7 +220,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmsac_vf_f32m1_tum(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -230,7 +230,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_f32m2_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmsac_vv_f32m2_tum(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -240,7 +240,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_f32m2_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmsac_vf_f32m2_tum(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -250,7 +250,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_f32m4_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmsac_vv_f32m4_tum(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -260,7 +260,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_f32m4_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmsac_vf_f32m4_tum(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -270,7 +270,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_f32m8_tum
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmsac_vv_f32m8_tum(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -280,7 +280,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_f32m8_tum
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmsac_vf_f32m8_tum(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -290,7 +290,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vv_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmsac_vv_f64m1_tum(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -300,7 +300,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vf_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmsac_vf_f64m1_tum(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -310,7 +310,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vv_f64m2_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmsac_vv_f64m2_tum(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -320,7 +320,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vf_f64m2_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmsac_vf_f64m2_tum(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -330,7 +330,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vv_f64m4_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmsac_vv_f64m4_tum(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -340,7 +340,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vf_f64m4_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmsac_vf_f64m4_tum(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -350,7 +350,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vv_f64m8_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmsac_vv_f64m8_tum(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -360,7 +360,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vf_f64m8_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmsac_vf_f64m8_tum(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
@@ -370,7 +370,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_f32mf2_tumu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmsac_vv_f32mf2_tumu(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -380,7 +380,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_f32mf2_tumu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmsac_vf_f32mf2_tumu(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -390,7 +390,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_f32m1_tumu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmsac_vv_f32m1_tumu(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -400,7 +400,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_f32m1_tumu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmsac_vf_f32m1_tumu(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -410,7 +410,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_f32m2_tumu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmsac_vv_f32m2_tumu(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -420,7 +420,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_f32m2_tumu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmsac_vf_f32m2_tumu(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -430,7 +430,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_f32m4_tumu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmsac_vv_f32m4_tumu(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -440,7 +440,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_f32m4_tumu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmsac_vf_f32m4_tumu(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -450,7 +450,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_f32m8_tumu
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmsac_vv_f32m8_tumu(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -460,7 +460,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_f32m8_tumu
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmsac_vf_f32m8_tumu(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -470,7 +470,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vv_f64m1_tumu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmsac_vv_f64m1_tumu(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -480,7 +480,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vf_f64m1_tumu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmsac_vf_f64m1_tumu(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -490,7 +490,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vv_f64m2_tumu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmsac_vv_f64m2_tumu(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -500,7 +500,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vf_f64m2_tumu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmsac_vf_f64m2_tumu(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -510,7 +510,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vv_f64m4_tumu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmsac_vv_f64m4_tumu(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -520,7 +520,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vf_f64m4_tumu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmsac_vf_f64m4_tumu(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -530,7 +530,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vv_f64m8_tumu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmsac_vv_f64m8_tumu(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -540,7 +540,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vf_f64m8_tumu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmsac_vf_f64m8_tumu(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
@@ -550,7 +550,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_f32mf2_mu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmsac_vv_f32mf2_mu(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -560,7 +560,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_f32mf2_mu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmsac_vf_f32mf2_mu(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -570,7 +570,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_f32m1_mu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmsac_vv_f32m1_mu(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -580,7 +580,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_f32m1_mu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmsac_vf_f32m1_mu(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -590,7 +590,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_f32m2_mu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmsac_vv_f32m2_mu(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -600,7 +600,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_f32m2_mu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmsac_vf_f32m2_mu(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -610,7 +610,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_f32m4_mu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmsac_vv_f32m4_mu(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -620,7 +620,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_f32m4_mu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmsac_vf_f32m4_mu(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -630,7 +630,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_f32m8_mu
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmsac_vv_f32m8_mu(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -640,7 +640,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_f32m8_mu
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmsac_vf_f32m8_mu(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -650,7 +650,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vv_f64m1_mu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmsac_vv_f64m1_mu(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -660,7 +660,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vf_f64m1_mu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmsac_vf_f64m1_mu(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -670,7 +670,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vv_f64m2_mu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmsac_vv_f64m2_mu(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -680,7 +680,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vf_f64m2_mu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmsac_vf_f64m2_mu(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -690,7 +690,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vv_f64m4_mu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmsac_vv_f64m4_mu(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -700,7 +700,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vf_f64m4_mu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmsac_vf_f64m4_mu(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -710,7 +710,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vv_f64m8_mu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmsac_vv_f64m8_mu(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -720,10 +720,730 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vf_f64m8_mu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmsac_vf_f64m8_mu(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
   return __riscv_vfwnmsac_vf_f64m8_mu(mask, vd, vs1, vs2, vl);
 }
 
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_f32mf2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vv_f32mf2_rm_tu(vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f32mf2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_f32mf2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vf_f32mf2_rm_tu(vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f32mf2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vv_f32m1_rm_tu(vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f32m1_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vf_f32m1_rm_tu(vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f32m1_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_f32m2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vv_f32m2_rm_tu(vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f32m2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_f32m2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vf_f32m2_rm_tu(vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f32m2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_f32m4_rm_tu
+// CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vv_f32m4_rm_tu(vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f32m4_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_f32m4_rm_tu
+// CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vf_f32m4_rm_tu(vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f32m4_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_f32m8_rm_tu
+// CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vv_f32m8_rm_tu(vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f32m8_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_f32m8_rm_tu
+// CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vf_f32m8_rm_tu(vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f32m8_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vv_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmsac_vv_f64m1_rm_tu(vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f64m1_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vf_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmsac_vf_f64m1_rm_tu(vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f64m1_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vv_f64m2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmsac_vv_f64m2_rm_tu(vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f64m2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vf_f64m2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmsac_vf_f64m2_rm_tu(vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f64m2_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vv_f64m4_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmsac_vv_f64m4_rm_tu(vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f64m4_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vf_f64m4_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmsac_vf_f64m4_rm_tu(vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f64m4_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vv_f64m8_rm_tu
+// CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmsac_vv_f64m8_rm_tu(vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f64m8_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vf_f64m8_rm_tu
+// CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmsac_vf_f64m8_rm_tu(vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f64m8_rm_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_f32mf2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vv_f32mf2_rm_tum(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f32mf2_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_f32mf2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vf_f32mf2_rm_tum(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f32mf2_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vv_f32m1_rm_tum(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f32m1_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vf_f32m1_rm_tum(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f32m1_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_f32m2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vv_f32m2_rm_tum(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f32m2_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_f32m2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vf_f32m2_rm_tum(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f32m2_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_f32m4_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vv_f32m4_rm_tum(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f32m4_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_f32m4_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vf_f32m4_rm_tum(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f32m4_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_f32m8_rm_tum
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vv_f32m8_rm_tum(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f32m8_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_f32m8_rm_tum
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vf_f32m8_rm_tum(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f32m8_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vv_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmsac_vv_f64m1_rm_tum(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f64m1_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vf_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmsac_vf_f64m1_rm_tum(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f64m1_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vv_f64m2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmsac_vv_f64m2_rm_tum(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f64m2_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vf_f64m2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmsac_vf_f64m2_rm_tum(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f64m2_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vv_f64m4_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmsac_vv_f64m4_rm_tum(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f64m4_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vf_f64m4_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmsac_vf_f64m4_rm_tum(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f64m4_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vv_f64m8_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmsac_vv_f64m8_rm_tum(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f64m8_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vf_f64m8_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmsac_vf_f64m8_rm_tum(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f64m8_rm_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_f32mf2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vv_f32mf2_rm_tumu(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f32mf2_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_f32mf2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vf_f32mf2_rm_tumu(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f32mf2_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_f32m1_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vv_f32m1_rm_tumu(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f32m1_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_f32m1_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vf_f32m1_rm_tumu(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f32m1_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_f32m2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vv_f32m2_rm_tumu(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f32m2_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_f32m2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vf_f32m2_rm_tumu(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f32m2_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_f32m4_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vv_f32m4_rm_tumu(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f32m4_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_f32m4_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vf_f32m4_rm_tumu(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f32m4_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_f32m8_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vv_f32m8_rm_tumu(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f32m8_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_f32m8_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vf_f32m8_rm_tumu(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f32m8_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vv_f64m1_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmsac_vv_f64m1_rm_tumu(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f64m1_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vf_f64m1_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmsac_vf_f64m1_rm_tumu(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f64m1_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vv_f64m2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmsac_vv_f64m2_rm_tumu(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f64m2_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vf_f64m2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmsac_vf_f64m2_rm_tumu(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f64m2_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vv_f64m4_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmsac_vv_f64m4_rm_tumu(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f64m4_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vf_f64m4_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmsac_vf_f64m4_rm_tumu(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f64m4_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vv_f64m8_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmsac_vv_f64m8_rm_tumu(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f64m8_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vf_f64m8_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmsac_vf_f64m8_rm_tumu(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f64m8_rm_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_f32mf2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vv_f32mf2_rm_mu(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f32mf2_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_f32mf2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vf_f32mf2_rm_mu(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f32mf2_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_f32m1_rm_mu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vv_f32m1_rm_mu(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f32m1_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_f32m1_rm_mu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vf_f32m1_rm_mu(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f32m1_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_f32m2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vv_f32m2_rm_mu(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f32m2_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_f32m2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vf_f32m2_rm_mu(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f32m2_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_f32m4_rm_mu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vv_f32m4_rm_mu(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f32m4_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_f32m4_rm_mu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vf_f32m4_rm_mu(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f32m4_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_f32m8_rm_mu
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vv_f32m8_rm_mu(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f32m8_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_f32m8_rm_mu
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vf_f32m8_rm_mu(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f32m8_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vv_f64m1_rm_mu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmsac_vv_f64m1_rm_mu(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f64m1_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vf_f64m1_rm_mu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmsac_vf_f64m1_rm_mu(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f64m1_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vv_f64m2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmsac_vv_f64m2_rm_mu(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f64m2_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vf_f64m2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmsac_vf_f64m2_rm_mu(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f64m2_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vv_f64m4_rm_mu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmsac_vv_f64m4_rm_mu(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f64m4_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vf_f64m4_rm_mu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmsac_vf_f64m4_rm_mu(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f64m4_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vv_f64m8_rm_mu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmsac_vv_f64m8_rm_mu(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vv_f64m8_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vf_f64m8_rm_mu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmsac_vf_f64m8_rm_mu(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_vf_f64m8_rm_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/vfwmacc.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/vfwmacc.c
--- a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/vfwmacc.c
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/vfwmacc.c
@@ -10,7 +10,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_f32mf2_tu
 // CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmacc_vv_f32mf2_tu(vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -20,7 +20,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_f32mf2_tu
 // CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmacc_vf_f32mf2_tu(vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -30,7 +30,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmacc_vv_f32m1_tu(vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -40,7 +40,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmacc_vf_f32m1_tu(vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -50,7 +50,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_f32m2_tu
 // CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmacc_vv_f32m2_tu(vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -60,7 +60,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_f32m2_tu
 // CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmacc_vf_f32m2_tu(vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -70,7 +70,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_f32m4_tu
 // CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmacc_vv_f32m4_tu(vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -80,7 +80,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_f32m4_tu
 // CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmacc_vf_f32m4_tu(vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -90,7 +90,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_f32m8_tu
 // CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmacc_vv_f32m8_tu(vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -100,7 +100,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_f32m8_tu
 // CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmacc_vf_f32m8_tu(vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -110,7 +110,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vv_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmacc_vv_f64m1_tu(vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -120,7 +120,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vf_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmacc_vf_f64m1_tu(vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -130,7 +130,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vv_f64m2_tu
 // CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmacc_vv_f64m2_tu(vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -140,7 +140,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vf_f64m2_tu
 // CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmacc_vf_f64m2_tu(vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -150,7 +150,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vv_f64m4_tu
 // CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmacc_vv_f64m4_tu(vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -160,7 +160,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vf_f64m4_tu
 // CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmacc_vf_f64m4_tu(vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -170,7 +170,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vv_f64m8_tu
 // CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmacc_vv_f64m8_tu(vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -180,7 +180,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vf_f64m8_tu
 // CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmacc_vf_f64m8_tu(vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
@@ -190,7 +190,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_f32mf2_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmacc_vv_f32mf2_tum(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -200,7 +200,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_f32mf2_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmacc_vf_f32mf2_tum(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -210,7 +210,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmacc_vv_f32m1_tum(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -220,7 +220,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmacc_vf_f32m1_tum(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -230,7 +230,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_f32m2_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmacc_vv_f32m2_tum(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -240,7 +240,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_f32m2_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmacc_vf_f32m2_tum(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -250,7 +250,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_f32m4_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmacc_vv_f32m4_tum(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -260,7 +260,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_f32m4_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmacc_vf_f32m4_tum(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -270,7 +270,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_f32m8_tum
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmacc_vv_f32m8_tum(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -280,7 +280,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_f32m8_tum
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmacc_vf_f32m8_tum(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -290,7 +290,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vv_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmacc_vv_f64m1_tum(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -300,7 +300,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vf_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmacc_vf_f64m1_tum(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -310,7 +310,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vv_f64m2_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmacc_vv_f64m2_tum(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -320,7 +320,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vf_f64m2_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmacc_vf_f64m2_tum(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -330,7 +330,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vv_f64m4_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmacc_vv_f64m4_tum(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -340,7 +340,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vf_f64m4_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmacc_vf_f64m4_tum(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -350,7 +350,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vv_f64m8_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmacc_vv_f64m8_tum(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -360,7 +360,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vf_f64m8_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmacc_vf_f64m8_tum(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
@@ -370,7 +370,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_f32mf2_tumu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmacc_vv_f32mf2_tumu(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -380,7 +380,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_f32mf2_tumu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmacc_vf_f32mf2_tumu(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -390,7 +390,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_f32m1_tumu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmacc_vv_f32m1_tumu(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -400,7 +400,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_f32m1_tumu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmacc_vf_f32m1_tumu(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -410,7 +410,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_f32m2_tumu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmacc_vv_f32m2_tumu(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -420,7 +420,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_f32m2_tumu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmacc_vf_f32m2_tumu(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -430,7 +430,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_f32m4_tumu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmacc_vv_f32m4_tumu(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -440,7 +440,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_f32m4_tumu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmacc_vf_f32m4_tumu(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -450,7 +450,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_f32m8_tumu
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmacc_vv_f32m8_tumu(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -460,7 +460,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_f32m8_tumu
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmacc_vf_f32m8_tumu(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -470,7 +470,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vv_f64m1_tumu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmacc_vv_f64m1_tumu(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -480,7 +480,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vf_f64m1_tumu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmacc_vf_f64m1_tumu(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -490,7 +490,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vv_f64m2_tumu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmacc_vv_f64m2_tumu(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -500,7 +500,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vf_f64m2_tumu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmacc_vf_f64m2_tumu(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -510,7 +510,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vv_f64m4_tumu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmacc_vv_f64m4_tumu(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -520,7 +520,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vf_f64m4_tumu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmacc_vf_f64m4_tumu(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -530,7 +530,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vv_f64m8_tumu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmacc_vv_f64m8_tumu(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -540,7 +540,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vf_f64m8_tumu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmacc_vf_f64m8_tumu(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
@@ -550,7 +550,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_f32mf2_mu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmacc_vv_f32mf2_mu(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -560,7 +560,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_f32mf2_mu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmacc_vf_f32mf2_mu(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -570,7 +570,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_f32m1_mu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmacc_vv_f32m1_mu(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -580,7 +580,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_f32m1_mu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmacc_vf_f32m1_mu(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -590,7 +590,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_f32m2_mu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmacc_vv_f32m2_mu(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -600,7 +600,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_f32m2_mu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmacc_vf_f32m2_mu(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -610,7 +610,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_f32m4_mu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmacc_vv_f32m4_mu(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -620,7 +620,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_f32m4_mu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmacc_vf_f32m4_mu(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -630,7 +630,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_f32m8_mu
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmacc_vv_f32m8_mu(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -640,7 +640,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_f32m8_mu
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmacc_vf_f32m8_mu(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -650,7 +650,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vv_f64m1_mu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmacc_vv_f64m1_mu(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -660,7 +660,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vf_f64m1_mu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmacc_vf_f64m1_mu(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -670,7 +670,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vv_f64m2_mu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmacc_vv_f64m2_mu(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -680,7 +680,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vf_f64m2_mu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmacc_vf_f64m2_mu(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -690,7 +690,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vv_f64m4_mu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmacc_vv_f64m4_mu(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -700,7 +700,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vf_f64m4_mu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmacc_vf_f64m4_mu(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -710,7 +710,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vv_f64m8_mu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmacc_vv_f64m8_mu(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -720,10 +720,730 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vf_f64m8_mu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmacc_vf_f64m8_mu(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
   return __riscv_vfwmacc_mu(mask, vd, vs1, vs2, vl);
 }
 
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_f32mf2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vv_f32mf2_rm_tu(vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_f32mf2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vf_f32mf2_rm_tu(vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vv_f32m1_rm_tu(vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vf_f32m1_rm_tu(vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_f32m2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vv_f32m2_rm_tu(vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_f32m2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vf_f32m2_rm_tu(vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_f32m4_rm_tu
+// CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vv_f32m4_rm_tu(vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_f32m4_rm_tu
+// CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vf_f32m4_rm_tu(vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_f32m8_rm_tu
+// CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vv_f32m8_rm_tu(vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_f32m8_rm_tu
+// CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vf_f32m8_rm_tu(vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vv_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmacc_vv_f64m1_rm_tu(vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vf_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmacc_vf_f64m1_rm_tu(vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vv_f64m2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmacc_vv_f64m2_rm_tu(vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vf_f64m2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmacc_vf_f64m2_rm_tu(vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vv_f64m4_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmacc_vv_f64m4_rm_tu(vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vf_f64m4_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmacc_vf_f64m4_rm_tu(vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vv_f64m8_rm_tu
+// CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmacc_vv_f64m8_rm_tu(vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vf_f64m8_rm_tu
+// CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmacc_vf_f64m8_rm_tu(vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_f32mf2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vv_f32mf2_rm_tum(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_f32mf2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vf_f32mf2_rm_tum(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vv_f32m1_rm_tum(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vf_f32m1_rm_tum(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_f32m2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vv_f32m2_rm_tum(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_f32m2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vf_f32m2_rm_tum(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_f32m4_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vv_f32m4_rm_tum(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_f32m4_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vf_f32m4_rm_tum(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_f32m8_rm_tum
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vv_f32m8_rm_tum(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_f32m8_rm_tum
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vf_f32m8_rm_tum(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vv_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmacc_vv_f64m1_rm_tum(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vf_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmacc_vf_f64m1_rm_tum(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vv_f64m2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmacc_vv_f64m2_rm_tum(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vf_f64m2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmacc_vf_f64m2_rm_tum(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vv_f64m4_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmacc_vv_f64m4_rm_tum(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vf_f64m4_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmacc_vf_f64m4_rm_tum(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vv_f64m8_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmacc_vv_f64m8_rm_tum(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vf_f64m8_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmacc_vf_f64m8_rm_tum(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_f32mf2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vv_f32mf2_rm_tumu(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_f32mf2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vf_f32mf2_rm_tumu(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_f32m1_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vv_f32m1_rm_tumu(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_f32m1_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vf_f32m1_rm_tumu(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_f32m2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vv_f32m2_rm_tumu(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_f32m2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vf_f32m2_rm_tumu(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_f32m4_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vv_f32m4_rm_tumu(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_f32m4_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vf_f32m4_rm_tumu(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_f32m8_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vv_f32m8_rm_tumu(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_f32m8_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vf_f32m8_rm_tumu(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vv_f64m1_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmacc_vv_f64m1_rm_tumu(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vf_f64m1_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmacc_vf_f64m1_rm_tumu(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vv_f64m2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmacc_vv_f64m2_rm_tumu(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vf_f64m2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmacc_vf_f64m2_rm_tumu(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vv_f64m4_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmacc_vv_f64m4_rm_tumu(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vf_f64m4_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmacc_vf_f64m4_rm_tumu(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vv_f64m8_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmacc_vv_f64m8_rm_tumu(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vf_f64m8_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmacc_vf_f64m8_rm_tumu(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vv_f32mf2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vv_f32mf2_rm_mu(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmacc_vf_f32mf2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmacc_vf_f32mf2_rm_mu(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vv_f32m1_rm_mu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vv_f32m1_rm_mu(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmacc_vf_f32m1_rm_mu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmacc_vf_f32m1_rm_mu(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vv_f32m2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vv_f32m2_rm_mu(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmacc_vf_f32m2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmacc_vf_f32m2_rm_mu(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vv_f32m4_rm_mu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vv_f32m4_rm_mu(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmacc_vf_f32m4_rm_mu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmacc_vf_f32m4_rm_mu(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vv_f32m8_rm_mu
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vv_f32m8_rm_mu(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmacc_vf_f32m8_rm_mu
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmacc_vf_f32m8_rm_mu(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vv_f64m1_rm_mu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmacc_vv_f64m1_rm_mu(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmacc_vf_f64m1_rm_mu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmacc_vf_f64m1_rm_mu(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vv_f64m2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmacc_vv_f64m2_rm_mu(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmacc_vf_f64m2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmacc_vf_f64m2_rm_mu(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmacc_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vv_f64m4_rm_mu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmacc_vv_f64m4_rm_mu(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmacc_vf_f64m4_rm_mu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmacc_vf_f64m4_rm_mu(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmacc_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vv_f64m8_rm_mu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmacc_vv_f64m8_rm_mu(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmacc_vf_f64m8_rm_mu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmacc_vf_f64m8_rm_mu(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmacc_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/vfwmsac.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/vfwmsac.c
--- a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/vfwmsac.c
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/vfwmsac.c
@@ -10,7 +10,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_f32mf2_tu
 // CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmsac_vv_f32mf2_tu(vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -20,7 +20,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_f32mf2_tu
 // CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmsac_vf_f32mf2_tu(vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -30,7 +30,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmsac_vv_f32m1_tu(vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -40,7 +40,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmsac_vf_f32m1_tu(vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -50,7 +50,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_f32m2_tu
 // CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmsac_vv_f32m2_tu(vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -60,7 +60,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_f32m2_tu
 // CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmsac_vf_f32m2_tu(vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -70,7 +70,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_f32m4_tu
 // CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmsac_vv_f32m4_tu(vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -80,7 +80,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_f32m4_tu
 // CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmsac_vf_f32m4_tu(vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -90,7 +90,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_f32m8_tu
 // CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmsac_vv_f32m8_tu(vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -100,7 +100,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_f32m8_tu
 // CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmsac_vf_f32m8_tu(vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -110,7 +110,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vv_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmsac_vv_f64m1_tu(vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -120,7 +120,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vf_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmsac_vf_f64m1_tu(vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -130,7 +130,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vv_f64m2_tu
 // CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmsac_vv_f64m2_tu(vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -140,7 +140,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vf_f64m2_tu
 // CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmsac_vf_f64m2_tu(vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -150,7 +150,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vv_f64m4_tu
 // CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmsac_vv_f64m4_tu(vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -160,7 +160,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vf_f64m4_tu
 // CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmsac_vf_f64m4_tu(vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -170,7 +170,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vv_f64m8_tu
 // CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmsac_vv_f64m8_tu(vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -180,7 +180,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vf_f64m8_tu
 // CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmsac_vf_f64m8_tu(vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
@@ -190,7 +190,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_f32mf2_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmsac_vv_f32mf2_tum(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -200,7 +200,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_f32mf2_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmsac_vf_f32mf2_tum(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -210,7 +210,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmsac_vv_f32m1_tum(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -220,7 +220,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmsac_vf_f32m1_tum(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -230,7 +230,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_f32m2_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmsac_vv_f32m2_tum(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -240,7 +240,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_f32m2_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmsac_vf_f32m2_tum(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -250,7 +250,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_f32m4_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmsac_vv_f32m4_tum(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -260,7 +260,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_f32m4_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmsac_vf_f32m4_tum(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -270,7 +270,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_f32m8_tum
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmsac_vv_f32m8_tum(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -280,7 +280,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_f32m8_tum
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmsac_vf_f32m8_tum(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -290,7 +290,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vv_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmsac_vv_f64m1_tum(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -300,7 +300,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vf_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmsac_vf_f64m1_tum(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -310,7 +310,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vv_f64m2_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmsac_vv_f64m2_tum(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -320,7 +320,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vf_f64m2_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmsac_vf_f64m2_tum(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -330,7 +330,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vv_f64m4_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmsac_vv_f64m4_tum(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -340,7 +340,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vf_f64m4_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmsac_vf_f64m4_tum(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -350,7 +350,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vv_f64m8_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmsac_vv_f64m8_tum(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -360,7 +360,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vf_f64m8_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmsac_vf_f64m8_tum(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
@@ -370,7 +370,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_f32mf2_tumu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmsac_vv_f32mf2_tumu(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -380,7 +380,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_f32mf2_tumu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmsac_vf_f32mf2_tumu(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -390,7 +390,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_f32m1_tumu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmsac_vv_f32m1_tumu(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -400,7 +400,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_f32m1_tumu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmsac_vf_f32m1_tumu(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -410,7 +410,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_f32m2_tumu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmsac_vv_f32m2_tumu(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -420,7 +420,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_f32m2_tumu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmsac_vf_f32m2_tumu(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -430,7 +430,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_f32m4_tumu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmsac_vv_f32m4_tumu(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -440,7 +440,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_f32m4_tumu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmsac_vf_f32m4_tumu(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -450,7 +450,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_f32m8_tumu
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmsac_vv_f32m8_tumu(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -460,7 +460,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_f32m8_tumu
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmsac_vf_f32m8_tumu(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -470,7 +470,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vv_f64m1_tumu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmsac_vv_f64m1_tumu(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -480,7 +480,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vf_f64m1_tumu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmsac_vf_f64m1_tumu(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -490,7 +490,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vv_f64m2_tumu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmsac_vv_f64m2_tumu(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -500,7 +500,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vf_f64m2_tumu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmsac_vf_f64m2_tumu(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -510,7 +510,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vv_f64m4_tumu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmsac_vv_f64m4_tumu(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -520,7 +520,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vf_f64m4_tumu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmsac_vf_f64m4_tumu(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -530,7 +530,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vv_f64m8_tumu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmsac_vv_f64m8_tumu(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -540,7 +540,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vf_f64m8_tumu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmsac_vf_f64m8_tumu(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
@@ -550,7 +550,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_f32mf2_mu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmsac_vv_f32mf2_mu(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -560,7 +560,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_f32mf2_mu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwmsac_vf_f32mf2_mu(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -570,7 +570,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_f32m1_mu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmsac_vv_f32m1_mu(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -580,7 +580,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_f32m1_mu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwmsac_vf_f32m1_mu(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -590,7 +590,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_f32m2_mu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmsac_vv_f32m2_mu(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -600,7 +600,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_f32m2_mu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwmsac_vf_f32m2_mu(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -610,7 +610,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_f32m4_mu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmsac_vv_f32m4_mu(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -620,7 +620,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_f32m4_mu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwmsac_vf_f32m4_mu(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -630,7 +630,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_f32m8_mu
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmsac_vv_f32m8_mu(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -640,7 +640,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_f32m8_mu
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwmsac_vf_f32m8_mu(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -650,7 +650,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vv_f64m1_mu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmsac_vv_f64m1_mu(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -660,7 +660,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vf_f64m1_mu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwmsac_vf_f64m1_mu(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -670,7 +670,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vv_f64m2_mu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmsac_vv_f64m2_mu(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -680,7 +680,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vf_f64m2_mu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwmsac_vf_f64m2_mu(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -690,7 +690,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vv_f64m4_mu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmsac_vv_f64m4_mu(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -700,7 +700,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vf_f64m4_mu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwmsac_vf_f64m4_mu(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -710,7 +710,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vv_f64m8_mu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmsac_vv_f64m8_mu(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -720,10 +720,730 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vf_f64m8_mu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwmsac_vf_f64m8_mu(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
   return __riscv_vfwmsac_mu(mask, vd, vs1, vs2, vl);
 }
 
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_f32mf2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vv_f32mf2_rm_tu(vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_f32mf2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vf_f32mf2_rm_tu(vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vv_f32m1_rm_tu(vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vf_f32m1_rm_tu(vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_f32m2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vv_f32m2_rm_tu(vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_f32m2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vf_f32m2_rm_tu(vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_f32m4_rm_tu
+// CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vv_f32m4_rm_tu(vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_f32m4_rm_tu
+// CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vf_f32m4_rm_tu(vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_f32m8_rm_tu
+// CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vv_f32m8_rm_tu(vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_f32m8_rm_tu
+// CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vf_f32m8_rm_tu(vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vv_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmsac_vv_f64m1_rm_tu(vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vf_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmsac_vf_f64m1_rm_tu(vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vv_f64m2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmsac_vv_f64m2_rm_tu(vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vf_f64m2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmsac_vf_f64m2_rm_tu(vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vv_f64m4_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmsac_vv_f64m4_rm_tu(vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vf_f64m4_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmsac_vf_f64m4_rm_tu(vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vv_f64m8_rm_tu
+// CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmsac_vv_f64m8_rm_tu(vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vf_f64m8_rm_tu
+// CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmsac_vf_f64m8_rm_tu(vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_f32mf2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vv_f32mf2_rm_tum(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_f32mf2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vf_f32mf2_rm_tum(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vv_f32m1_rm_tum(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vf_f32m1_rm_tum(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_f32m2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vv_f32m2_rm_tum(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_f32m2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vf_f32m2_rm_tum(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_f32m4_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vv_f32m4_rm_tum(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_f32m4_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vf_f32m4_rm_tum(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_f32m8_rm_tum
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vv_f32m8_rm_tum(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_f32m8_rm_tum
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vf_f32m8_rm_tum(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vv_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmsac_vv_f64m1_rm_tum(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vf_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmsac_vf_f64m1_rm_tum(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vv_f64m2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmsac_vv_f64m2_rm_tum(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vf_f64m2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmsac_vf_f64m2_rm_tum(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vv_f64m4_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmsac_vv_f64m4_rm_tum(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vf_f64m4_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmsac_vf_f64m4_rm_tum(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vv_f64m8_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmsac_vv_f64m8_rm_tum(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vf_f64m8_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmsac_vf_f64m8_rm_tum(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_f32mf2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vv_f32mf2_rm_tumu(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_f32mf2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vf_f32mf2_rm_tumu(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_f32m1_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vv_f32m1_rm_tumu(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_f32m1_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vf_f32m1_rm_tumu(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_f32m2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vv_f32m2_rm_tumu(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_f32m2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vf_f32m2_rm_tumu(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_f32m4_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vv_f32m4_rm_tumu(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_f32m4_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vf_f32m4_rm_tumu(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_f32m8_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vv_f32m8_rm_tumu(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_f32m8_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vf_f32m8_rm_tumu(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vv_f64m1_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmsac_vv_f64m1_rm_tumu(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vf_f64m1_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmsac_vf_f64m1_rm_tumu(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vv_f64m2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmsac_vv_f64m2_rm_tumu(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vf_f64m2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmsac_vf_f64m2_rm_tumu(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vv_f64m4_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmsac_vv_f64m4_rm_tumu(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vf_f64m4_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmsac_vf_f64m4_rm_tumu(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vv_f64m8_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmsac_vv_f64m8_rm_tumu(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vf_f64m8_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmsac_vf_f64m8_rm_tumu(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vv_f32mf2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vv_f32mf2_rm_mu(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwmsac_vf_f32mf2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwmsac_vf_f32mf2_rm_mu(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vv_f32m1_rm_mu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vv_f32m1_rm_mu(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwmsac_vf_f32m1_rm_mu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwmsac_vf_f32m1_rm_mu(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vv_f32m2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vv_f32m2_rm_mu(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwmsac_vf_f32m2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwmsac_vf_f32m2_rm_mu(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vv_f32m4_rm_mu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vv_f32m4_rm_mu(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwmsac_vf_f32m4_rm_mu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwmsac_vf_f32m4_rm_mu(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vv_f32m8_rm_mu
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vv_f32m8_rm_mu(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwmsac_vf_f32m8_rm_mu
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwmsac_vf_f32m8_rm_mu(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vv_f64m1_rm_mu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmsac_vv_f64m1_rm_mu(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwmsac_vf_f64m1_rm_mu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwmsac_vf_f64m1_rm_mu(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vv_f64m2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmsac_vv_f64m2_rm_mu(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwmsac_vf_f64m2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwmsac_vf_f64m2_rm_mu(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwmsac_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vv_f64m4_rm_mu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmsac_vv_f64m4_rm_mu(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwmsac_vf_f64m4_rm_mu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwmsac_vf_f64m4_rm_mu(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwmsac_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vv_f64m8_rm_mu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmsac_vv_f64m8_rm_mu(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwmsac_vf_f64m8_rm_mu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwmsac_vf_f64m8_rm_mu(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwmsac_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/vfwnmacc.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/vfwnmacc.c
--- a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/vfwnmacc.c
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/vfwnmacc.c
@@ -10,7 +10,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_f32mf2_tu
 // CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmacc_vv_f32mf2_tu(vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -20,7 +20,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_f32mf2_tu
 // CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmacc_vf_f32mf2_tu(vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -30,7 +30,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmacc_vv_f32m1_tu(vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -40,7 +40,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmacc_vf_f32m1_tu(vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -50,7 +50,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_f32m2_tu
 // CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmacc_vv_f32m2_tu(vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -60,7 +60,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_f32m2_tu
 // CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmacc_vf_f32m2_tu(vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -70,7 +70,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_f32m4_tu
 // CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmacc_vv_f32m4_tu(vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -80,7 +80,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_f32m4_tu
 // CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmacc_vf_f32m4_tu(vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -90,7 +90,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_f32m8_tu
 // CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmacc_vv_f32m8_tu(vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -100,7 +100,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_f32m8_tu
 // CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmacc_vf_f32m8_tu(vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -110,7 +110,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vv_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmacc_vv_f64m1_tu(vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -120,7 +120,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vf_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmacc_vf_f64m1_tu(vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -130,7 +130,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vv_f64m2_tu
 // CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmacc_vv_f64m2_tu(vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -140,7 +140,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vf_f64m2_tu
 // CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmacc_vf_f64m2_tu(vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -150,7 +150,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vv_f64m4_tu
 // CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmacc_vv_f64m4_tu(vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -160,7 +160,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vf_f64m4_tu
 // CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmacc_vf_f64m4_tu(vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -170,7 +170,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vv_f64m8_tu
 // CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmacc_vv_f64m8_tu(vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -180,7 +180,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vf_f64m8_tu
 // CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmacc_vf_f64m8_tu(vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
@@ -190,7 +190,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_f32mf2_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmacc_vv_f32mf2_tum(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -200,7 +200,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_f32mf2_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmacc_vf_f32mf2_tum(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -210,7 +210,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmacc_vv_f32m1_tum(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -220,7 +220,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmacc_vf_f32m1_tum(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -230,7 +230,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_f32m2_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmacc_vv_f32m2_tum(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -240,7 +240,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_f32m2_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmacc_vf_f32m2_tum(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -250,7 +250,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_f32m4_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmacc_vv_f32m4_tum(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -260,7 +260,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_f32m4_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmacc_vf_f32m4_tum(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -270,7 +270,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_f32m8_tum
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmacc_vv_f32m8_tum(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -280,7 +280,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_f32m8_tum
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmacc_vf_f32m8_tum(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -290,7 +290,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vv_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmacc_vv_f64m1_tum(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -300,7 +300,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vf_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmacc_vf_f64m1_tum(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -310,7 +310,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vv_f64m2_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmacc_vv_f64m2_tum(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -320,7 +320,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vf_f64m2_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmacc_vf_f64m2_tum(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -330,7 +330,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vv_f64m4_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmacc_vv_f64m4_tum(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -340,7 +340,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vf_f64m4_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmacc_vf_f64m4_tum(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -350,7 +350,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vv_f64m8_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmacc_vv_f64m8_tum(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -360,7 +360,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vf_f64m8_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmacc_vf_f64m8_tum(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
@@ -370,7 +370,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_f32mf2_tumu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmacc_vv_f32mf2_tumu(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -380,7 +380,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_f32mf2_tumu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmacc_vf_f32mf2_tumu(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -390,7 +390,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_f32m1_tumu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmacc_vv_f32m1_tumu(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -400,7 +400,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_f32m1_tumu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmacc_vf_f32m1_tumu(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -410,7 +410,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_f32m2_tumu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmacc_vv_f32m2_tumu(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -420,7 +420,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_f32m2_tumu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmacc_vf_f32m2_tumu(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -430,7 +430,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_f32m4_tumu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmacc_vv_f32m4_tumu(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -440,7 +440,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_f32m4_tumu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmacc_vf_f32m4_tumu(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -450,7 +450,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_f32m8_tumu
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmacc_vv_f32m8_tumu(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -460,7 +460,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_f32m8_tumu
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmacc_vf_f32m8_tumu(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -470,7 +470,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vv_f64m1_tumu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmacc_vv_f64m1_tumu(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -480,7 +480,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vf_f64m1_tumu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmacc_vf_f64m1_tumu(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -490,7 +490,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vv_f64m2_tumu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmacc_vv_f64m2_tumu(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -500,7 +500,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vf_f64m2_tumu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmacc_vf_f64m2_tumu(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -510,7 +510,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vv_f64m4_tumu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmacc_vv_f64m4_tumu(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -520,7 +520,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vf_f64m4_tumu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmacc_vf_f64m4_tumu(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -530,7 +530,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vv_f64m8_tumu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmacc_vv_f64m8_tumu(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -540,7 +540,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vf_f64m8_tumu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmacc_vf_f64m8_tumu(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
@@ -550,7 +550,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_f32mf2_mu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmacc_vv_f32mf2_mu(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -560,7 +560,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_f32mf2_mu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmacc_vf_f32mf2_mu(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -570,7 +570,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_f32m1_mu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmacc_vv_f32m1_mu(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -580,7 +580,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_f32m1_mu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmacc_vf_f32m1_mu(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -590,7 +590,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_f32m2_mu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmacc_vv_f32m2_mu(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -600,7 +600,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_f32m2_mu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmacc_vf_f32m2_mu(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -610,7 +610,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_f32m4_mu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmacc_vv_f32m4_mu(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -620,7 +620,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_f32m4_mu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmacc_vf_f32m4_mu(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -630,7 +630,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_f32m8_mu
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmacc_vv_f32m8_mu(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -640,7 +640,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_f32m8_mu
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmacc_vf_f32m8_mu(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -650,7 +650,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vv_f64m1_mu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmacc_vv_f64m1_mu(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -660,7 +660,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vf_f64m1_mu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmacc_vf_f64m1_mu(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -670,7 +670,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vv_f64m2_mu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmacc_vv_f64m2_mu(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -680,7 +680,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vf_f64m2_mu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmacc_vf_f64m2_mu(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -690,7 +690,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vv_f64m4_mu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmacc_vv_f64m4_mu(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -700,7 +700,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vf_f64m4_mu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmacc_vf_f64m4_mu(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -710,7 +710,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vv_f64m8_mu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmacc_vv_f64m8_mu(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -720,10 +720,730 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vf_f64m8_mu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmacc_vf_f64m8_mu(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
   return __riscv_vfwnmacc_mu(mask, vd, vs1, vs2, vl);
 }
 
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_f32mf2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vv_f32mf2_rm_tu(vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_f32mf2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vf_f32mf2_rm_tu(vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vv_f32m1_rm_tu(vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vf_f32m1_rm_tu(vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_f32m2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vv_f32m2_rm_tu(vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_f32m2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vf_f32m2_rm_tu(vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_f32m4_rm_tu
+// CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vv_f32m4_rm_tu(vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_f32m4_rm_tu
+// CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vf_f32m4_rm_tu(vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_f32m8_rm_tu
+// CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vv_f32m8_rm_tu(vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_f32m8_rm_tu
+// CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vf_f32m8_rm_tu(vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vv_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmacc_vv_f64m1_rm_tu(vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vf_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmacc_vf_f64m1_rm_tu(vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vv_f64m2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmacc_vv_f64m2_rm_tu(vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vf_f64m2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmacc_vf_f64m2_rm_tu(vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vv_f64m4_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmacc_vv_f64m4_rm_tu(vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vf_f64m4_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmacc_vf_f64m4_rm_tu(vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vv_f64m8_rm_tu
+// CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmacc_vv_f64m8_rm_tu(vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vf_f64m8_rm_tu
+// CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmacc_vf_f64m8_rm_tu(vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_f32mf2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vv_f32mf2_rm_tum(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_f32mf2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vf_f32mf2_rm_tum(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vv_f32m1_rm_tum(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vf_f32m1_rm_tum(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_f32m2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vv_f32m2_rm_tum(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_f32m2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vf_f32m2_rm_tum(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_f32m4_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vv_f32m4_rm_tum(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_f32m4_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vf_f32m4_rm_tum(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_f32m8_rm_tum
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vv_f32m8_rm_tum(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_f32m8_rm_tum
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vf_f32m8_rm_tum(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vv_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmacc_vv_f64m1_rm_tum(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vf_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmacc_vf_f64m1_rm_tum(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vv_f64m2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmacc_vv_f64m2_rm_tum(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vf_f64m2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmacc_vf_f64m2_rm_tum(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vv_f64m4_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmacc_vv_f64m4_rm_tum(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vf_f64m4_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmacc_vf_f64m4_rm_tum(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vv_f64m8_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmacc_vv_f64m8_rm_tum(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vf_f64m8_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmacc_vf_f64m8_rm_tum(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_f32mf2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vv_f32mf2_rm_tumu(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_f32mf2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vf_f32mf2_rm_tumu(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_f32m1_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vv_f32m1_rm_tumu(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_f32m1_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vf_f32m1_rm_tumu(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_f32m2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vv_f32m2_rm_tumu(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_f32m2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vf_f32m2_rm_tumu(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_f32m4_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vv_f32m4_rm_tumu(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_f32m4_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vf_f32m4_rm_tumu(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_f32m8_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vv_f32m8_rm_tumu(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_f32m8_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vf_f32m8_rm_tumu(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vv_f64m1_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmacc_vv_f64m1_rm_tumu(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vf_f64m1_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmacc_vf_f64m1_rm_tumu(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vv_f64m2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmacc_vv_f64m2_rm_tumu(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vf_f64m2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmacc_vf_f64m2_rm_tumu(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vv_f64m4_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmacc_vv_f64m4_rm_tumu(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vf_f64m4_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmacc_vf_f64m4_rm_tumu(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vv_f64m8_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmacc_vv_f64m8_rm_tumu(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vf_f64m8_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmacc_vf_f64m8_rm_tumu(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vv_f32mf2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vv_f32mf2_rm_mu(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmacc_vf_f32mf2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmacc_vf_f32mf2_rm_mu(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vv_f32m1_rm_mu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vv_f32m1_rm_mu(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmacc_vf_f32m1_rm_mu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmacc_vf_f32m1_rm_mu(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vv_f32m2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vv_f32m2_rm_mu(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmacc_vf_f32m2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmacc_vf_f32m2_rm_mu(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vv_f32m4_rm_mu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vv_f32m4_rm_mu(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmacc_vf_f32m4_rm_mu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmacc_vf_f32m4_rm_mu(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vv_f32m8_rm_mu
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vv_f32m8_rm_mu(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmacc_vf_f32m8_rm_mu
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmacc_vf_f32m8_rm_mu(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vv_f64m1_rm_mu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmacc_vv_f64m1_rm_mu(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmacc_vf_f64m1_rm_mu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmacc_vf_f64m1_rm_mu(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vv_f64m2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmacc_vv_f64m2_rm_mu(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmacc_vf_f64m2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmacc_vf_f64m2_rm_mu(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vv_f64m4_rm_mu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmacc_vv_f64m4_rm_mu(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmacc_vf_f64m4_rm_mu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmacc_vf_f64m4_rm_mu(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vv_f64m8_rm_mu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmacc_vv_f64m8_rm_mu(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmacc_vf_f64m8_rm_mu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmacc_vf_f64m8_rm_mu(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmacc_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/vfwnmsac.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/vfwnmsac.c
--- a/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/vfwnmsac.c
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/policy/overloaded/vfwnmsac.c
@@ -10,7 +10,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_f32mf2_tu
 // CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0:[0-9]+]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmsac_vv_f32mf2_tu(vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -20,7 +20,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_f32mf2_tu
 // CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmsac_vf_f32mf2_tu(vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -30,7 +30,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmsac_vv_f32m1_tu(vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -40,7 +40,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_f32m1_tu
 // CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmsac_vf_f32m1_tu(vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -50,7 +50,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_f32m2_tu
 // CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmsac_vv_f32m2_tu(vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -60,7 +60,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_f32m2_tu
 // CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmsac_vf_f32m2_tu(vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -70,7 +70,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_f32m4_tu
 // CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmsac_vv_f32m4_tu(vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -80,7 +80,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_f32m4_tu
 // CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmsac_vf_f32m4_tu(vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -90,7 +90,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_f32m8_tu
 // CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmsac_vv_f32m8_tu(vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -100,7 +100,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_f32m8_tu
 // CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmsac_vf_f32m8_tu(vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -110,7 +110,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vv_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmsac_vv_f64m1_tu(vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -120,7 +120,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vf_f64m1_tu
 // CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmsac_vf_f64m1_tu(vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -130,7 +130,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vv_f64m2_tu
 // CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmsac_vv_f64m2_tu(vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -140,7 +140,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vf_f64m2_tu
 // CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmsac_vf_f64m2_tu(vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -150,7 +150,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vv_f64m4_tu
 // CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmsac_vv_f64m4_tu(vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -160,7 +160,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vf_f64m4_tu
 // CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmsac_vf_f64m4_tu(vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -170,7 +170,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vv_f64m8_tu
 // CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmsac_vv_f64m8_tu(vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -180,7 +180,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vf_f64m8_tu
 // CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmsac_vf_f64m8_tu(vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
@@ -190,7 +190,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_f32mf2_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmsac_vv_f32mf2_tum(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -200,7 +200,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_f32mf2_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmsac_vf_f32mf2_tum(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -210,7 +210,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmsac_vv_f32m1_tum(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -220,7 +220,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_f32m1_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmsac_vf_f32m1_tum(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -230,7 +230,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_f32m2_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmsac_vv_f32m2_tum(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -240,7 +240,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_f32m2_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmsac_vf_f32m2_tum(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -250,7 +250,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_f32m4_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmsac_vv_f32m4_tum(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -260,7 +260,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_f32m4_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmsac_vf_f32m4_tum(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -270,7 +270,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_f32m8_tum
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmsac_vv_f32m8_tum(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -280,7 +280,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_f32m8_tum
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmsac_vf_f32m8_tum(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -290,7 +290,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vv_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmsac_vv_f64m1_tum(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -300,7 +300,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vf_f64m1_tum
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmsac_vf_f64m1_tum(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -310,7 +310,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vv_f64m2_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmsac_vv_f64m2_tum(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -320,7 +320,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vf_f64m2_tum
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmsac_vf_f64m2_tum(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -330,7 +330,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vv_f64m4_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmsac_vv_f64m4_tum(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -340,7 +340,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vf_f64m4_tum
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmsac_vf_f64m4_tum(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -350,7 +350,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vv_f64m8_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmsac_vv_f64m8_tum(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -360,7 +360,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vf_f64m8_tum
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 2)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmsac_vf_f64m8_tum(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
@@ -370,7 +370,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_f32mf2_tumu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmsac_vv_f32mf2_tumu(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -380,7 +380,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_f32mf2_tumu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmsac_vf_f32mf2_tumu(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -390,7 +390,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_f32m1_tumu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmsac_vv_f32m1_tumu(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -400,7 +400,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_f32m1_tumu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmsac_vf_f32m1_tumu(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -410,7 +410,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_f32m2_tumu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmsac_vv_f32m2_tumu(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -420,7 +420,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_f32m2_tumu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmsac_vf_f32m2_tumu(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -430,7 +430,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_f32m4_tumu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmsac_vv_f32m4_tumu(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -440,7 +440,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_f32m4_tumu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmsac_vf_f32m4_tumu(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -450,7 +450,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_f32m8_tumu
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmsac_vv_f32m8_tumu(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -460,7 +460,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_f32m8_tumu
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmsac_vf_f32m8_tumu(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -470,7 +470,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vv_f64m1_tumu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmsac_vv_f64m1_tumu(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -480,7 +480,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vf_f64m1_tumu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmsac_vf_f64m1_tumu(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -490,7 +490,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vv_f64m2_tumu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmsac_vv_f64m2_tumu(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -500,7 +500,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vf_f64m2_tumu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmsac_vf_f64m2_tumu(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -510,7 +510,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vv_f64m4_tumu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmsac_vv_f64m4_tumu(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -520,7 +520,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vf_f64m4_tumu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmsac_vf_f64m4_tumu(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -530,7 +530,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vv_f64m8_tumu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmsac_vv_f64m8_tumu(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -540,7 +540,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vf_f64m8_tumu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 0)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmsac_vf_f64m8_tumu(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
@@ -550,7 +550,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_f32mf2_mu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmsac_vv_f32mf2_mu(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -560,7 +560,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_f32mf2_mu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
 //
 vfloat32mf2_t test_vfwnmsac_vf_f32mf2_mu(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
@@ -570,7 +570,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_f32m1_mu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmsac_vv_f32m1_mu(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -580,7 +580,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_f32m1_mu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
 //
 vfloat32m1_t test_vfwnmsac_vf_f32m1_mu(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
@@ -590,7 +590,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_f32m2_mu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmsac_vv_f32m2_mu(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
@@ -600,7 +600,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_f32m2_mu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 vfloat32m2_t test_vfwnmsac_vf_f32m2_mu(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
@@ -610,7 +610,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_f32m4_mu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmsac_vv_f32m4_mu(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
@@ -620,7 +620,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_f32m4_mu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
 //
 vfloat32m4_t test_vfwnmsac_vf_f32m4_mu(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
@@ -630,7 +630,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_f32m8_mu
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmsac_vv_f32m8_mu(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
@@ -640,7 +640,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_f32m8_mu
 // CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
 //
 vfloat32m8_t test_vfwnmsac_vf_f32m8_mu(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
@@ -650,7 +650,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vv_f64m1_mu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmsac_vv_f64m1_mu(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -660,7 +660,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vf_f64m1_mu
 // CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
 //
 vfloat64m1_t test_vfwnmsac_vf_f64m1_mu(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
@@ -670,7 +670,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vv_f64m2_mu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmsac_vv_f64m2_mu(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
@@ -680,7 +680,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vf_f64m2_mu
 // CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 vfloat64m2_t test_vfwnmsac_vf_f64m2_mu(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
@@ -690,7 +690,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vv_f64m4_mu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmsac_vv_f64m4_mu(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
@@ -700,7 +700,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vf_f64m4_mu
 // CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
 //
 vfloat64m4_t test_vfwnmsac_vf_f64m4_mu(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
@@ -710,7 +710,7 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vv_f64m8_mu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmsac_vv_f64m8_mu(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
@@ -720,10 +720,730 @@
 // CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vf_f64m8_mu
 // CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
 // CHECK-RV64-NEXT:  entry:
-// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 7, i64 [[VL]], i64 1)
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfwnmsac_vf_f64m8_mu(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
   return __riscv_vfwnmsac_mu(mask, vd, vs1, vs2, vl);
 }
 
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_f32mf2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vv_f32mf2_rm_tu(vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_f32mf2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vf_f32mf2_rm_tu(vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vv_f32m1_rm_tu(vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_f32m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vf_f32m1_rm_tu(vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_f32m2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vv_f32m2_rm_tu(vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_f32m2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vf_f32m2_rm_tu(vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_f32m4_rm_tu
+// CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vv_f32m4_rm_tu(vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_f32m4_rm_tu
+// CHECK-RV64-SAME: (<vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vf_f32m4_rm_tu(vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_f32m8_rm_tu
+// CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vv_f32m8_rm_tu(vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_f32m8_rm_tu
+// CHECK-RV64-SAME: (<vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vf_f32m8_rm_tu(vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vv_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmsac_vv_f64m1_rm_tu(vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vf_f64m1_rm_tu
+// CHECK-RV64-SAME: (<vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmsac_vf_f64m1_rm_tu(vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vv_f64m2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmsac_vv_f64m2_rm_tu(vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vf_f64m2_rm_tu
+// CHECK-RV64-SAME: (<vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmsac_vf_f64m2_rm_tu(vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vv_f64m4_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmsac_vv_f64m4_rm_tu(vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vf_f64m4_rm_tu
+// CHECK-RV64-SAME: (<vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmsac_vf_f64m4_rm_tu(vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vv_f64m8_rm_tu
+// CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmsac_vv_f64m8_rm_tu(vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vf_f64m8_rm_tu
+// CHECK-RV64-SAME: (<vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmsac_vf_f64m8_rm_tu(vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tu(vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_f32mf2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vv_f32mf2_rm_tum(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_f32mf2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vf_f32mf2_rm_tum(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vv_f32m1_rm_tum(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_f32m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vf_f32m1_rm_tum(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_f32m2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vv_f32m2_rm_tum(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_f32m2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vf_f32m2_rm_tum(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_f32m4_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vv_f32m4_rm_tum(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_f32m4_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vf_f32m4_rm_tum(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_f32m8_rm_tum
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vv_f32m8_rm_tum(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_f32m8_rm_tum
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vf_f32m8_rm_tum(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vv_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmsac_vv_f64m1_rm_tum(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vf_f64m1_rm_tum
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmsac_vf_f64m1_rm_tum(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vv_f64m2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmsac_vv_f64m2_rm_tum(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vf_f64m2_rm_tum
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmsac_vf_f64m2_rm_tum(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vv_f64m4_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmsac_vv_f64m4_rm_tum(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vf_f64m4_rm_tum
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmsac_vf_f64m4_rm_tum(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vv_f64m8_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmsac_vv_f64m8_rm_tum(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vf_f64m8_rm_tum
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 2)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmsac_vf_f64m8_rm_tum(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tum(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_f32mf2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vv_f32mf2_rm_tumu(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_f32mf2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vf_f32mf2_rm_tumu(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_f32m1_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vv_f32m1_rm_tumu(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_f32m1_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vf_f32m1_rm_tumu(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_f32m2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vv_f32m2_rm_tumu(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_f32m2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vf_f32m2_rm_tumu(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_f32m4_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vv_f32m4_rm_tumu(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_f32m4_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vf_f32m4_rm_tumu(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_f32m8_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vv_f32m8_rm_tumu(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_f32m8_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vf_f32m8_rm_tumu(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vv_f64m1_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmsac_vv_f64m1_rm_tumu(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vf_f64m1_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmsac_vf_f64m1_rm_tumu(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vv_f64m2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmsac_vv_f64m2_rm_tumu(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vf_f64m2_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmsac_vf_f64m2_rm_tumu(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vv_f64m4_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmsac_vv_f64m4_rm_tumu(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vf_f64m4_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmsac_vf_f64m4_rm_tumu(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vv_f64m8_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmsac_vv_f64m8_rm_tumu(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vf_f64m8_rm_tumu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 0)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmsac_vf_f64m8_rm_tumu(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_tumu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vv_f32mf2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], <vscale x 1 x half> [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], <vscale x 1 x half> [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vv_f32mf2_rm_mu(vbool64_t mask, vfloat32mf2_t vd, vfloat16mf4_t vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x float> @test_vfwnmsac_vf_f32mf2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 1 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.f16.nxv1f16.i64(<vscale x 1 x float> [[VD]], half [[VS1]], <vscale x 1 x half> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfwnmsac_vf_f32mf2_rm_mu(vbool64_t mask, vfloat32mf2_t vd, _Float16 vs1, vfloat16mf4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vv_f32m1_rm_mu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], <vscale x 2 x half> [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], <vscale x 2 x half> [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vv_f32m1_rm_mu(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x float> @test_vfwnmsac_vf_f32m1_rm_mu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 2 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.f16.nxv2f16.i64(<vscale x 2 x float> [[VD]], half [[VS1]], <vscale x 2 x half> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x float> [[TMP0]]
+//
+vfloat32m1_t test_vfwnmsac_vf_f32m1_rm_mu(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vv_f32m2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], <vscale x 4 x half> [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], <vscale x 4 x half> [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vv_f32m2_rm_mu(vbool16_t mask, vfloat32m2_t vd, vfloat16m1_t vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x float> @test_vfwnmsac_vf_f32m2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 4 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.f16.nxv4f16.i64(<vscale x 4 x float> [[VD]], half [[VS1]], <vscale x 4 x half> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x float> [[TMP0]]
+//
+vfloat32m2_t test_vfwnmsac_vf_f32m2_rm_mu(vbool16_t mask, vfloat32m2_t vd, _Float16 vs1, vfloat16m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vv_f32m4_rm_mu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], <vscale x 8 x half> [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], <vscale x 8 x half> [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vv_f32m4_rm_mu(vbool8_t mask, vfloat32m4_t vd, vfloat16m2_t vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x float> @test_vfwnmsac_vf_f32m4_rm_mu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 8 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.f16.nxv8f16.i64(<vscale x 8 x float> [[VD]], half [[VS1]], <vscale x 8 x half> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x float> [[TMP0]]
+//
+vfloat32m4_t test_vfwnmsac_vf_f32m4_rm_mu(vbool8_t mask, vfloat32m4_t vd, _Float16 vs1, vfloat16m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vv_f32m8_rm_mu
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], <vscale x 16 x half> [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], <vscale x 16 x half> [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vv_f32m8_rm_mu(vbool4_t mask, vfloat32m8_t vd, vfloat16m4_t vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 16 x float> @test_vfwnmsac_vf_f32m8_rm_mu
+// CHECK-RV64-SAME: (<vscale x 16 x i1> [[MASK:%.*]], <vscale x 16 x float> [[VD:%.*]], half noundef [[VS1:%.*]], <vscale x 16 x half> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.f16.nxv16f16.i64(<vscale x 16 x float> [[VD]], half [[VS1]], <vscale x 16 x half> [[VS2]], <vscale x 16 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 16 x float> [[TMP0]]
+//
+vfloat32m8_t test_vfwnmsac_vf_f32m8_rm_mu(vbool4_t mask, vfloat32m8_t vd, _Float16 vs1, vfloat16m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vv_f64m1_rm_mu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], <vscale x 1 x float> [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.nxv1f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], <vscale x 1 x float> [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmsac_vv_f64m1_rm_mu(vbool64_t mask, vfloat64m1_t vd, vfloat32mf2_t vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 1 x double> @test_vfwnmsac_vf_f64m1_rm_mu
+// CHECK-RV64-SAME: (<vscale x 1 x i1> [[MASK:%.*]], <vscale x 1 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 1 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.f32.nxv1f32.i64(<vscale x 1 x double> [[VD]], float [[VS1]], <vscale x 1 x float> [[VS2]], <vscale x 1 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 1 x double> [[TMP0]]
+//
+vfloat64m1_t test_vfwnmsac_vf_f64m1_rm_mu(vbool64_t mask, vfloat64m1_t vd, float vs1, vfloat32mf2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vv_f64m2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], <vscale x 2 x float> [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.nxv2f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], <vscale x 2 x float> [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmsac_vv_f64m2_rm_mu(vbool32_t mask, vfloat64m2_t vd, vfloat32m1_t vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x double> @test_vfwnmsac_vf_f64m2_rm_mu
+// CHECK-RV64-SAME: (<vscale x 2 x i1> [[MASK:%.*]], <vscale x 2 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 2 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.f32.nxv2f32.i64(<vscale x 2 x double> [[VD]], float [[VS1]], <vscale x 2 x float> [[VS2]], <vscale x 2 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 2 x double> [[TMP0]]
+//
+vfloat64m2_t test_vfwnmsac_vf_f64m2_rm_mu(vbool32_t mask, vfloat64m2_t vd, float vs1, vfloat32m1_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vv_f64m4_rm_mu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], <vscale x 4 x float> [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.nxv4f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], <vscale x 4 x float> [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmsac_vv_f64m4_rm_mu(vbool16_t mask, vfloat64m4_t vd, vfloat32m2_t vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 4 x double> @test_vfwnmsac_vf_f64m4_rm_mu
+// CHECK-RV64-SAME: (<vscale x 4 x i1> [[MASK:%.*]], <vscale x 4 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 4 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.f32.nxv4f32.i64(<vscale x 4 x double> [[VD]], float [[VS1]], <vscale x 4 x float> [[VS2]], <vscale x 4 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 4 x double> [[TMP0]]
+//
+vfloat64m4_t test_vfwnmsac_vf_f64m4_rm_mu(vbool16_t mask, vfloat64m4_t vd, float vs1, vfloat32m2_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vv_f64m8_rm_mu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], <vscale x 8 x float> [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.nxv8f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], <vscale x 8 x float> [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmsac_vv_f64m8_rm_mu(vbool8_t mask, vfloat64m8_t vd, vfloat32m4_t vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 8 x double> @test_vfwnmsac_vf_f64m8_rm_mu
+// CHECK-RV64-SAME: (<vscale x 8 x i1> [[MASK:%.*]], <vscale x 8 x double> [[VD:%.*]], float noundef [[VS1:%.*]], <vscale x 8 x float> [[VS2:%.*]], i64 noundef [[VL:%.*]]) #[[ATTR0]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.f32.nxv8f32.i64(<vscale x 8 x double> [[VD]], float [[VS1]], <vscale x 8 x float> [[VS2]], <vscale x 8 x i1> [[MASK]], i64 0, i64 [[VL]], i64 1)
+// CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
+//
+vfloat64m8_t test_vfwnmsac_vf_f64m8_rm_mu(vbool8_t mask, vfloat64m8_t vd, float vs1, vfloat32m4_t vs2, size_t vl) {
+  return __riscv_vfwnmsac_mu(mask, vd, vs1, vs2, __RISCV_FRM_RNE, vl);
+}
+
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-handcrafted/vfwmacc-out-of-range.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-handcrafted/vfwmacc-out-of-range.c
new file mode 100644
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-handcrafted/vfwmacc-out-of-range.c
@@ -0,0 +1,66 @@
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +f -target-feature +d \
+// RUN:   -target-feature +v -target-feature +zfh -target-feature +experimental-zvfh \
+// RUN:   -fsyntax-only -verify %s
+
+#include <riscv_vector.h>
+
+vfloat32m1_t test_vfwmacc_vv_f32m1_rm(vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwmacc_vv_f32m1_rm(vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwmacc_vf_f32m1_rm(vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwmacc_vf_f32m1_rm(vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwmacc_vv_f32m1_rm_m(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwmacc_vv_f32m1_rm_m(mask, vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwmacc_vf_f32m1_rm_m(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwmacc_vf_f32m1_rm_m(mask, vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwmacc_vv_f32m1_rm_tu(vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwmacc_vv_f32m1_rm_tu(vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwmacc_vf_f32m1_rm_tu(vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwmacc_vf_f32m1_rm_tu(vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwmacc_vv_f32m1_rm_tum(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwmacc_vv_f32m1_rm_tum(mask, vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwmacc_vf_f32m1_rm_tum(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwmacc_vf_f32m1_rm_tum(mask, vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwmacc_vv_f32m1_rm_tumu(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwmacc_vv_f32m1_rm_tumu(mask, vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwmacc_vf_f32m1_rm_tumu(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwmacc_vf_f32m1_rm_tumu(mask, vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwmacc_vv_f32m1_rm_mu(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwmacc_vv_f32m1_rm_mu(mask, vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwmacc_vf_f32m1_rm_mu(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwmacc_vf_f32m1_rm_mu(mask, vd, vs1, vs2, 5, vl);
+}
\ No newline at end of file
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-handcrafted/vfwmsac-out-of-range.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-handcrafted/vfwmsac-out-of-range.c
new file mode 100644
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-handcrafted/vfwmsac-out-of-range.c
@@ -0,0 +1,66 @@
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +f -target-feature +d \
+// RUN:   -target-feature +v -target-feature +zfh -target-feature +experimental-zvfh \
+// RUN:   -fsyntax-only -verify %s
+
+#include <riscv_vector.h>
+
+vfloat32m1_t test_vfwmsac_vv_f32m1_rm(vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwmsac_vv_f32m1_rm(vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwmsac_vf_f32m1_rm(vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwmsac_vf_f32m1_rm(vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwmsac_vv_f32m1_rm_m(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwmsac_vv_f32m1_rm_m(mask, vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwmsac_vf_f32m1_rm_m(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwmsac_vf_f32m1_rm_m(mask, vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwmsac_vv_f32m1_rm_tu(vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwmsac_vv_f32m1_rm_tu(vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwmsac_vf_f32m1_rm_tu(vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwmsac_vf_f32m1_rm_tu(vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwmsac_vv_f32m1_rm_tum(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwmsac_vv_f32m1_rm_tum(mask, vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwmsac_vf_f32m1_rm_tum(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwmsac_vf_f32m1_rm_tum(mask, vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwmsac_vv_f32m1_rm_tumu(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwmsac_vv_f32m1_rm_tumu(mask, vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwmsac_vf_f32m1_rm_tumu(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwmsac_vf_f32m1_rm_tumu(mask, vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwmsac_vv_f32m1_rm_mu(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwmsac_vv_f32m1_rm_mu(mask, vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwmsac_vf_f32m1_rm_mu(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwmsac_vf_f32m1_rm_mu(mask, vd, vs1, vs2, 5, vl);
+}
\ No newline at end of file
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-handcrafted/vfwnmacc-out-of-range.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-handcrafted/vfwnmacc-out-of-range.c
new file mode 100644
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-handcrafted/vfwnmacc-out-of-range.c
@@ -0,0 +1,66 @@
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +f -target-feature +d \
+// RUN:   -target-feature +v -target-feature +zfh -target-feature +experimental-zvfh \
+// RUN:   -fsyntax-only -verify %s
+
+#include <riscv_vector.h>
+
+vfloat32m1_t test_vfwnmacc_vv_f32m1_rm(vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwnmacc_vv_f32m1_rm(vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwnmacc_vf_f32m1_rm(vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwnmacc_vf_f32m1_rm(vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwnmacc_vv_f32m1_rm_m(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwnmacc_vv_f32m1_rm_m(mask, vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwnmacc_vf_f32m1_rm_m(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwnmacc_vf_f32m1_rm_m(mask, vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwnmacc_vv_f32m1_rm_tu(vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwnmacc_vv_f32m1_rm_tu(vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwnmacc_vf_f32m1_rm_tu(vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwnmacc_vf_f32m1_rm_tu(vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwnmacc_vv_f32m1_rm_tum(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwnmacc_vv_f32m1_rm_tum(mask, vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwnmacc_vf_f32m1_rm_tum(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwnmacc_vf_f32m1_rm_tum(mask, vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwnmacc_vv_f32m1_rm_tumu(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwnmacc_vv_f32m1_rm_tumu(mask, vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwnmacc_vf_f32m1_rm_tumu(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwnmacc_vf_f32m1_rm_tumu(mask, vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwnmacc_vv_f32m1_rm_mu(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwnmacc_vv_f32m1_rm_mu(mask, vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwnmacc_vf_f32m1_rm_mu(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwnmacc_vf_f32m1_rm_mu(mask, vd, vs1, vs2, 5, vl);
+}
\ No newline at end of file
diff --git a/clang/test/CodeGen/RISCV/rvv-intrinsics-handcrafted/vfwnmsac-out-of-range.c b/clang/test/CodeGen/RISCV/rvv-intrinsics-handcrafted/vfwnmsac-out-of-range.c
new file mode 100644
--- /dev/null
+++ b/clang/test/CodeGen/RISCV/rvv-intrinsics-handcrafted/vfwnmsac-out-of-range.c
@@ -0,0 +1,66 @@
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +f -target-feature +d \
+// RUN:   -target-feature +v -target-feature +zfh -target-feature +experimental-zvfh \
+// RUN:   -fsyntax-only -verify %s
+
+#include <riscv_vector.h>
+
+vfloat32m1_t test_vfwnmsac_vv_f32m1_rm(vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwnmsac_vv_f32m1_rm(vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwnmsac_vf_f32m1_rm(vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwnmsac_vf_f32m1_rm(vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwnmsac_vv_f32m1_rm_m(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwnmsac_vv_f32m1_rm_m(mask, vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwnmsac_vf_f32m1_rm_m(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwnmsac_vf_f32m1_rm_m(mask, vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwnmsac_vv_f32m1_rm_tu(vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwnmsac_vv_f32m1_rm_tu(vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwnmsac_vf_f32m1_rm_tu(vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwnmsac_vf_f32m1_rm_tu(vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwnmsac_vv_f32m1_rm_tum(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwnmsac_vv_f32m1_rm_tum(mask, vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwnmsac_vf_f32m1_rm_tum(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwnmsac_vf_f32m1_rm_tum(mask, vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwnmsac_vv_f32m1_rm_tumu(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwnmsac_vv_f32m1_rm_tumu(mask, vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwnmsac_vf_f32m1_rm_tumu(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwnmsac_vf_f32m1_rm_tumu(mask, vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwnmsac_vv_f32m1_rm_mu(vbool32_t mask, vfloat32m1_t vd, vfloat16mf2_t vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwnmsac_vv_f32m1_rm_mu(mask, vd, vs1, vs2, 5, vl);
+}
+
+vfloat32m1_t test_vfwnmsac_vf_f32m1_rm_mu(vbool32_t mask, vfloat32m1_t vd, _Float16 vs1, vfloat16mf2_t vs2, size_t vl) {
+  // expected-error@+1 {{argument value 5 is outside the valid range [0, 4]}}
+  return __riscv_vfwnmsac_vf_f32m1_rm_mu(mask, vd, vs1, vs2, 5, vl);
+}
\ No newline at end of file
diff --git a/llvm/include/llvm/IR/IntrinsicsRISCV.td b/llvm/include/llvm/IR/IntrinsicsRISCV.td
--- a/llvm/include/llvm/IR/IntrinsicsRISCV.td
+++ b/llvm/include/llvm/IR/IntrinsicsRISCV.td
@@ -825,6 +825,29 @@
     let ScalarOperand = 1;
     let VLOperand = 4;
   }
+  // UnMasked Widening Vector Multiply-Add operations, its first operand can not be undef.
+  // Input: (vector_in, vector_in/scalar, vector_in, frm, vl, policy)
+  class RISCVTernaryWideUnMaskedRoundingMode
+        : DefaultAttrsIntrinsic< [llvm_anyvector_ty],
+                     [LLVMMatchType<0>, llvm_any_ty, llvm_anyvector_ty,
+                      llvm_anyint_ty, LLVMMatchType<3>, LLVMMatchType<3>],
+                     [ImmArg<ArgIndex<3>>, ImmArg<ArgIndex<5>>, IntrNoMem] >,
+                     RISCVVIntrinsic {
+    let ScalarOperand = 1;
+    let VLOperand = 4;
+  }
+  // Masked Widening Vector Multiply-Add operations, its first operand can not be undef.
+  // Input: (vector_in, vector_in/scalar, vector_in, mask, frm, vl, policy
+  class RISCVTernaryWideMaskedRoundingMode
+        : DefaultAttrsIntrinsic< [llvm_anyvector_ty],
+                     [LLVMMatchType<0>, llvm_any_ty, llvm_anyvector_ty,
+                      LLVMScalarOrSameVectorWidth<0, llvm_i1_ty>,
+                      llvm_anyint_ty, LLVMMatchType<3>, LLVMMatchType<3>],
+                     [ImmArg<ArgIndex<4>>, ImmArg<ArgIndex<6>>, IntrNoMem]>,
+                     RISCVVIntrinsic {
+    let ScalarOperand = 1;
+    let VLOperand = 5;
+  }
   // For Reduction ternary operations.
   // For destination vector type is the same as first and third source vector.
   // Input: (vector_in, vector_in, vector_in, vl)
@@ -1240,6 +1263,10 @@
     def "int_riscv_" # NAME : RISCVTernaryWideUnMasked;
     def "int_riscv_" # NAME # "_mask" : RISCVTernaryWideMasked;
   }
+  multiclass RISCVTernaryWideRoundingMode {
+    def "int_riscv_" # NAME : RISCVTernaryWideUnMaskedRoundingMode;
+    def "int_riscv_" # NAME # "_mask" : RISCVTernaryWideMaskedRoundingMode;
+  }
   multiclass RISCVReduction {
     def "int_riscv_" # NAME : RISCVReductionUnMasked;
     def "int_riscv_" # NAME # "_mask" : RISCVReductionMasked;
@@ -1454,10 +1481,10 @@
   defm vfmsub : RISCVTernaryAAXARoundingMode;
   defm vfnmsub : RISCVTernaryAAXARoundingMode;
 
-  defm vfwmacc : RISCVTernaryWide;
-  defm vfwnmacc : RISCVTernaryWide;
-  defm vfwmsac : RISCVTernaryWide;
-  defm vfwnmsac : RISCVTernaryWide;
+  defm vfwmacc : RISCVTernaryWideRoundingMode;
+  defm vfwnmacc : RISCVTernaryWideRoundingMode;
+  defm vfwmsac : RISCVTernaryWideRoundingMode;
+  defm vfwnmsac : RISCVTernaryWideRoundingMode;
 
   defm vfsqrt : RISCVUnaryAA;
   defm vfrsqrt7 : RISCVUnaryAA;
diff --git a/llvm/lib/Target/RISCV/RISCVInstrInfoVPseudos.td b/llvm/lib/Target/RISCV/RISCVInstrInfoVPseudos.td
--- a/llvm/lib/Target/RISCV/RISCVInstrInfoVPseudos.td
+++ b/llvm/lib/Target/RISCV/RISCVInstrInfoVPseudos.td
@@ -3275,6 +3275,12 @@
                                       constraint>;
 }
 
+multiclass VPseudoTernaryW_VV_RM<LMULInfo m> {
+  defvar constraint = "@earlyclobber $rd";
+  defm _VV : VPseudoTernaryWithPolicyRoundingMode<m.wvrclass, m.vrclass, m.vrclass, m,
+                                                  constraint>;
+}
+
 multiclass VPseudoTernaryW_VX<LMULInfo m> {
   defvar constraint = "@earlyclobber $rd";
   defm "_VX" : VPseudoTernaryWithPolicy<m.wvrclass, GPR, m.vrclass, m,
@@ -3287,6 +3293,12 @@
                                               m.vrclass, m, constraint>;
 }
 
+multiclass VPseudoTernaryW_VF_RM<LMULInfo m, FPR_Info f> {
+  defvar constraint = "@earlyclobber $rd";
+  defm "_V" # f.FX : VPseudoTernaryWithPolicyRoundingMode<m.wvrclass, f.fprclass,
+                                                          m.vrclass, m, constraint>;
+}
+
 multiclass VPseudoVSLDVWithPolicy<VReg RetClass,
                                   RegisterClass Op1Class,
                                   DAGOperand Op2Class,
@@ -3416,13 +3428,13 @@
   }
 }
 
-multiclass VPseudoVWMAC_VV_VF {
+multiclass VPseudoVWMAC_VV_VF_RM {
   foreach m = MxListFW in {
     defvar mx = m.MX;
     defvar WriteVFWMulAddV_MX = !cast<SchedWrite>("WriteVFWMulAddV_" # mx);
     defvar ReadVFWMulAddV_MX = !cast<SchedRead>("ReadVFWMulAddV_" # mx);
 
-    defm "" : VPseudoTernaryW_VV<m>,
+    defm "" : VPseudoTernaryW_VV_RM<m>,
               Sched<[WriteVFWMulAddV_MX, ReadVFWMulAddV_MX,
                      ReadVFWMulAddV_MX, ReadVFWMulAddV_MX, ReadVMask]>;
   }
@@ -3434,7 +3446,7 @@
       defvar ReadVFWMulAddV_MX = !cast<SchedRead>("ReadVFWMulAddV_" # mx);
       defvar ReadVFWMulAddF_MX = !cast<SchedRead>("ReadVFWMulAddF_" # mx);
 
-      defm "" : VPseudoTernaryW_VF<m, f>,
+      defm "" : VPseudoTernaryW_VF_RM<m, f>,
                 Sched<[WriteVFWMulAddF_MX, ReadVFWMulAddV_MX,
                        ReadVFWMulAddV_MX, ReadVFWMulAddF_MX, ReadVMask]>;
     }
@@ -5572,6 +5584,20 @@
   }
 }
 
+multiclass VPatTernaryW_VV_RM<string intrinsic, string instruction,
+                           list<VTypeInfoToWide> vtilist> {
+  foreach vtiToWti = vtilist in {
+    defvar vti = vtiToWti.Vti;
+    defvar wti = vtiToWti.Wti;
+    let Predicates = !listconcat(GetVTypePredicates<vti>.Predicates,
+                                 GetVTypePredicates<wti>.Predicates) in
+    defm : VPatTernaryWithPolicyRoundingMode<intrinsic, instruction, "VV",
+                                             wti.Vector, vti.Vector, vti.Vector,
+                                             vti.Mask, vti.Log2SEW, vti.LMul,
+                                             wti.RegClass, vti.RegClass, vti.RegClass>;
+  }
+}
+
 multiclass VPatTernaryW_VX<string intrinsic, string instruction,
                            list<VTypeInfoToWide> vtilist> {
   foreach vtiToWti = vtilist in {
@@ -5587,6 +5613,22 @@
   }
 }
 
+multiclass VPatTernaryW_VX_RM<string intrinsic, string instruction,
+                           list<VTypeInfoToWide> vtilist> {
+  foreach vtiToWti = vtilist in {
+    defvar vti = vtiToWti.Vti;
+    defvar wti = vtiToWti.Wti;
+    let Predicates = !listconcat(GetVTypePredicates<vti>.Predicates,
+                                 GetVTypePredicates<wti>.Predicates) in
+    defm : VPatTernaryWithPolicyRoundingMode<intrinsic, instruction,
+                                             "V"#vti.ScalarSuffix,
+                                             wti.Vector, vti.Scalar, vti.Vector,
+                                             vti.Mask, vti.Log2SEW, vti.LMul,
+                                             wti.RegClass, vti.ScalarRegClass,
+                                             vti.RegClass>;
+  }
+}
+
 multiclass VPatTernaryV_VV_VX_AAXA<string intrinsic, string instruction,
                               list<VTypeInfo> vtilist>
     : VPatTernaryV_VV_AAXA<intrinsic, instruction, vtilist>,
@@ -5614,6 +5656,11 @@
     : VPatTernaryW_VV<intrinsic, instruction, vtilist>,
       VPatTernaryW_VX<intrinsic, instruction, vtilist>;
 
+multiclass VPatTernaryW_VV_VX_RM<string intrinsic, string instruction,
+                              list<VTypeInfoToWide> vtilist>
+    : VPatTernaryW_VV_RM<intrinsic, instruction, vtilist>,
+      VPatTernaryW_VX_RM<intrinsic, instruction, vtilist>;
+
 multiclass VPatBinaryM_VV_VX<string intrinsic, string instruction,
                              list<VTypeInfo> vtilist>
     : VPatBinaryM_VV<intrinsic, instruction, vtilist>,
@@ -6206,11 +6253,12 @@
 //===----------------------------------------------------------------------===//
 // 13.7. Vector Widening Floating-Point Fused Multiply-Add Instructions
 //===----------------------------------------------------------------------===//
-let Uses = [FRM], mayRaiseFPException = true in {
-defm PseudoVFWMACC  : VPseudoVWMAC_VV_VF;
-defm PseudoVFWNMACC : VPseudoVWMAC_VV_VF;
-defm PseudoVFWMSAC  : VPseudoVWMAC_VV_VF;
-defm PseudoVFWNMSAC : VPseudoVWMAC_VV_VF;
+let mayRaiseFPException = true, hasSideEffects = 0,
+    hasPostISelHook = 1 in {
+defm PseudoVFWMACC  : VPseudoVWMAC_VV_VF_RM;
+defm PseudoVFWNMACC : VPseudoVWMAC_VV_VF_RM;
+defm PseudoVFWMSAC  : VPseudoVWMAC_VV_VF_RM;
+defm PseudoVFWNMSAC : VPseudoVWMAC_VV_VF_RM;
 }
 
 //===----------------------------------------------------------------------===//
@@ -6866,10 +6914,14 @@
 //===----------------------------------------------------------------------===//
 // 13.7. Vector Widening Floating-Point Fused Multiply-Add Instructions
 //===----------------------------------------------------------------------===//
-defm : VPatTernaryW_VV_VX<"int_riscv_vfwmacc", "PseudoVFWMACC", AllWidenableFloatVectors>;
-defm : VPatTernaryW_VV_VX<"int_riscv_vfwnmacc", "PseudoVFWNMACC", AllWidenableFloatVectors>;
-defm : VPatTernaryW_VV_VX<"int_riscv_vfwmsac", "PseudoVFWMSAC", AllWidenableFloatVectors>;
-defm : VPatTernaryW_VV_VX<"int_riscv_vfwnmsac", "PseudoVFWNMSAC", AllWidenableFloatVectors>;
+defm : VPatTernaryW_VV_VX_RM<"int_riscv_vfwmacc", "PseudoVFWMACC",
+                             AllWidenableFloatVectors>;
+defm : VPatTernaryW_VV_VX_RM<"int_riscv_vfwnmacc", "PseudoVFWNMACC",
+                             AllWidenableFloatVectors>;
+defm : VPatTernaryW_VV_VX_RM<"int_riscv_vfwmsac", "PseudoVFWMSAC",
+                             AllWidenableFloatVectors>;
+defm : VPatTernaryW_VV_VX_RM<"int_riscv_vfwnmsac", "PseudoVFWNMSAC",
+                             AllWidenableFloatVectors>;
 
 //===----------------------------------------------------------------------===//
 // 13.8. Vector Floating-Point Square-Root Instruction
diff --git a/llvm/lib/Target/RISCV/RISCVInstrInfoVSDPatterns.td b/llvm/lib/Target/RISCV/RISCVInstrInfoVSDPatterns.td
--- a/llvm/lib/Target/RISCV/RISCVInstrInfoVSDPatterns.td
+++ b/llvm/lib/Target/RISCV/RISCVInstrInfoVSDPatterns.td
@@ -693,7 +693,7 @@
     : VPatWidenBinaryFPSDNode_VV_VF_RM<op, instruction_name>,
       VPatWidenBinaryFPSDNode_WV_WF_RM<op, instruction_name>;
 
-multiclass VPatWidenFPMulAccSDNode_VV_VF<string instruction_name> {
+multiclass VPatWidenFPMulAccSDNode_VV_VF_RM<string instruction_name> {
   foreach vtiToWti = AllWidenableFloatVectors in {
     defvar vti = vtiToWti.Vti;
     defvar wti = vtiToWti.Wti;
@@ -708,6 +708,9 @@
                      (wti.Vector wti.RegClass:$rd)),
                 (!cast<Instruction>(instruction_name#"_VV_"#vti.LMul.MX)
                    wti.RegClass:$rd, vti.RegClass:$rs1, vti.RegClass:$rs2,
+                   // Value to indicate no rounding mode change in
+                   // RISCVInsertReadWriteCSR
+                   FRM_DYN,
                    vti.AVL, vti.Log2SEW, TAIL_AGNOSTIC)>;
       def : Pat<(fma (wti.Vector (SplatFPOp
                                       (fpext_oneuse vti.ScalarRegClass:$rs1))),
@@ -717,12 +720,15 @@
                      (wti.Vector wti.RegClass:$rd)),
                 (!cast<Instruction>(instruction_name#"_V"#vti.ScalarSuffix#"_"#vti.LMul.MX)
                    wti.RegClass:$rd, vti.ScalarRegClass:$rs1, vti.RegClass:$rs2,
+                   // Value to indicate no rounding mode change in
+                   // RISCVInsertReadWriteCSR
+                   FRM_DYN,
                    vti.AVL, vti.Log2SEW, TAIL_AGNOSTIC)>;
     }
   }
 }
 
-multiclass VPatWidenFPNegMulAccSDNode_VV_VF<string instruction_name> {
+multiclass VPatWidenFPNegMulAccSDNode_VV_VF_RM<string instruction_name> {
   foreach vtiToWti = AllWidenableFloatVectors in {
     defvar vti = vtiToWti.Vti;
     defvar wti = vtiToWti.Wti;
@@ -736,6 +742,9 @@
                      (fneg wti.RegClass:$rd)),
                 (!cast<Instruction>(instruction_name#"_VV_"#vti.LMul.MX)
                    wti.RegClass:$rd, vti.RegClass:$rs1, vti.RegClass:$rs2,
+                   // Value to indicate no rounding mode change in
+                   // RISCVInsertReadWriteCSR
+                   FRM_DYN,
                    vti.AVL, vti.Log2SEW, TAIL_AGNOSTIC)>;
       def : Pat<(fma (SplatFPOp (fpext_oneuse vti.ScalarRegClass:$rs1)),
                      (fneg (wti.Vector (riscv_fpextend_vl_oneuse
@@ -744,6 +753,9 @@
                      (fneg wti.RegClass:$rd)),
                 (!cast<Instruction>(instruction_name#"_V"#vti.ScalarSuffix#"_"#vti.LMul.MX)
                    wti.RegClass:$rd, vti.ScalarRegClass:$rs1, vti.RegClass:$rs2,
+                   // Value to indicate no rounding mode change in
+                   // RISCVInsertReadWriteCSR
+                   FRM_DYN,
                    vti.AVL, vti.Log2SEW, TAIL_AGNOSTIC)>;
       def : Pat<(fma (fneg (wti.Vector (SplatFPOp (fpext_oneuse vti.ScalarRegClass:$rs1)))),
                      (riscv_fpextend_vl_oneuse (vti.Vector vti.RegClass:$rs2),
@@ -751,12 +763,15 @@
                      (fneg wti.RegClass:$rd)),
                 (!cast<Instruction>(instruction_name#"_V"#vti.ScalarSuffix#"_"#vti.LMul.MX)
                    wti.RegClass:$rd, vti.ScalarRegClass:$rs1, vti.RegClass:$rs2,
+                   // Value to indicate no rounding mode change in
+                   // RISCVInsertReadWriteCSR
+                   FRM_DYN,
                    vti.AVL, vti.Log2SEW, TAIL_AGNOSTIC)>;
     }
   }
 }
 
-multiclass VPatWidenFPMulSacSDNode_VV_VF<string instruction_name> {
+multiclass VPatWidenFPMulSacSDNode_VV_VF_RM<string instruction_name> {
   foreach vtiToWti = AllWidenableFloatVectors in {
     defvar vti = vtiToWti.Vti;
     defvar wti = vtiToWti.Wti;
@@ -770,6 +785,9 @@
                      (fneg wti.RegClass:$rd)),
                 (!cast<Instruction>(instruction_name#"_VV_"#vti.LMul.MX)
                    wti.RegClass:$rd, vti.RegClass:$rs1, vti.RegClass:$rs2,
+                   // Value to indicate no rounding mode change in
+                   // RISCVInsertReadWriteCSR
+                   FRM_DYN,
                    vti.AVL, vti.Log2SEW, TAIL_AGNOSTIC)>;
       def : Pat<(fma (wti.Vector (SplatFPOp (fpext_oneuse vti.ScalarRegClass:$rs1))),
                      (riscv_fpextend_vl_oneuse (vti.Vector vti.RegClass:$rs2),
@@ -777,12 +795,15 @@
                      (fneg wti.RegClass:$rd)),
                 (!cast<Instruction>(instruction_name#"_V"#vti.ScalarSuffix#"_"#vti.LMul.MX)
                    wti.RegClass:$rd, vti.ScalarRegClass:$rs1, vti.RegClass:$rs2,
+                   // Value to indicate no rounding mode change in
+                   // RISCVInsertReadWriteCSR
+                   FRM_DYN,
                    vti.AVL, vti.Log2SEW, TAIL_AGNOSTIC)>;
     }
   }
 }
 
-multiclass VPatWidenFPNegMulSacSDNode_VV_VF<string instruction_name> {
+multiclass VPatWidenFPNegMulSacSDNode_VV_VF_RM<string instruction_name> {
   foreach vtiToWti = AllWidenableFloatVectors in {
     defvar vti = vtiToWti.Vti;
     defvar wti = vtiToWti.Wti;
@@ -796,6 +817,9 @@
                      wti.RegClass:$rd),
                 (!cast<Instruction>(instruction_name#"_VV_"#vti.LMul.MX)
                    wti.RegClass:$rd, vti.RegClass:$rs1, vti.RegClass:$rs2,
+                   // Value to indicate no rounding mode change in
+                   // RISCVInsertReadWriteCSR
+                   FRM_DYN,
                    vti.AVL, vti.Log2SEW, TAIL_AGNOSTIC)>;
       def : Pat<(fma (wti.Vector (SplatFPOp (fpext_oneuse vti.ScalarRegClass:$rs1))),
                      (fneg (wti.Vector (riscv_fpextend_vl_oneuse
@@ -804,6 +828,9 @@
                      wti.RegClass:$rd),
                 (!cast<Instruction>(instruction_name#"_V"#vti.ScalarSuffix#"_"#vti.LMul.MX)
                    wti.RegClass:$rd, vti.ScalarRegClass:$rs1, vti.RegClass:$rs2,
+                   // Value to indicate no rounding mode change in
+                   // RISCVInsertReadWriteCSR
+                   FRM_DYN,
                    vti.AVL, vti.Log2SEW, TAIL_AGNOSTIC)>;
       def : Pat<(fma (fneg (wti.Vector (SplatFPOp (fpext_oneuse vti.ScalarRegClass:$rs1)))),
                      (riscv_fpextend_vl_oneuse (vti.Vector vti.RegClass:$rs2),
@@ -811,6 +838,9 @@
                      wti.RegClass:$rd),
                 (!cast<Instruction>(instruction_name#"_V"#vti.ScalarSuffix#"_"#vti.LMul.MX)
                    wti.RegClass:$rd, vti.ScalarRegClass:$rs1, vti.RegClass:$rs2,
+                   // Value to indicate no rounding mode change in
+                   // RISCVInsertReadWriteCSR
+                   FRM_DYN,
                    vti.AVL, vti.Log2SEW, TAIL_AGNOSTIC)>;
     }
   }
@@ -1225,10 +1255,10 @@
 }
 
 // 13.7. Vector Widening Floating-Point Fused Multiply-Add Instructions
-defm : VPatWidenFPMulAccSDNode_VV_VF<"PseudoVFWMACC">;
-defm : VPatWidenFPNegMulAccSDNode_VV_VF<"PseudoVFWNMACC">;
-defm : VPatWidenFPMulSacSDNode_VV_VF<"PseudoVFWMSAC">;
-defm : VPatWidenFPNegMulSacSDNode_VV_VF<"PseudoVFWNMSAC">;
+defm : VPatWidenFPMulAccSDNode_VV_VF_RM<"PseudoVFWMACC">;
+defm : VPatWidenFPNegMulAccSDNode_VV_VF_RM<"PseudoVFWNMACC">;
+defm : VPatWidenFPMulSacSDNode_VV_VF_RM<"PseudoVFWMSAC">;
+defm : VPatWidenFPNegMulSacSDNode_VV_VF_RM<"PseudoVFWNMSAC">;
 
 foreach vti = AllFloatVectors in {
   let Predicates = GetVTypePredicates<vti>.Predicates in {
diff --git a/llvm/lib/Target/RISCV/RISCVInstrInfoVVLPatterns.td b/llvm/lib/Target/RISCV/RISCVInstrInfoVVLPatterns.td
--- a/llvm/lib/Target/RISCV/RISCVInstrInfoVVLPatterns.td
+++ b/llvm/lib/Target/RISCV/RISCVInstrInfoVVLPatterns.td
@@ -1801,6 +1801,38 @@
   }
 }
 
+multiclass VPatWidenFPMulAccVL_VV_VF_RM<SDNode vop, string instruction_name> {
+  foreach vtiToWti = AllWidenableFloatVectors in {
+    defvar vti = vtiToWti.Vti;
+    defvar wti = vtiToWti.Wti;
+    let Predicates = !listconcat(GetVTypePredicates<vti>.Predicates,
+                                 GetVTypePredicates<wti>.Predicates) in {
+      def : Pat<(vop (vti.Vector vti.RegClass:$rs1),
+                     (vti.Vector vti.RegClass:$rs2),
+                     (wti.Vector wti.RegClass:$rd), (vti.Mask V0),
+                     VLOpFrag),
+                (!cast<Instruction>(instruction_name#"_VV_"#vti.LMul.MX #"_MASK")
+                   wti.RegClass:$rd, vti.RegClass:$rs1, vti.RegClass:$rs2,
+                   (vti.Mask V0),
+                   // Value to indicate no rounding mode change in
+                   // RISCVInsertReadWriteCSR
+                   FRM_DYN,
+                   GPR:$vl, vti.Log2SEW, TA_MA)>;
+      def : Pat<(vop (vti.Vector (SplatFPOp vti.ScalarRegClass:$rs1)),
+                     (vti.Vector vti.RegClass:$rs2),
+                     (wti.Vector wti.RegClass:$rd), (vti.Mask V0),
+                     VLOpFrag),
+                (!cast<Instruction>(instruction_name#"_V"#vti.ScalarSuffix#"_"#vti.LMul.MX #"_MASK")
+                   wti.RegClass:$rd, vti.ScalarRegClass:$rs1, vti.RegClass:$rs2,
+                   (vti.Mask V0),
+                   // Value to indicate no rounding mode change in
+                   // RISCVInsertReadWriteCSR
+                   FRM_DYN,
+                   GPR:$vl, vti.Log2SEW, TA_MA)>;
+    }
+  }
+}
+
 //===----------------------------------------------------------------------===//
 // Patterns.
 //===----------------------------------------------------------------------===//
@@ -2126,10 +2158,10 @@
 defm : VPatFPMulAccVL_VV_VF_RM<riscv_vfnmsub_vl_oneuse, "PseudoVFNMSAC">;
 
 // 13.7. Vector Widening Floating-Point Fused Multiply-Add Instructions
-defm : VPatWidenFPMulAccVL_VV_VF<riscv_vfwmadd_vl, "PseudoVFWMACC">;
-defm : VPatWidenFPMulAccVL_VV_VF<riscv_vfwnmadd_vl, "PseudoVFWNMACC">;
-defm : VPatWidenFPMulAccVL_VV_VF<riscv_vfwmsub_vl, "PseudoVFWMSAC">;
-defm : VPatWidenFPMulAccVL_VV_VF<riscv_vfwnmsub_vl, "PseudoVFWNMSAC">;
+defm : VPatWidenFPMulAccVL_VV_VF_RM<riscv_vfwmadd_vl, "PseudoVFWMACC">;
+defm : VPatWidenFPMulAccVL_VV_VF_RM<riscv_vfwnmadd_vl, "PseudoVFWNMACC">;
+defm : VPatWidenFPMulAccVL_VV_VF_RM<riscv_vfwmsub_vl, "PseudoVFWMSAC">;
+defm : VPatWidenFPMulAccVL_VV_VF_RM<riscv_vfwnmsub_vl, "PseudoVFWNMSAC">;
 
 // 13.11. Vector Floating-Point MIN/MAX Instructions
 defm : VPatBinaryFPVL_VV_VF<riscv_fminnum_vl, "PseudoVFMIN">;
diff --git a/llvm/test/CodeGen/RISCV/rvv/masked-tama.ll b/llvm/test/CodeGen/RISCV/rvv/masked-tama.ll
--- a/llvm/test/CodeGen/RISCV/rvv/masked-tama.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/masked-tama.ll
@@ -787,7 +787,7 @@
   <vscale x 1 x half>,
   <vscale x 1 x half>,
   <vscale x 1 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 1 x float>  @intrinsic_vfwmacc_mask_vv_nxv1f32_nxv1f16_nxv1f16(<vscale x 1 x half> %0, <vscale x 1 x half> %1, <vscale x 1 x i1> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmacc_mask_vv_nxv1f32_nxv1f16_nxv1f16:
@@ -802,7 +802,7 @@
     <vscale x 1 x half> %0,
     <vscale x 1 x half> %1,
     <vscale x 1 x i1> %2,
-    iXLen %3, iXLen 3);
+    iXLen 7, iXLen %3, iXLen 3);
 
   ret <vscale x 1 x float> %a
 }
diff --git a/llvm/test/CodeGen/RISCV/rvv/masked-tamu.ll b/llvm/test/CodeGen/RISCV/rvv/masked-tamu.ll
--- a/llvm/test/CodeGen/RISCV/rvv/masked-tamu.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/masked-tamu.ll
@@ -748,7 +748,7 @@
   <vscale x 1 x half>,
   <vscale x 1 x half>,
   <vscale x 1 x i1>,
-  iXLen, iXLen)
+  iXLen, iXLen, iXLen)
 define <vscale x 1 x float>  @intrinsic_vfwmacc_mask_vv_nxv1f32_nxv1f16_nxv1f16(<vscale x 1 x float> %0, <vscale x 1 x half> %1, <vscale x 1 x half> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmacc_mask_vv_nxv1f32_nxv1f16_nxv1f16:
 ; CHECK:       # %bb.0: # %entry
@@ -761,7 +761,7 @@
     <vscale x 1 x half> %1,
     <vscale x 1 x half> %2,
     <vscale x 1 x i1> %3,
-    iXLen %4, iXLen 1)
+    iXLen 7, iXLen %4, iXLen 1)
   ret <vscale x 1 x float> %a
 }
 
diff --git a/llvm/test/CodeGen/RISCV/rvv/masked-tuma.ll b/llvm/test/CodeGen/RISCV/rvv/masked-tuma.ll
--- a/llvm/test/CodeGen/RISCV/rvv/masked-tuma.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/masked-tuma.ll
@@ -748,7 +748,7 @@
   <vscale x 1 x half>,
   <vscale x 1 x half>,
   <vscale x 1 x i1>,
-  iXLen, iXLen)
+  iXLen, iXLen, iXLen)
 define <vscale x 1 x float>  @intrinsic_vfwmacc_mask_vv_nxv1f32_nxv1f16_nxv1f16(<vscale x 1 x float> %0, <vscale x 1 x half> %1, <vscale x 1 x half> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmacc_mask_vv_nxv1f32_nxv1f16_nxv1f16:
 ; CHECK:       # %bb.0: # %entry
@@ -761,7 +761,7 @@
     <vscale x 1 x half> %1,
     <vscale x 1 x half> %2,
     <vscale x 1 x i1> %3,
-    iXLen %4, iXLen 2)
+    iXLen 7, iXLen %4, iXLen 2)
   ret <vscale x 1 x float> %a
 }
 
diff --git a/llvm/test/CodeGen/RISCV/rvv/masked-tumu.ll b/llvm/test/CodeGen/RISCV/rvv/masked-tumu.ll
--- a/llvm/test/CodeGen/RISCV/rvv/masked-tumu.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/masked-tumu.ll
@@ -748,7 +748,7 @@
   <vscale x 1 x half>,
   <vscale x 1 x half>,
   <vscale x 1 x i1>,
-  iXLen, iXLen)
+  iXLen, iXLen, iXLen)
 define <vscale x 1 x float>  @intrinsic_vfwmacc_mask_vv_nxv1f32_nxv1f16_nxv1f16(<vscale x 1 x float> %0, <vscale x 1 x half> %1, <vscale x 1 x half> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmacc_mask_vv_nxv1f32_nxv1f16_nxv1f16:
 ; CHECK:       # %bb.0: # %entry
@@ -761,7 +761,7 @@
     <vscale x 1 x half> %1,
     <vscale x 1 x half> %2,
     <vscale x 1 x i1> %3,
-    iXLen %4, iXLen 0)
+    iXLen 7, iXLen %4, iXLen 0)
   ret <vscale x 1 x float> %a
 }
 
diff --git a/llvm/test/CodeGen/RISCV/rvv/unmasked-ta.ll b/llvm/test/CodeGen/RISCV/rvv/unmasked-ta.ll
--- a/llvm/test/CodeGen/RISCV/rvv/unmasked-ta.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/unmasked-ta.ll
@@ -184,8 +184,7 @@
   <vscale x 1 x float>,
   <vscale x 1 x half>,
   <vscale x 1 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 1 x float>  @intrinsic_vfwmacc_vv_nxv1f32_nxv1f16_nxv1f16(<vscale x 1 x float> %0, <vscale x 1 x half> %1, <vscale x 1 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmacc_vv_nxv1f32_nxv1f16_nxv1f16:
@@ -198,7 +197,7 @@
     <vscale x 1 x float> %0,
     <vscale x 1 x half> %1,
     <vscale x 1 x half> %2,
-    iXLen %3, iXLen 1)
+    iXLen 7, iXLen %3, iXLen 1)
 
   ret <vscale x 1 x float> %a
 }
@@ -207,8 +206,7 @@
   <vscale x 1 x float>,
   <vscale x 1 x half>,
   <vscale x 1 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 1 x float>  @intrinsic_vfwmsac_vv_nxv1f32_nxv1f16_nxv1f16(<vscale x 1 x float> %0, <vscale x 1 x half> %1, <vscale x 1 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmsac_vv_nxv1f32_nxv1f16_nxv1f16:
@@ -221,7 +219,7 @@
     <vscale x 1 x float> %0,
     <vscale x 1 x half> %1,
     <vscale x 1 x half> %2,
-    iXLen %3, iXLen 1)
+    iXLen 7, iXLen %3, iXLen 1)
 
   ret <vscale x 1 x float> %a
 }
@@ -230,8 +228,7 @@
   <vscale x 1 x float>,
   <vscale x 1 x half>,
   <vscale x 1 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 1 x float>  @intrinsic_vfwnmacc_vv_nxv1f32_nxv1f16_nxv1f16(<vscale x 1 x float> %0, <vscale x 1 x half> %1, <vscale x 1 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmacc_vv_nxv1f32_nxv1f16_nxv1f16:
@@ -244,7 +241,7 @@
     <vscale x 1 x float> %0,
     <vscale x 1 x half> %1,
     <vscale x 1 x half> %2,
-    iXLen %3, iXLen 1)
+    iXLen 7, iXLen %3, iXLen 1)
 
   ret <vscale x 1 x float> %a
 }
@@ -253,8 +250,7 @@
   <vscale x 1 x float>,
   <vscale x 1 x half>,
   <vscale x 1 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 1 x float>  @intrinsic_vfwnmsac_vv_nxv1f32_nxv1f16_nxv1f16(<vscale x 1 x float> %0, <vscale x 1 x half> %1, <vscale x 1 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmsac_vv_nxv1f32_nxv1f16_nxv1f16:
@@ -267,7 +263,7 @@
     <vscale x 1 x float> %0,
     <vscale x 1 x half> %1,
     <vscale x 1 x half> %2,
-    iXLen %3, iXLen 1)
+    iXLen 7, iXLen %3, iXLen 1)
 
   ret <vscale x 1 x float> %a
 }
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfwmacc.ll b/llvm/test/CodeGen/RISCV/rvv/vfwmacc.ll
--- a/llvm/test/CodeGen/RISCV/rvv/vfwmacc.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/vfwmacc.ll
@@ -7,21 +7,22 @@
   <vscale x 1 x float>,
   <vscale x 1 x half>,
   <vscale x 1 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 1 x float>  @intrinsic_vfwmacc_vv_nxv1f32_nxv1f16_nxv1f16(<vscale x 1 x float> %0, <vscale x 1 x half> %1, <vscale x 1 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmacc_vv_nxv1f32_nxv1f16_nxv1f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmacc.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.nxv1f16(
     <vscale x 1 x float> %0,
     <vscale x 1 x half> %1,
     <vscale x 1 x half> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 1 x float> %a
 }
@@ -31,13 +32,15 @@
   <vscale x 1 x half>,
   <vscale x 1 x half>,
   <vscale x 1 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 1 x float>  @intrinsic_vfwmacc_mask_vv_nxv1f32_nxv1f16_nxv1f16(<vscale x 1 x float> %0, <vscale x 1 x half> %1, <vscale x 1 x half> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmacc_mask_vv_nxv1f32_nxv1f16_nxv1f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmacc.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.nxv1f16(
@@ -45,7 +48,7 @@
     <vscale x 1 x half> %1,
     <vscale x 1 x half> %2,
     <vscale x 1 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 1 x float> %a
 }
@@ -54,21 +57,22 @@
   <vscale x 2 x float>,
   <vscale x 2 x half>,
   <vscale x 2 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 2 x float>  @intrinsic_vfwmacc_vv_nxv2f32_nxv2f16_nxv2f16(<vscale x 2 x float> %0, <vscale x 2 x half> %1, <vscale x 2 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmacc_vv_nxv2f32_nxv2f16_nxv2f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmacc.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.nxv2f16(
     <vscale x 2 x float> %0,
     <vscale x 2 x half> %1,
     <vscale x 2 x half> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 2 x float> %a
 }
@@ -78,13 +82,15 @@
   <vscale x 2 x half>,
   <vscale x 2 x half>,
   <vscale x 2 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 2 x float>  @intrinsic_vfwmacc_mask_vv_nxv2f32_nxv2f16_nxv2f16(<vscale x 2 x float> %0, <vscale x 2 x half> %1, <vscale x 2 x half> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmacc_mask_vv_nxv2f32_nxv2f16_nxv2f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmacc.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.nxv2f16(
@@ -92,7 +98,7 @@
     <vscale x 2 x half> %1,
     <vscale x 2 x half> %2,
     <vscale x 2 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 2 x float> %a
 }
@@ -101,21 +107,22 @@
   <vscale x 4 x float>,
   <vscale x 4 x half>,
   <vscale x 4 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 4 x float>  @intrinsic_vfwmacc_vv_nxv4f32_nxv4f16_nxv4f16(<vscale x 4 x float> %0, <vscale x 4 x half> %1, <vscale x 4 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmacc_vv_nxv4f32_nxv4f16_nxv4f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m1, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmacc.vv v8, v10, v11
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.nxv4f16(
     <vscale x 4 x float> %0,
     <vscale x 4 x half> %1,
     <vscale x 4 x half> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 4 x float> %a
 }
@@ -125,13 +132,15 @@
   <vscale x 4 x half>,
   <vscale x 4 x half>,
   <vscale x 4 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 4 x float>  @intrinsic_vfwmacc_mask_vv_nxv4f32_nxv4f16_nxv4f16(<vscale x 4 x float> %0, <vscale x 4 x half> %1, <vscale x 4 x half> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmacc_mask_vv_nxv4f32_nxv4f16_nxv4f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m1, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmacc.vv v8, v10, v11, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.nxv4f16(
@@ -139,7 +148,7 @@
     <vscale x 4 x half> %1,
     <vscale x 4 x half> %2,
     <vscale x 4 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 4 x float> %a
 }
@@ -148,21 +157,22 @@
   <vscale x 8 x float>,
   <vscale x 8 x half>,
   <vscale x 8 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 8 x float>  @intrinsic_vfwmacc_vv_nxv8f32_nxv8f16_nxv8f16(<vscale x 8 x float> %0, <vscale x 8 x half> %1, <vscale x 8 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmacc_vv_nxv8f32_nxv8f16_nxv8f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m2, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmacc.vv v8, v12, v14
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.nxv8f16(
     <vscale x 8 x float> %0,
     <vscale x 8 x half> %1,
     <vscale x 8 x half> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 8 x float> %a
 }
@@ -172,13 +182,15 @@
   <vscale x 8 x half>,
   <vscale x 8 x half>,
   <vscale x 8 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 8 x float>  @intrinsic_vfwmacc_mask_vv_nxv8f32_nxv8f16_nxv8f16(<vscale x 8 x float> %0, <vscale x 8 x half> %1, <vscale x 8 x half> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmacc_mask_vv_nxv8f32_nxv8f16_nxv8f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m2, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmacc.vv v8, v12, v14, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.nxv8f16(
@@ -186,7 +198,7 @@
     <vscale x 8 x half> %1,
     <vscale x 8 x half> %2,
     <vscale x 8 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 8 x float> %a
 }
@@ -195,21 +207,22 @@
   <vscale x 16 x float>,
   <vscale x 16 x half>,
   <vscale x 16 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 16 x float>  @intrinsic_vfwmacc_vv_nxv16f32_nxv16f16_nxv16f16(<vscale x 16 x float> %0, <vscale x 16 x half> %1, <vscale x 16 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmacc_vv_nxv16f32_nxv16f16_nxv16f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m4, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmacc.vv v8, v16, v20
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.nxv16f16(
     <vscale x 16 x float> %0,
     <vscale x 16 x half> %1,
     <vscale x 16 x half> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 16 x float> %a
 }
@@ -219,13 +232,15 @@
   <vscale x 16 x half>,
   <vscale x 16 x half>,
   <vscale x 16 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 16 x float>  @intrinsic_vfwmacc_mask_vv_nxv16f32_nxv16f16_nxv16f16(<vscale x 16 x float> %0, <vscale x 16 x half> %1, <vscale x 16 x half> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmacc_mask_vv_nxv16f32_nxv16f16_nxv16f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m4, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmacc.vv v8, v16, v20, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.nxv16f16(
@@ -233,7 +248,7 @@
     <vscale x 16 x half> %1,
     <vscale x 16 x half> %2,
     <vscale x 16 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 16 x float> %a
 }
@@ -242,21 +257,22 @@
   <vscale x 1 x double>,
   <vscale x 1 x float>,
   <vscale x 1 x float>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 1 x double>  @intrinsic_vfwmacc_vv_nxv1f64_nxv1f32_nxv1f32(<vscale x 1 x double> %0, <vscale x 1 x float> %1, <vscale x 1 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmacc_vv_nxv1f64_nxv1f32_nxv1f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmacc.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfwmacc.nxv1f64.nxv1f32(
     <vscale x 1 x double> %0,
     <vscale x 1 x float> %1,
     <vscale x 1 x float> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 1 x double> %a
 }
@@ -266,13 +282,15 @@
   <vscale x 1 x float>,
   <vscale x 1 x float>,
   <vscale x 1 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 1 x double>  @intrinsic_vfwmacc_mask_vv_nxv1f64_nxv1f32_nxv1f32(<vscale x 1 x double> %0, <vscale x 1 x float> %1, <vscale x 1 x float> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmacc_mask_vv_nxv1f64_nxv1f32_nxv1f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmacc.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.nxv1f32(
@@ -280,7 +298,7 @@
     <vscale x 1 x float> %1,
     <vscale x 1 x float> %2,
     <vscale x 1 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 1 x double> %a
 }
@@ -289,21 +307,22 @@
   <vscale x 2 x double>,
   <vscale x 2 x float>,
   <vscale x 2 x float>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 2 x double>  @intrinsic_vfwmacc_vv_nxv2f64_nxv2f32_nxv2f32(<vscale x 2 x double> %0, <vscale x 2 x float> %1, <vscale x 2 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmacc_vv_nxv2f64_nxv2f32_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m1, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmacc.vv v8, v10, v11
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x double> @llvm.riscv.vfwmacc.nxv2f64.nxv2f32(
     <vscale x 2 x double> %0,
     <vscale x 2 x float> %1,
     <vscale x 2 x float> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 2 x double> %a
 }
@@ -313,13 +332,15 @@
   <vscale x 2 x float>,
   <vscale x 2 x float>,
   <vscale x 2 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 2 x double>  @intrinsic_vfwmacc_mask_vv_nxv2f64_nxv2f32_nxv2f32(<vscale x 2 x double> %0, <vscale x 2 x float> %1, <vscale x 2 x float> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmacc_mask_vv_nxv2f64_nxv2f32_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m1, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmacc.vv v8, v10, v11, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.nxv2f32(
@@ -327,7 +348,7 @@
     <vscale x 2 x float> %1,
     <vscale x 2 x float> %2,
     <vscale x 2 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 2 x double> %a
 }
@@ -336,21 +357,22 @@
   <vscale x 4 x double>,
   <vscale x 4 x float>,
   <vscale x 4 x float>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 4 x double>  @intrinsic_vfwmacc_vv_nxv4f64_nxv4f32_nxv4f32(<vscale x 4 x double> %0, <vscale x 4 x float> %1, <vscale x 4 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmacc_vv_nxv4f64_nxv4f32_nxv4f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m2, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmacc.vv v8, v12, v14
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x double> @llvm.riscv.vfwmacc.nxv4f64.nxv4f32(
     <vscale x 4 x double> %0,
     <vscale x 4 x float> %1,
     <vscale x 4 x float> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 4 x double> %a
 }
@@ -360,13 +382,15 @@
   <vscale x 4 x float>,
   <vscale x 4 x float>,
   <vscale x 4 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 4 x double>  @intrinsic_vfwmacc_mask_vv_nxv4f64_nxv4f32_nxv4f32(<vscale x 4 x double> %0, <vscale x 4 x float> %1, <vscale x 4 x float> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmacc_mask_vv_nxv4f64_nxv4f32_nxv4f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m2, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmacc.vv v8, v12, v14, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.nxv4f32(
@@ -374,7 +398,7 @@
     <vscale x 4 x float> %1,
     <vscale x 4 x float> %2,
     <vscale x 4 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 4 x double> %a
 }
@@ -383,21 +407,22 @@
   <vscale x 8 x double>,
   <vscale x 8 x float>,
   <vscale x 8 x float>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 8 x double>  @intrinsic_vfwmacc_vv_nxv8f64_nxv8f32_nxv8f32(<vscale x 8 x double> %0, <vscale x 8 x float> %1, <vscale x 8 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmacc_vv_nxv8f64_nxv8f32_nxv8f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m4, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmacc.vv v8, v16, v20
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 8 x double> @llvm.riscv.vfwmacc.nxv8f64.nxv8f32(
     <vscale x 8 x double> %0,
     <vscale x 8 x float> %1,
     <vscale x 8 x float> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 8 x double> %a
 }
@@ -407,13 +432,15 @@
   <vscale x 8 x float>,
   <vscale x 8 x float>,
   <vscale x 8 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 8 x double>  @intrinsic_vfwmacc_mask_vv_nxv8f64_nxv8f32_nxv8f32(<vscale x 8 x double> %0, <vscale x 8 x float> %1, <vscale x 8 x float> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmacc_mask_vv_nxv8f64_nxv8f32_nxv8f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m4, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmacc.vv v8, v16, v20, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.nxv8f32(
@@ -421,7 +448,7 @@
     <vscale x 8 x float> %1,
     <vscale x 8 x float> %2,
     <vscale x 8 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 8 x double> %a
 }
@@ -430,21 +457,22 @@
   <vscale x 1 x float>,
   half,
   <vscale x 1 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 1 x float>  @intrinsic_vfwmacc_vf_nxv1f32_f16_nxv1f16(<vscale x 1 x float> %0, half %1, <vscale x 1 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmacc_vf_nxv1f32_f16_nxv1f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmacc.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x float> @llvm.riscv.vfwmacc.nxv1f32.f16(
     <vscale x 1 x float> %0,
     half %1,
     <vscale x 1 x half> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 1 x float> %a
 }
@@ -454,13 +482,15 @@
   half,
   <vscale x 1 x half>,
   <vscale x 1 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 1 x float> @intrinsic_vfwmacc_mask_vf_nxv1f32_f16_nxv1f16(<vscale x 1 x float> %0, half %1, <vscale x 1 x half> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmacc_mask_vf_nxv1f32_f16_nxv1f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmacc.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x float> @llvm.riscv.vfwmacc.mask.nxv1f32.f16(
@@ -468,7 +498,7 @@
     half %1,
     <vscale x 1 x half> %2,
     <vscale x 1 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 1 x float> %a
 }
@@ -477,21 +507,22 @@
   <vscale x 2 x float>,
   half,
   <vscale x 2 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 2 x float>  @intrinsic_vfwmacc_vf_nxv2f32_f16_nxv2f16(<vscale x 2 x float> %0, half %1, <vscale x 2 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmacc_vf_nxv2f32_f16_nxv2f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmacc.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfwmacc.nxv2f32.f16(
     <vscale x 2 x float> %0,
     half %1,
     <vscale x 2 x half> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 2 x float> %a
 }
@@ -501,13 +532,15 @@
   half,
   <vscale x 2 x half>,
   <vscale x 2 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfwmacc_mask_vf_nxv2f32_f16_nxv2f16(<vscale x 2 x float> %0, half %1, <vscale x 2 x half> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmacc_mask_vf_nxv2f32_f16_nxv2f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmacc.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfwmacc.mask.nxv2f32.f16(
@@ -515,7 +548,7 @@
     half %1,
     <vscale x 2 x half> %2,
     <vscale x 2 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 2 x float> %a
 }
@@ -524,21 +557,22 @@
   <vscale x 4 x float>,
   half,
   <vscale x 4 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 4 x float>  @intrinsic_vfwmacc_vf_nxv4f32_f16_nxv4f16(<vscale x 4 x float> %0, half %1, <vscale x 4 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmacc_vf_nxv4f32_f16_nxv4f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m1, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmacc.vf v8, fa0, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x float> @llvm.riscv.vfwmacc.nxv4f32.f16(
     <vscale x 4 x float> %0,
     half %1,
     <vscale x 4 x half> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 4 x float> %a
 }
@@ -548,13 +582,15 @@
   half,
   <vscale x 4 x half>,
   <vscale x 4 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 4 x float> @intrinsic_vfwmacc_mask_vf_nxv4f32_f16_nxv4f16(<vscale x 4 x float> %0, half %1, <vscale x 4 x half> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmacc_mask_vf_nxv4f32_f16_nxv4f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m1, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmacc.vf v8, fa0, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x float> @llvm.riscv.vfwmacc.mask.nxv4f32.f16(
@@ -562,7 +598,7 @@
     half %1,
     <vscale x 4 x half> %2,
     <vscale x 4 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 4 x float> %a
 }
@@ -571,21 +607,22 @@
   <vscale x 8 x float>,
   half,
   <vscale x 8 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 8 x float>  @intrinsic_vfwmacc_vf_nxv8f32_f16_nxv8f16(<vscale x 8 x float> %0, half %1, <vscale x 8 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmacc_vf_nxv8f32_f16_nxv8f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m2, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmacc.vf v8, fa0, v12
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 8 x float> @llvm.riscv.vfwmacc.nxv8f32.f16(
     <vscale x 8 x float> %0,
     half %1,
     <vscale x 8 x half> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 8 x float> %a
 }
@@ -595,13 +632,15 @@
   half,
   <vscale x 8 x half>,
   <vscale x 8 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 8 x float> @intrinsic_vfwmacc_mask_vf_nxv8f32_f16_nxv8f16(<vscale x 8 x float> %0, half %1, <vscale x 8 x half> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmacc_mask_vf_nxv8f32_f16_nxv8f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m2, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmacc.vf v8, fa0, v12, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 8 x float> @llvm.riscv.vfwmacc.mask.nxv8f32.f16(
@@ -609,7 +648,7 @@
     half %1,
     <vscale x 8 x half> %2,
     <vscale x 8 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 8 x float> %a
 }
@@ -618,21 +657,22 @@
   <vscale x 16 x float>,
   half,
   <vscale x 16 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 16 x float>  @intrinsic_vfwmacc_vf_nxv16f32_f16_nxv16f16(<vscale x 16 x float> %0, half %1, <vscale x 16 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmacc_vf_nxv16f32_f16_nxv16f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m4, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmacc.vf v8, fa0, v16
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 16 x float> @llvm.riscv.vfwmacc.nxv16f32.f16(
     <vscale x 16 x float> %0,
     half %1,
     <vscale x 16 x half> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 16 x float> %a
 }
@@ -642,13 +682,15 @@
   half,
   <vscale x 16 x half>,
   <vscale x 16 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 16 x float> @intrinsic_vfwmacc_mask_vf_nxv16f32_f16_nxv16f16(<vscale x 16 x float> %0, half %1, <vscale x 16 x half> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmacc_mask_vf_nxv16f32_f16_nxv16f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m4, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmacc.vf v8, fa0, v16, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 16 x float> @llvm.riscv.vfwmacc.mask.nxv16f32.f16(
@@ -656,7 +698,7 @@
     half %1,
     <vscale x 16 x half> %2,
     <vscale x 16 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 16 x float> %a
 }
@@ -665,21 +707,22 @@
   <vscale x 1 x double>,
   float,
   <vscale x 1 x float>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 1 x double>  @intrinsic_vfwmacc_vf_nxv1f64_f32_nxv1f32(<vscale x 1 x double> %0, float %1, <vscale x 1 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmacc_vf_nxv1f64_f32_nxv1f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmacc.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfwmacc.nxv1f64.f32(
     <vscale x 1 x double> %0,
     float %1,
     <vscale x 1 x float> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 1 x double> %a
 }
@@ -689,13 +732,15 @@
   float,
   <vscale x 1 x float>,
   <vscale x 1 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 1 x double> @intrinsic_vfwmacc_mask_vf_nxv1f64_f32_nxv1f32(<vscale x 1 x double> %0, float %1, <vscale x 1 x float> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmacc_mask_vf_nxv1f64_f32_nxv1f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmacc.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfwmacc.mask.nxv1f64.f32(
@@ -703,7 +748,7 @@
     float %1,
     <vscale x 1 x float> %2,
     <vscale x 1 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 1 x double> %a
 }
@@ -712,21 +757,22 @@
   <vscale x 2 x double>,
   float,
   <vscale x 2 x float>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 2 x double>  @intrinsic_vfwmacc_vf_nxv2f64_f32_nxv2f32(<vscale x 2 x double> %0, float %1, <vscale x 2 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmacc_vf_nxv2f64_f32_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m1, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmacc.vf v8, fa0, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x double> @llvm.riscv.vfwmacc.nxv2f64.f32(
     <vscale x 2 x double> %0,
     float %1,
     <vscale x 2 x float> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 2 x double> %a
 }
@@ -736,13 +782,15 @@
   float,
   <vscale x 2 x float>,
   <vscale x 2 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 2 x double> @intrinsic_vfwmacc_mask_vf_nxv2f64_f32_nxv2f32(<vscale x 2 x double> %0, float %1, <vscale x 2 x float> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmacc_mask_vf_nxv2f64_f32_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m1, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmacc.vf v8, fa0, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x double> @llvm.riscv.vfwmacc.mask.nxv2f64.f32(
@@ -750,7 +798,7 @@
     float %1,
     <vscale x 2 x float> %2,
     <vscale x 2 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 2 x double> %a
 }
@@ -759,21 +807,22 @@
   <vscale x 4 x double>,
   float,
   <vscale x 4 x float>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 4 x double>  @intrinsic_vfwmacc_vf_nxv4f64_f32_nxv4f32(<vscale x 4 x double> %0, float %1, <vscale x 4 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmacc_vf_nxv4f64_f32_nxv4f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m2, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmacc.vf v8, fa0, v12
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x double> @llvm.riscv.vfwmacc.nxv4f64.f32(
     <vscale x 4 x double> %0,
     float %1,
     <vscale x 4 x float> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 4 x double> %a
 }
@@ -783,13 +832,15 @@
   float,
   <vscale x 4 x float>,
   <vscale x 4 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 4 x double> @intrinsic_vfwmacc_mask_vf_nxv4f64_f32_nxv4f32(<vscale x 4 x double> %0, float %1, <vscale x 4 x float> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmacc_mask_vf_nxv4f64_f32_nxv4f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m2, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmacc.vf v8, fa0, v12, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x double> @llvm.riscv.vfwmacc.mask.nxv4f64.f32(
@@ -797,7 +848,7 @@
     float %1,
     <vscale x 4 x float> %2,
     <vscale x 4 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 4 x double> %a
 }
@@ -806,21 +857,22 @@
   <vscale x 8 x double>,
   float,
   <vscale x 8 x float>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 8 x double>  @intrinsic_vfwmacc_vf_nxv8f64_f32_nxv8f32(<vscale x 8 x double> %0, float %1, <vscale x 8 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmacc_vf_nxv8f64_f32_nxv8f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m4, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmacc.vf v8, fa0, v16
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 8 x double> @llvm.riscv.vfwmacc.nxv8f64.f32(
     <vscale x 8 x double> %0,
     float %1,
     <vscale x 8 x float> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 8 x double> %a
 }
@@ -830,13 +882,15 @@
   float,
   <vscale x 8 x float>,
   <vscale x 8 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 8 x double> @intrinsic_vfwmacc_mask_vf_nxv8f64_f32_nxv8f32(<vscale x 8 x double> %0, float %1, <vscale x 8 x float> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmacc_mask_vf_nxv8f64_f32_nxv8f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m4, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmacc.vf v8, fa0, v16, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 8 x double> @llvm.riscv.vfwmacc.mask.nxv8f64.f32(
@@ -844,7 +898,7 @@
     float %1,
     <vscale x 8 x float> %2,
     <vscale x 8 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 8 x double> %a
 }
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfwmsac.ll b/llvm/test/CodeGen/RISCV/rvv/vfwmsac.ll
--- a/llvm/test/CodeGen/RISCV/rvv/vfwmsac.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/vfwmsac.ll
@@ -7,21 +7,22 @@
   <vscale x 1 x float>,
   <vscale x 1 x half>,
   <vscale x 1 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 1 x float>  @intrinsic_vfwmsac_vv_nxv1f32_nxv1f16_nxv1f16(<vscale x 1 x float> %0, <vscale x 1 x half> %1, <vscale x 1 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmsac_vv_nxv1f32_nxv1f16_nxv1f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmsac.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.nxv1f16(
     <vscale x 1 x float> %0,
     <vscale x 1 x half> %1,
     <vscale x 1 x half> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 1 x float> %a
 }
@@ -31,13 +32,15 @@
   <vscale x 1 x half>,
   <vscale x 1 x half>,
   <vscale x 1 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 1 x float>  @intrinsic_vfwmsac_mask_vv_nxv1f32_nxv1f16_nxv1f16(<vscale x 1 x float> %0, <vscale x 1 x half> %1, <vscale x 1 x half> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmsac_mask_vv_nxv1f32_nxv1f16_nxv1f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmsac.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.nxv1f16(
@@ -45,7 +48,7 @@
     <vscale x 1 x half> %1,
     <vscale x 1 x half> %2,
     <vscale x 1 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 1 x float> %a
 }
@@ -54,21 +57,22 @@
   <vscale x 2 x float>,
   <vscale x 2 x half>,
   <vscale x 2 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 2 x float>  @intrinsic_vfwmsac_vv_nxv2f32_nxv2f16_nxv2f16(<vscale x 2 x float> %0, <vscale x 2 x half> %1, <vscale x 2 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmsac_vv_nxv2f32_nxv2f16_nxv2f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmsac.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.nxv2f16(
     <vscale x 2 x float> %0,
     <vscale x 2 x half> %1,
     <vscale x 2 x half> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 2 x float> %a
 }
@@ -78,13 +82,15 @@
   <vscale x 2 x half>,
   <vscale x 2 x half>,
   <vscale x 2 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 2 x float>  @intrinsic_vfwmsac_mask_vv_nxv2f32_nxv2f16_nxv2f16(<vscale x 2 x float> %0, <vscale x 2 x half> %1, <vscale x 2 x half> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmsac_mask_vv_nxv2f32_nxv2f16_nxv2f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmsac.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.nxv2f16(
@@ -92,7 +98,7 @@
     <vscale x 2 x half> %1,
     <vscale x 2 x half> %2,
     <vscale x 2 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 2 x float> %a
 }
@@ -101,21 +107,22 @@
   <vscale x 4 x float>,
   <vscale x 4 x half>,
   <vscale x 4 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 4 x float>  @intrinsic_vfwmsac_vv_nxv4f32_nxv4f16_nxv4f16(<vscale x 4 x float> %0, <vscale x 4 x half> %1, <vscale x 4 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmsac_vv_nxv4f32_nxv4f16_nxv4f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m1, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmsac.vv v8, v10, v11
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.nxv4f16(
     <vscale x 4 x float> %0,
     <vscale x 4 x half> %1,
     <vscale x 4 x half> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 4 x float> %a
 }
@@ -125,13 +132,15 @@
   <vscale x 4 x half>,
   <vscale x 4 x half>,
   <vscale x 4 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 4 x float>  @intrinsic_vfwmsac_mask_vv_nxv4f32_nxv4f16_nxv4f16(<vscale x 4 x float> %0, <vscale x 4 x half> %1, <vscale x 4 x half> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmsac_mask_vv_nxv4f32_nxv4f16_nxv4f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m1, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmsac.vv v8, v10, v11, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.nxv4f16(
@@ -139,7 +148,7 @@
     <vscale x 4 x half> %1,
     <vscale x 4 x half> %2,
     <vscale x 4 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 4 x float> %a
 }
@@ -148,21 +157,22 @@
   <vscale x 8 x float>,
   <vscale x 8 x half>,
   <vscale x 8 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 8 x float>  @intrinsic_vfwmsac_vv_nxv8f32_nxv8f16_nxv8f16(<vscale x 8 x float> %0, <vscale x 8 x half> %1, <vscale x 8 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmsac_vv_nxv8f32_nxv8f16_nxv8f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m2, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmsac.vv v8, v12, v14
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.nxv8f16(
     <vscale x 8 x float> %0,
     <vscale x 8 x half> %1,
     <vscale x 8 x half> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 8 x float> %a
 }
@@ -172,13 +182,15 @@
   <vscale x 8 x half>,
   <vscale x 8 x half>,
   <vscale x 8 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 8 x float>  @intrinsic_vfwmsac_mask_vv_nxv8f32_nxv8f16_nxv8f16(<vscale x 8 x float> %0, <vscale x 8 x half> %1, <vscale x 8 x half> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmsac_mask_vv_nxv8f32_nxv8f16_nxv8f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m2, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmsac.vv v8, v12, v14, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.nxv8f16(
@@ -186,7 +198,7 @@
     <vscale x 8 x half> %1,
     <vscale x 8 x half> %2,
     <vscale x 8 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 8 x float> %a
 }
@@ -195,21 +207,22 @@
   <vscale x 16 x float>,
   <vscale x 16 x half>,
   <vscale x 16 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 16 x float>  @intrinsic_vfwmsac_vv_nxv16f32_nxv16f16_nxv16f16(<vscale x 16 x float> %0, <vscale x 16 x half> %1, <vscale x 16 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmsac_vv_nxv16f32_nxv16f16_nxv16f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m4, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmsac.vv v8, v16, v20
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.nxv16f16(
     <vscale x 16 x float> %0,
     <vscale x 16 x half> %1,
     <vscale x 16 x half> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 16 x float> %a
 }
@@ -219,13 +232,15 @@
   <vscale x 16 x half>,
   <vscale x 16 x half>,
   <vscale x 16 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 16 x float>  @intrinsic_vfwmsac_mask_vv_nxv16f32_nxv16f16_nxv16f16(<vscale x 16 x float> %0, <vscale x 16 x half> %1, <vscale x 16 x half> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmsac_mask_vv_nxv16f32_nxv16f16_nxv16f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m4, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmsac.vv v8, v16, v20, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.nxv16f16(
@@ -233,7 +248,7 @@
     <vscale x 16 x half> %1,
     <vscale x 16 x half> %2,
     <vscale x 16 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 16 x float> %a
 }
@@ -242,21 +257,22 @@
   <vscale x 1 x double>,
   <vscale x 1 x float>,
   <vscale x 1 x float>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 1 x double>  @intrinsic_vfwmsac_vv_nxv1f64_nxv1f32_nxv1f32(<vscale x 1 x double> %0, <vscale x 1 x float> %1, <vscale x 1 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmsac_vv_nxv1f64_nxv1f32_nxv1f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmsac.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfwmsac.nxv1f64.nxv1f32(
     <vscale x 1 x double> %0,
     <vscale x 1 x float> %1,
     <vscale x 1 x float> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 1 x double> %a
 }
@@ -266,13 +282,15 @@
   <vscale x 1 x float>,
   <vscale x 1 x float>,
   <vscale x 1 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 1 x double>  @intrinsic_vfwmsac_mask_vv_nxv1f64_nxv1f32_nxv1f32(<vscale x 1 x double> %0, <vscale x 1 x float> %1, <vscale x 1 x float> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmsac_mask_vv_nxv1f64_nxv1f32_nxv1f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmsac.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.nxv1f32(
@@ -280,7 +298,7 @@
     <vscale x 1 x float> %1,
     <vscale x 1 x float> %2,
     <vscale x 1 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 1 x double> %a
 }
@@ -289,21 +307,22 @@
   <vscale x 2 x double>,
   <vscale x 2 x float>,
   <vscale x 2 x float>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 2 x double>  @intrinsic_vfwmsac_vv_nxv2f64_nxv2f32_nxv2f32(<vscale x 2 x double> %0, <vscale x 2 x float> %1, <vscale x 2 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmsac_vv_nxv2f64_nxv2f32_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m1, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmsac.vv v8, v10, v11
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x double> @llvm.riscv.vfwmsac.nxv2f64.nxv2f32(
     <vscale x 2 x double> %0,
     <vscale x 2 x float> %1,
     <vscale x 2 x float> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 2 x double> %a
 }
@@ -313,13 +332,15 @@
   <vscale x 2 x float>,
   <vscale x 2 x float>,
   <vscale x 2 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 2 x double>  @intrinsic_vfwmsac_mask_vv_nxv2f64_nxv2f32_nxv2f32(<vscale x 2 x double> %0, <vscale x 2 x float> %1, <vscale x 2 x float> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmsac_mask_vv_nxv2f64_nxv2f32_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m1, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmsac.vv v8, v10, v11, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.nxv2f32(
@@ -327,7 +348,7 @@
     <vscale x 2 x float> %1,
     <vscale x 2 x float> %2,
     <vscale x 2 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 2 x double> %a
 }
@@ -336,21 +357,22 @@
   <vscale x 4 x double>,
   <vscale x 4 x float>,
   <vscale x 4 x float>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 4 x double>  @intrinsic_vfwmsac_vv_nxv4f64_nxv4f32_nxv4f32(<vscale x 4 x double> %0, <vscale x 4 x float> %1, <vscale x 4 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmsac_vv_nxv4f64_nxv4f32_nxv4f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m2, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmsac.vv v8, v12, v14
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x double> @llvm.riscv.vfwmsac.nxv4f64.nxv4f32(
     <vscale x 4 x double> %0,
     <vscale x 4 x float> %1,
     <vscale x 4 x float> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 4 x double> %a
 }
@@ -360,13 +382,15 @@
   <vscale x 4 x float>,
   <vscale x 4 x float>,
   <vscale x 4 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 4 x double>  @intrinsic_vfwmsac_mask_vv_nxv4f64_nxv4f32_nxv4f32(<vscale x 4 x double> %0, <vscale x 4 x float> %1, <vscale x 4 x float> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmsac_mask_vv_nxv4f64_nxv4f32_nxv4f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m2, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmsac.vv v8, v12, v14, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.nxv4f32(
@@ -374,7 +398,7 @@
     <vscale x 4 x float> %1,
     <vscale x 4 x float> %2,
     <vscale x 4 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 4 x double> %a
 }
@@ -383,21 +407,22 @@
   <vscale x 8 x double>,
   <vscale x 8 x float>,
   <vscale x 8 x float>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 8 x double>  @intrinsic_vfwmsac_vv_nxv8f64_nxv8f32_nxv8f32(<vscale x 8 x double> %0, <vscale x 8 x float> %1, <vscale x 8 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmsac_vv_nxv8f64_nxv8f32_nxv8f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m4, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmsac.vv v8, v16, v20
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 8 x double> @llvm.riscv.vfwmsac.nxv8f64.nxv8f32(
     <vscale x 8 x double> %0,
     <vscale x 8 x float> %1,
     <vscale x 8 x float> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 8 x double> %a
 }
@@ -407,13 +432,15 @@
   <vscale x 8 x float>,
   <vscale x 8 x float>,
   <vscale x 8 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 8 x double>  @intrinsic_vfwmsac_mask_vv_nxv8f64_nxv8f32_nxv8f32(<vscale x 8 x double> %0, <vscale x 8 x float> %1, <vscale x 8 x float> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmsac_mask_vv_nxv8f64_nxv8f32_nxv8f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m4, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmsac.vv v8, v16, v20, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.nxv8f32(
@@ -421,7 +448,7 @@
     <vscale x 8 x float> %1,
     <vscale x 8 x float> %2,
     <vscale x 8 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 8 x double> %a
 }
@@ -430,21 +457,22 @@
   <vscale x 1 x float>,
   half,
   <vscale x 1 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 1 x float>  @intrinsic_vfwmsac_vf_nxv1f32_f16_nxv1f16(<vscale x 1 x float> %0, half %1, <vscale x 1 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmsac_vf_nxv1f32_f16_nxv1f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmsac.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x float> @llvm.riscv.vfwmsac.nxv1f32.f16(
     <vscale x 1 x float> %0,
     half %1,
     <vscale x 1 x half> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 1 x float> %a
 }
@@ -454,13 +482,15 @@
   half,
   <vscale x 1 x half>,
   <vscale x 1 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 1 x float> @intrinsic_vfwmsac_mask_vf_nxv1f32_f16_nxv1f16(<vscale x 1 x float> %0, half %1, <vscale x 1 x half> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmsac_mask_vf_nxv1f32_f16_nxv1f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmsac.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x float> @llvm.riscv.vfwmsac.mask.nxv1f32.f16(
@@ -468,7 +498,7 @@
     half %1,
     <vscale x 1 x half> %2,
     <vscale x 1 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 1 x float> %a
 }
@@ -477,21 +507,22 @@
   <vscale x 2 x float>,
   half,
   <vscale x 2 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 2 x float>  @intrinsic_vfwmsac_vf_nxv2f32_f16_nxv2f16(<vscale x 2 x float> %0, half %1, <vscale x 2 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmsac_vf_nxv2f32_f16_nxv2f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmsac.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfwmsac.nxv2f32.f16(
     <vscale x 2 x float> %0,
     half %1,
     <vscale x 2 x half> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 2 x float> %a
 }
@@ -501,13 +532,15 @@
   half,
   <vscale x 2 x half>,
   <vscale x 2 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfwmsac_mask_vf_nxv2f32_f16_nxv2f16(<vscale x 2 x float> %0, half %1, <vscale x 2 x half> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmsac_mask_vf_nxv2f32_f16_nxv2f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmsac.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfwmsac.mask.nxv2f32.f16(
@@ -515,7 +548,7 @@
     half %1,
     <vscale x 2 x half> %2,
     <vscale x 2 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 2 x float> %a
 }
@@ -524,21 +557,22 @@
   <vscale x 4 x float>,
   half,
   <vscale x 4 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 4 x float>  @intrinsic_vfwmsac_vf_nxv4f32_f16_nxv4f16(<vscale x 4 x float> %0, half %1, <vscale x 4 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmsac_vf_nxv4f32_f16_nxv4f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m1, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmsac.vf v8, fa0, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x float> @llvm.riscv.vfwmsac.nxv4f32.f16(
     <vscale x 4 x float> %0,
     half %1,
     <vscale x 4 x half> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 4 x float> %a
 }
@@ -548,13 +582,15 @@
   half,
   <vscale x 4 x half>,
   <vscale x 4 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 4 x float> @intrinsic_vfwmsac_mask_vf_nxv4f32_f16_nxv4f16(<vscale x 4 x float> %0, half %1, <vscale x 4 x half> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmsac_mask_vf_nxv4f32_f16_nxv4f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m1, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmsac.vf v8, fa0, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x float> @llvm.riscv.vfwmsac.mask.nxv4f32.f16(
@@ -562,7 +598,7 @@
     half %1,
     <vscale x 4 x half> %2,
     <vscale x 4 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 4 x float> %a
 }
@@ -571,21 +607,22 @@
   <vscale x 8 x float>,
   half,
   <vscale x 8 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 8 x float>  @intrinsic_vfwmsac_vf_nxv8f32_f16_nxv8f16(<vscale x 8 x float> %0, half %1, <vscale x 8 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmsac_vf_nxv8f32_f16_nxv8f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m2, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmsac.vf v8, fa0, v12
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 8 x float> @llvm.riscv.vfwmsac.nxv8f32.f16(
     <vscale x 8 x float> %0,
     half %1,
     <vscale x 8 x half> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 8 x float> %a
 }
@@ -595,13 +632,15 @@
   half,
   <vscale x 8 x half>,
   <vscale x 8 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 8 x float> @intrinsic_vfwmsac_mask_vf_nxv8f32_f16_nxv8f16(<vscale x 8 x float> %0, half %1, <vscale x 8 x half> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmsac_mask_vf_nxv8f32_f16_nxv8f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m2, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmsac.vf v8, fa0, v12, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 8 x float> @llvm.riscv.vfwmsac.mask.nxv8f32.f16(
@@ -609,7 +648,7 @@
     half %1,
     <vscale x 8 x half> %2,
     <vscale x 8 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 8 x float> %a
 }
@@ -618,21 +657,22 @@
   <vscale x 16 x float>,
   half,
   <vscale x 16 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 16 x float>  @intrinsic_vfwmsac_vf_nxv16f32_f16_nxv16f16(<vscale x 16 x float> %0, half %1, <vscale x 16 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmsac_vf_nxv16f32_f16_nxv16f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m4, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmsac.vf v8, fa0, v16
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 16 x float> @llvm.riscv.vfwmsac.nxv16f32.f16(
     <vscale x 16 x float> %0,
     half %1,
     <vscale x 16 x half> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 16 x float> %a
 }
@@ -642,13 +682,15 @@
   half,
   <vscale x 16 x half>,
   <vscale x 16 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 16 x float> @intrinsic_vfwmsac_mask_vf_nxv16f32_f16_nxv16f16(<vscale x 16 x float> %0, half %1, <vscale x 16 x half> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmsac_mask_vf_nxv16f32_f16_nxv16f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m4, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmsac.vf v8, fa0, v16, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 16 x float> @llvm.riscv.vfwmsac.mask.nxv16f32.f16(
@@ -656,7 +698,7 @@
     half %1,
     <vscale x 16 x half> %2,
     <vscale x 16 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 16 x float> %a
 }
@@ -665,21 +707,22 @@
   <vscale x 1 x double>,
   float,
   <vscale x 1 x float>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 1 x double>  @intrinsic_vfwmsac_vf_nxv1f64_f32_nxv1f32(<vscale x 1 x double> %0, float %1, <vscale x 1 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmsac_vf_nxv1f64_f32_nxv1f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmsac.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfwmsac.nxv1f64.f32(
     <vscale x 1 x double> %0,
     float %1,
     <vscale x 1 x float> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 1 x double> %a
 }
@@ -689,13 +732,15 @@
   float,
   <vscale x 1 x float>,
   <vscale x 1 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 1 x double> @intrinsic_vfwmsac_mask_vf_nxv1f64_f32_nxv1f32(<vscale x 1 x double> %0, float %1, <vscale x 1 x float> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmsac_mask_vf_nxv1f64_f32_nxv1f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmsac.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfwmsac.mask.nxv1f64.f32(
@@ -703,7 +748,7 @@
     float %1,
     <vscale x 1 x float> %2,
     <vscale x 1 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 1 x double> %a
 }
@@ -712,21 +757,22 @@
   <vscale x 2 x double>,
   float,
   <vscale x 2 x float>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 2 x double>  @intrinsic_vfwmsac_vf_nxv2f64_f32_nxv2f32(<vscale x 2 x double> %0, float %1, <vscale x 2 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmsac_vf_nxv2f64_f32_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m1, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmsac.vf v8, fa0, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x double> @llvm.riscv.vfwmsac.nxv2f64.f32(
     <vscale x 2 x double> %0,
     float %1,
     <vscale x 2 x float> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 2 x double> %a
 }
@@ -736,13 +782,15 @@
   float,
   <vscale x 2 x float>,
   <vscale x 2 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 2 x double> @intrinsic_vfwmsac_mask_vf_nxv2f64_f32_nxv2f32(<vscale x 2 x double> %0, float %1, <vscale x 2 x float> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmsac_mask_vf_nxv2f64_f32_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m1, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmsac.vf v8, fa0, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x double> @llvm.riscv.vfwmsac.mask.nxv2f64.f32(
@@ -750,7 +798,7 @@
     float %1,
     <vscale x 2 x float> %2,
     <vscale x 2 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 2 x double> %a
 }
@@ -759,21 +807,22 @@
   <vscale x 4 x double>,
   float,
   <vscale x 4 x float>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 4 x double>  @intrinsic_vfwmsac_vf_nxv4f64_f32_nxv4f32(<vscale x 4 x double> %0, float %1, <vscale x 4 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmsac_vf_nxv4f64_f32_nxv4f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m2, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmsac.vf v8, fa0, v12
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x double> @llvm.riscv.vfwmsac.nxv4f64.f32(
     <vscale x 4 x double> %0,
     float %1,
     <vscale x 4 x float> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 4 x double> %a
 }
@@ -783,13 +832,15 @@
   float,
   <vscale x 4 x float>,
   <vscale x 4 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 4 x double> @intrinsic_vfwmsac_mask_vf_nxv4f64_f32_nxv4f32(<vscale x 4 x double> %0, float %1, <vscale x 4 x float> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmsac_mask_vf_nxv4f64_f32_nxv4f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m2, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmsac.vf v8, fa0, v12, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x double> @llvm.riscv.vfwmsac.mask.nxv4f64.f32(
@@ -797,7 +848,7 @@
     float %1,
     <vscale x 4 x float> %2,
     <vscale x 4 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 4 x double> %a
 }
@@ -806,21 +857,22 @@
   <vscale x 8 x double>,
   float,
   <vscale x 8 x float>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 8 x double>  @intrinsic_vfwmsac_vf_nxv8f64_f32_nxv8f32(<vscale x 8 x double> %0, float %1, <vscale x 8 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmsac_vf_nxv8f64_f32_nxv8f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m4, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmsac.vf v8, fa0, v16
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 8 x double> @llvm.riscv.vfwmsac.nxv8f64.f32(
     <vscale x 8 x double> %0,
     float %1,
     <vscale x 8 x float> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 8 x double> %a
 }
@@ -830,13 +882,15 @@
   float,
   <vscale x 8 x float>,
   <vscale x 8 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 8 x double> @intrinsic_vfwmsac_mask_vf_nxv8f64_f32_nxv8f32(<vscale x 8 x double> %0, float %1, <vscale x 8 x float> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwmsac_mask_vf_nxv8f64_f32_nxv8f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m4, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwmsac.vf v8, fa0, v16, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 8 x double> @llvm.riscv.vfwmsac.mask.nxv8f64.f32(
@@ -844,7 +898,7 @@
     float %1,
     <vscale x 8 x float> %2,
     <vscale x 8 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 8 x double> %a
 }
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfwnmacc.ll b/llvm/test/CodeGen/RISCV/rvv/vfwnmacc.ll
--- a/llvm/test/CodeGen/RISCV/rvv/vfwnmacc.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/vfwnmacc.ll
@@ -7,21 +7,22 @@
   <vscale x 1 x float>,
   <vscale x 1 x half>,
   <vscale x 1 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 1 x float>  @intrinsic_vfwnmacc_vv_nxv1f32_nxv1f16_nxv1f16(<vscale x 1 x float> %0, <vscale x 1 x half> %1, <vscale x 1 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmacc_vv_nxv1f32_nxv1f16_nxv1f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmacc.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.nxv1f16(
     <vscale x 1 x float> %0,
     <vscale x 1 x half> %1,
     <vscale x 1 x half> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 1 x float> %a
 }
@@ -31,13 +32,15 @@
   <vscale x 1 x half>,
   <vscale x 1 x half>,
   <vscale x 1 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 1 x float>  @intrinsic_vfwnmacc_mask_vv_nxv1f32_nxv1f16_nxv1f16(<vscale x 1 x float> %0, <vscale x 1 x half> %1, <vscale x 1 x half> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmacc_mask_vv_nxv1f32_nxv1f16_nxv1f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmacc.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.nxv1f16(
@@ -45,7 +48,7 @@
     <vscale x 1 x half> %1,
     <vscale x 1 x half> %2,
     <vscale x 1 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 1 x float> %a
 }
@@ -54,21 +57,22 @@
   <vscale x 2 x float>,
   <vscale x 2 x half>,
   <vscale x 2 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 2 x float>  @intrinsic_vfwnmacc_vv_nxv2f32_nxv2f16_nxv2f16(<vscale x 2 x float> %0, <vscale x 2 x half> %1, <vscale x 2 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmacc_vv_nxv2f32_nxv2f16_nxv2f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmacc.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.nxv2f16(
     <vscale x 2 x float> %0,
     <vscale x 2 x half> %1,
     <vscale x 2 x half> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 2 x float> %a
 }
@@ -78,13 +82,15 @@
   <vscale x 2 x half>,
   <vscale x 2 x half>,
   <vscale x 2 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 2 x float>  @intrinsic_vfwnmacc_mask_vv_nxv2f32_nxv2f16_nxv2f16(<vscale x 2 x float> %0, <vscale x 2 x half> %1, <vscale x 2 x half> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmacc_mask_vv_nxv2f32_nxv2f16_nxv2f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmacc.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.nxv2f16(
@@ -92,7 +98,7 @@
     <vscale x 2 x half> %1,
     <vscale x 2 x half> %2,
     <vscale x 2 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 2 x float> %a
 }
@@ -101,21 +107,22 @@
   <vscale x 4 x float>,
   <vscale x 4 x half>,
   <vscale x 4 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 4 x float>  @intrinsic_vfwnmacc_vv_nxv4f32_nxv4f16_nxv4f16(<vscale x 4 x float> %0, <vscale x 4 x half> %1, <vscale x 4 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmacc_vv_nxv4f32_nxv4f16_nxv4f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m1, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmacc.vv v8, v10, v11
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.nxv4f16(
     <vscale x 4 x float> %0,
     <vscale x 4 x half> %1,
     <vscale x 4 x half> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 4 x float> %a
 }
@@ -125,13 +132,15 @@
   <vscale x 4 x half>,
   <vscale x 4 x half>,
   <vscale x 4 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 4 x float>  @intrinsic_vfwnmacc_mask_vv_nxv4f32_nxv4f16_nxv4f16(<vscale x 4 x float> %0, <vscale x 4 x half> %1, <vscale x 4 x half> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmacc_mask_vv_nxv4f32_nxv4f16_nxv4f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m1, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmacc.vv v8, v10, v11, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.nxv4f16(
@@ -139,7 +148,7 @@
     <vscale x 4 x half> %1,
     <vscale x 4 x half> %2,
     <vscale x 4 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 4 x float> %a
 }
@@ -148,21 +157,22 @@
   <vscale x 8 x float>,
   <vscale x 8 x half>,
   <vscale x 8 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 8 x float>  @intrinsic_vfwnmacc_vv_nxv8f32_nxv8f16_nxv8f16(<vscale x 8 x float> %0, <vscale x 8 x half> %1, <vscale x 8 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmacc_vv_nxv8f32_nxv8f16_nxv8f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m2, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmacc.vv v8, v12, v14
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.nxv8f16(
     <vscale x 8 x float> %0,
     <vscale x 8 x half> %1,
     <vscale x 8 x half> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 8 x float> %a
 }
@@ -172,13 +182,15 @@
   <vscale x 8 x half>,
   <vscale x 8 x half>,
   <vscale x 8 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 8 x float>  @intrinsic_vfwnmacc_mask_vv_nxv8f32_nxv8f16_nxv8f16(<vscale x 8 x float> %0, <vscale x 8 x half> %1, <vscale x 8 x half> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmacc_mask_vv_nxv8f32_nxv8f16_nxv8f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m2, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmacc.vv v8, v12, v14, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.nxv8f16(
@@ -186,7 +198,7 @@
     <vscale x 8 x half> %1,
     <vscale x 8 x half> %2,
     <vscale x 8 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 8 x float> %a
 }
@@ -195,21 +207,22 @@
   <vscale x 16 x float>,
   <vscale x 16 x half>,
   <vscale x 16 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 16 x float>  @intrinsic_vfwnmacc_vv_nxv16f32_nxv16f16_nxv16f16(<vscale x 16 x float> %0, <vscale x 16 x half> %1, <vscale x 16 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmacc_vv_nxv16f32_nxv16f16_nxv16f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m4, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmacc.vv v8, v16, v20
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.nxv16f16(
     <vscale x 16 x float> %0,
     <vscale x 16 x half> %1,
     <vscale x 16 x half> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 16 x float> %a
 }
@@ -219,13 +232,15 @@
   <vscale x 16 x half>,
   <vscale x 16 x half>,
   <vscale x 16 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 16 x float>  @intrinsic_vfwnmacc_mask_vv_nxv16f32_nxv16f16_nxv16f16(<vscale x 16 x float> %0, <vscale x 16 x half> %1, <vscale x 16 x half> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmacc_mask_vv_nxv16f32_nxv16f16_nxv16f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m4, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmacc.vv v8, v16, v20, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.nxv16f16(
@@ -233,7 +248,7 @@
     <vscale x 16 x half> %1,
     <vscale x 16 x half> %2,
     <vscale x 16 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 16 x float> %a
 }
@@ -242,21 +257,22 @@
   <vscale x 1 x double>,
   <vscale x 1 x float>,
   <vscale x 1 x float>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 1 x double>  @intrinsic_vfwnmacc_vv_nxv1f64_nxv1f32_nxv1f32(<vscale x 1 x double> %0, <vscale x 1 x float> %1, <vscale x 1 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmacc_vv_nxv1f64_nxv1f32_nxv1f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmacc.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.nxv1f64.nxv1f32(
     <vscale x 1 x double> %0,
     <vscale x 1 x float> %1,
     <vscale x 1 x float> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 1 x double> %a
 }
@@ -266,13 +282,15 @@
   <vscale x 1 x float>,
   <vscale x 1 x float>,
   <vscale x 1 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 1 x double>  @intrinsic_vfwnmacc_mask_vv_nxv1f64_nxv1f32_nxv1f32(<vscale x 1 x double> %0, <vscale x 1 x float> %1, <vscale x 1 x float> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmacc_mask_vv_nxv1f64_nxv1f32_nxv1f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmacc.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.nxv1f32(
@@ -280,7 +298,7 @@
     <vscale x 1 x float> %1,
     <vscale x 1 x float> %2,
     <vscale x 1 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 1 x double> %a
 }
@@ -289,21 +307,22 @@
   <vscale x 2 x double>,
   <vscale x 2 x float>,
   <vscale x 2 x float>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 2 x double>  @intrinsic_vfwnmacc_vv_nxv2f64_nxv2f32_nxv2f32(<vscale x 2 x double> %0, <vscale x 2 x float> %1, <vscale x 2 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmacc_vv_nxv2f64_nxv2f32_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m1, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmacc.vv v8, v10, v11
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.nxv2f64.nxv2f32(
     <vscale x 2 x double> %0,
     <vscale x 2 x float> %1,
     <vscale x 2 x float> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 2 x double> %a
 }
@@ -313,13 +332,15 @@
   <vscale x 2 x float>,
   <vscale x 2 x float>,
   <vscale x 2 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 2 x double>  @intrinsic_vfwnmacc_mask_vv_nxv2f64_nxv2f32_nxv2f32(<vscale x 2 x double> %0, <vscale x 2 x float> %1, <vscale x 2 x float> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmacc_mask_vv_nxv2f64_nxv2f32_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m1, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmacc.vv v8, v10, v11, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.nxv2f32(
@@ -327,7 +348,7 @@
     <vscale x 2 x float> %1,
     <vscale x 2 x float> %2,
     <vscale x 2 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 2 x double> %a
 }
@@ -336,21 +357,22 @@
   <vscale x 4 x double>,
   <vscale x 4 x float>,
   <vscale x 4 x float>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 4 x double>  @intrinsic_vfwnmacc_vv_nxv4f64_nxv4f32_nxv4f32(<vscale x 4 x double> %0, <vscale x 4 x float> %1, <vscale x 4 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmacc_vv_nxv4f64_nxv4f32_nxv4f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m2, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmacc.vv v8, v12, v14
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.nxv4f64.nxv4f32(
     <vscale x 4 x double> %0,
     <vscale x 4 x float> %1,
     <vscale x 4 x float> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 4 x double> %a
 }
@@ -360,13 +382,15 @@
   <vscale x 4 x float>,
   <vscale x 4 x float>,
   <vscale x 4 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 4 x double>  @intrinsic_vfwnmacc_mask_vv_nxv4f64_nxv4f32_nxv4f32(<vscale x 4 x double> %0, <vscale x 4 x float> %1, <vscale x 4 x float> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmacc_mask_vv_nxv4f64_nxv4f32_nxv4f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m2, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmacc.vv v8, v12, v14, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.nxv4f32(
@@ -374,7 +398,7 @@
     <vscale x 4 x float> %1,
     <vscale x 4 x float> %2,
     <vscale x 4 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 4 x double> %a
 }
@@ -383,21 +407,22 @@
   <vscale x 8 x double>,
   <vscale x 8 x float>,
   <vscale x 8 x float>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 8 x double>  @intrinsic_vfwnmacc_vv_nxv8f64_nxv8f32_nxv8f32(<vscale x 8 x double> %0, <vscale x 8 x float> %1, <vscale x 8 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmacc_vv_nxv8f64_nxv8f32_nxv8f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m4, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmacc.vv v8, v16, v20
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.nxv8f64.nxv8f32(
     <vscale x 8 x double> %0,
     <vscale x 8 x float> %1,
     <vscale x 8 x float> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 8 x double> %a
 }
@@ -407,13 +432,15 @@
   <vscale x 8 x float>,
   <vscale x 8 x float>,
   <vscale x 8 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 8 x double>  @intrinsic_vfwnmacc_mask_vv_nxv8f64_nxv8f32_nxv8f32(<vscale x 8 x double> %0, <vscale x 8 x float> %1, <vscale x 8 x float> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmacc_mask_vv_nxv8f64_nxv8f32_nxv8f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m4, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmacc.vv v8, v16, v20, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.nxv8f32(
@@ -421,7 +448,7 @@
     <vscale x 8 x float> %1,
     <vscale x 8 x float> %2,
     <vscale x 8 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 8 x double> %a
 }
@@ -430,21 +457,22 @@
   <vscale x 1 x float>,
   half,
   <vscale x 1 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 1 x float>  @intrinsic_vfwnmacc_vf_nxv1f32_f16_nxv1f16(<vscale x 1 x float> %0, half %1, <vscale x 1 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmacc_vf_nxv1f32_f16_nxv1f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.nxv1f32.f16(
     <vscale x 1 x float> %0,
     half %1,
     <vscale x 1 x half> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 1 x float> %a
 }
@@ -454,13 +482,15 @@
   half,
   <vscale x 1 x half>,
   <vscale x 1 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 1 x float> @intrinsic_vfwnmacc_mask_vf_nxv1f32_f16_nxv1f16(<vscale x 1 x float> %0, half %1, <vscale x 1 x half> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmacc_mask_vf_nxv1f32_f16_nxv1f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x float> @llvm.riscv.vfwnmacc.mask.nxv1f32.f16(
@@ -468,7 +498,7 @@
     half %1,
     <vscale x 1 x half> %2,
     <vscale x 1 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 1 x float> %a
 }
@@ -477,21 +507,22 @@
   <vscale x 2 x float>,
   half,
   <vscale x 2 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 2 x float>  @intrinsic_vfwnmacc_vf_nxv2f32_f16_nxv2f16(<vscale x 2 x float> %0, half %1, <vscale x 2 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmacc_vf_nxv2f32_f16_nxv2f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.nxv2f32.f16(
     <vscale x 2 x float> %0,
     half %1,
     <vscale x 2 x half> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 2 x float> %a
 }
@@ -501,13 +532,15 @@
   half,
   <vscale x 2 x half>,
   <vscale x 2 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfwnmacc_mask_vf_nxv2f32_f16_nxv2f16(<vscale x 2 x float> %0, half %1, <vscale x 2 x half> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmacc_mask_vf_nxv2f32_f16_nxv2f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfwnmacc.mask.nxv2f32.f16(
@@ -515,7 +548,7 @@
     half %1,
     <vscale x 2 x half> %2,
     <vscale x 2 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 2 x float> %a
 }
@@ -524,21 +557,22 @@
   <vscale x 4 x float>,
   half,
   <vscale x 4 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 4 x float>  @intrinsic_vfwnmacc_vf_nxv4f32_f16_nxv4f16(<vscale x 4 x float> %0, half %1, <vscale x 4 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmacc_vf_nxv4f32_f16_nxv4f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m1, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.nxv4f32.f16(
     <vscale x 4 x float> %0,
     half %1,
     <vscale x 4 x half> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 4 x float> %a
 }
@@ -548,13 +582,15 @@
   half,
   <vscale x 4 x half>,
   <vscale x 4 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 4 x float> @intrinsic_vfwnmacc_mask_vf_nxv4f32_f16_nxv4f16(<vscale x 4 x float> %0, half %1, <vscale x 4 x half> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmacc_mask_vf_nxv4f32_f16_nxv4f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m1, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x float> @llvm.riscv.vfwnmacc.mask.nxv4f32.f16(
@@ -562,7 +598,7 @@
     half %1,
     <vscale x 4 x half> %2,
     <vscale x 4 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 4 x float> %a
 }
@@ -571,21 +607,22 @@
   <vscale x 8 x float>,
   half,
   <vscale x 8 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 8 x float>  @intrinsic_vfwnmacc_vf_nxv8f32_f16_nxv8f16(<vscale x 8 x float> %0, half %1, <vscale x 8 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmacc_vf_nxv8f32_f16_nxv8f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m2, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v12
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.nxv8f32.f16(
     <vscale x 8 x float> %0,
     half %1,
     <vscale x 8 x half> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 8 x float> %a
 }
@@ -595,13 +632,15 @@
   half,
   <vscale x 8 x half>,
   <vscale x 8 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 8 x float> @intrinsic_vfwnmacc_mask_vf_nxv8f32_f16_nxv8f16(<vscale x 8 x float> %0, half %1, <vscale x 8 x half> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmacc_mask_vf_nxv8f32_f16_nxv8f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m2, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v12, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 8 x float> @llvm.riscv.vfwnmacc.mask.nxv8f32.f16(
@@ -609,7 +648,7 @@
     half %1,
     <vscale x 8 x half> %2,
     <vscale x 8 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 8 x float> %a
 }
@@ -618,21 +657,22 @@
   <vscale x 16 x float>,
   half,
   <vscale x 16 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 16 x float>  @intrinsic_vfwnmacc_vf_nxv16f32_f16_nxv16f16(<vscale x 16 x float> %0, half %1, <vscale x 16 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmacc_vf_nxv16f32_f16_nxv16f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m4, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v16
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.nxv16f32.f16(
     <vscale x 16 x float> %0,
     half %1,
     <vscale x 16 x half> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 16 x float> %a
 }
@@ -642,13 +682,15 @@
   half,
   <vscale x 16 x half>,
   <vscale x 16 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 16 x float> @intrinsic_vfwnmacc_mask_vf_nxv16f32_f16_nxv16f16(<vscale x 16 x float> %0, half %1, <vscale x 16 x half> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmacc_mask_vf_nxv16f32_f16_nxv16f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m4, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v16, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 16 x float> @llvm.riscv.vfwnmacc.mask.nxv16f32.f16(
@@ -656,7 +698,7 @@
     half %1,
     <vscale x 16 x half> %2,
     <vscale x 16 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 16 x float> %a
 }
@@ -665,21 +707,22 @@
   <vscale x 1 x double>,
   float,
   <vscale x 1 x float>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 1 x double>  @intrinsic_vfwnmacc_vf_nxv1f64_f32_nxv1f32(<vscale x 1 x double> %0, float %1, <vscale x 1 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmacc_vf_nxv1f64_f32_nxv1f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.nxv1f64.f32(
     <vscale x 1 x double> %0,
     float %1,
     <vscale x 1 x float> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 1 x double> %a
 }
@@ -689,13 +732,15 @@
   float,
   <vscale x 1 x float>,
   <vscale x 1 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 1 x double> @intrinsic_vfwnmacc_mask_vf_nxv1f64_f32_nxv1f32(<vscale x 1 x double> %0, float %1, <vscale x 1 x float> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmacc_mask_vf_nxv1f64_f32_nxv1f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfwnmacc.mask.nxv1f64.f32(
@@ -703,7 +748,7 @@
     float %1,
     <vscale x 1 x float> %2,
     <vscale x 1 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 1 x double> %a
 }
@@ -712,21 +757,22 @@
   <vscale x 2 x double>,
   float,
   <vscale x 2 x float>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 2 x double>  @intrinsic_vfwnmacc_vf_nxv2f64_f32_nxv2f32(<vscale x 2 x double> %0, float %1, <vscale x 2 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmacc_vf_nxv2f64_f32_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m1, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.nxv2f64.f32(
     <vscale x 2 x double> %0,
     float %1,
     <vscale x 2 x float> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 2 x double> %a
 }
@@ -736,13 +782,15 @@
   float,
   <vscale x 2 x float>,
   <vscale x 2 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 2 x double> @intrinsic_vfwnmacc_mask_vf_nxv2f64_f32_nxv2f32(<vscale x 2 x double> %0, float %1, <vscale x 2 x float> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmacc_mask_vf_nxv2f64_f32_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m1, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x double> @llvm.riscv.vfwnmacc.mask.nxv2f64.f32(
@@ -750,7 +798,7 @@
     float %1,
     <vscale x 2 x float> %2,
     <vscale x 2 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 2 x double> %a
 }
@@ -759,21 +807,22 @@
   <vscale x 4 x double>,
   float,
   <vscale x 4 x float>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 4 x double>  @intrinsic_vfwnmacc_vf_nxv4f64_f32_nxv4f32(<vscale x 4 x double> %0, float %1, <vscale x 4 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmacc_vf_nxv4f64_f32_nxv4f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m2, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v12
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.nxv4f64.f32(
     <vscale x 4 x double> %0,
     float %1,
     <vscale x 4 x float> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 4 x double> %a
 }
@@ -783,13 +832,15 @@
   float,
   <vscale x 4 x float>,
   <vscale x 4 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 4 x double> @intrinsic_vfwnmacc_mask_vf_nxv4f64_f32_nxv4f32(<vscale x 4 x double> %0, float %1, <vscale x 4 x float> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmacc_mask_vf_nxv4f64_f32_nxv4f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m2, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v12, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x double> @llvm.riscv.vfwnmacc.mask.nxv4f64.f32(
@@ -797,7 +848,7 @@
     float %1,
     <vscale x 4 x float> %2,
     <vscale x 4 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 4 x double> %a
 }
@@ -806,21 +857,22 @@
   <vscale x 8 x double>,
   float,
   <vscale x 8 x float>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 8 x double>  @intrinsic_vfwnmacc_vf_nxv8f64_f32_nxv8f32(<vscale x 8 x double> %0, float %1, <vscale x 8 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmacc_vf_nxv8f64_f32_nxv8f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m4, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v16
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.nxv8f64.f32(
     <vscale x 8 x double> %0,
     float %1,
     <vscale x 8 x float> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 8 x double> %a
 }
@@ -830,13 +882,15 @@
   float,
   <vscale x 8 x float>,
   <vscale x 8 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 8 x double> @intrinsic_vfwnmacc_mask_vf_nxv8f64_f32_nxv8f32(<vscale x 8 x double> %0, float %1, <vscale x 8 x float> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmacc_mask_vf_nxv8f64_f32_nxv8f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m4, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v16, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 8 x double> @llvm.riscv.vfwnmacc.mask.nxv8f64.f32(
@@ -844,7 +898,7 @@
     float %1,
     <vscale x 8 x float> %2,
     <vscale x 8 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 8 x double> %a
 }
diff --git a/llvm/test/CodeGen/RISCV/rvv/vfwnmsac.ll b/llvm/test/CodeGen/RISCV/rvv/vfwnmsac.ll
--- a/llvm/test/CodeGen/RISCV/rvv/vfwnmsac.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/vfwnmsac.ll
@@ -7,21 +7,22 @@
   <vscale x 1 x float>,
   <vscale x 1 x half>,
   <vscale x 1 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 1 x float>  @intrinsic_vfwnmsac_vv_nxv1f32_nxv1f16_nxv1f16(<vscale x 1 x float> %0, <vscale x 1 x half> %1, <vscale x 1 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmsac_vv_nxv1f32_nxv1f16_nxv1f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmsac.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.nxv1f16(
     <vscale x 1 x float> %0,
     <vscale x 1 x half> %1,
     <vscale x 1 x half> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 1 x float> %a
 }
@@ -31,13 +32,15 @@
   <vscale x 1 x half>,
   <vscale x 1 x half>,
   <vscale x 1 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 1 x float>  @intrinsic_vfwnmsac_mask_vv_nxv1f32_nxv1f16_nxv1f16(<vscale x 1 x float> %0, <vscale x 1 x half> %1, <vscale x 1 x half> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmsac_mask_vv_nxv1f32_nxv1f16_nxv1f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmsac.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.nxv1f16(
@@ -45,7 +48,7 @@
     <vscale x 1 x half> %1,
     <vscale x 1 x half> %2,
     <vscale x 1 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 1 x float> %a
 }
@@ -54,21 +57,22 @@
   <vscale x 2 x float>,
   <vscale x 2 x half>,
   <vscale x 2 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 2 x float>  @intrinsic_vfwnmsac_vv_nxv2f32_nxv2f16_nxv2f16(<vscale x 2 x float> %0, <vscale x 2 x half> %1, <vscale x 2 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmsac_vv_nxv2f32_nxv2f16_nxv2f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmsac.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.nxv2f16(
     <vscale x 2 x float> %0,
     <vscale x 2 x half> %1,
     <vscale x 2 x half> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 2 x float> %a
 }
@@ -78,13 +82,15 @@
   <vscale x 2 x half>,
   <vscale x 2 x half>,
   <vscale x 2 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 2 x float>  @intrinsic_vfwnmsac_mask_vv_nxv2f32_nxv2f16_nxv2f16(<vscale x 2 x float> %0, <vscale x 2 x half> %1, <vscale x 2 x half> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmsac_mask_vv_nxv2f32_nxv2f16_nxv2f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmsac.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.nxv2f16(
@@ -92,7 +98,7 @@
     <vscale x 2 x half> %1,
     <vscale x 2 x half> %2,
     <vscale x 2 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 2 x float> %a
 }
@@ -101,21 +107,22 @@
   <vscale x 4 x float>,
   <vscale x 4 x half>,
   <vscale x 4 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 4 x float>  @intrinsic_vfwnmsac_vv_nxv4f32_nxv4f16_nxv4f16(<vscale x 4 x float> %0, <vscale x 4 x half> %1, <vscale x 4 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmsac_vv_nxv4f32_nxv4f16_nxv4f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m1, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmsac.vv v8, v10, v11
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.nxv4f16(
     <vscale x 4 x float> %0,
     <vscale x 4 x half> %1,
     <vscale x 4 x half> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 4 x float> %a
 }
@@ -125,13 +132,15 @@
   <vscale x 4 x half>,
   <vscale x 4 x half>,
   <vscale x 4 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 4 x float>  @intrinsic_vfwnmsac_mask_vv_nxv4f32_nxv4f16_nxv4f16(<vscale x 4 x float> %0, <vscale x 4 x half> %1, <vscale x 4 x half> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmsac_mask_vv_nxv4f32_nxv4f16_nxv4f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m1, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmsac.vv v8, v10, v11, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.nxv4f16(
@@ -139,7 +148,7 @@
     <vscale x 4 x half> %1,
     <vscale x 4 x half> %2,
     <vscale x 4 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 4 x float> %a
 }
@@ -148,21 +157,22 @@
   <vscale x 8 x float>,
   <vscale x 8 x half>,
   <vscale x 8 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 8 x float>  @intrinsic_vfwnmsac_vv_nxv8f32_nxv8f16_nxv8f16(<vscale x 8 x float> %0, <vscale x 8 x half> %1, <vscale x 8 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmsac_vv_nxv8f32_nxv8f16_nxv8f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m2, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmsac.vv v8, v12, v14
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.nxv8f16(
     <vscale x 8 x float> %0,
     <vscale x 8 x half> %1,
     <vscale x 8 x half> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 8 x float> %a
 }
@@ -172,13 +182,15 @@
   <vscale x 8 x half>,
   <vscale x 8 x half>,
   <vscale x 8 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 8 x float>  @intrinsic_vfwnmsac_mask_vv_nxv8f32_nxv8f16_nxv8f16(<vscale x 8 x float> %0, <vscale x 8 x half> %1, <vscale x 8 x half> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmsac_mask_vv_nxv8f32_nxv8f16_nxv8f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m2, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmsac.vv v8, v12, v14, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.nxv8f16(
@@ -186,7 +198,7 @@
     <vscale x 8 x half> %1,
     <vscale x 8 x half> %2,
     <vscale x 8 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 8 x float> %a
 }
@@ -195,21 +207,22 @@
   <vscale x 16 x float>,
   <vscale x 16 x half>,
   <vscale x 16 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 16 x float>  @intrinsic_vfwnmsac_vv_nxv16f32_nxv16f16_nxv16f16(<vscale x 16 x float> %0, <vscale x 16 x half> %1, <vscale x 16 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmsac_vv_nxv16f32_nxv16f16_nxv16f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m4, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmsac.vv v8, v16, v20
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.nxv16f16(
     <vscale x 16 x float> %0,
     <vscale x 16 x half> %1,
     <vscale x 16 x half> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 16 x float> %a
 }
@@ -219,13 +232,15 @@
   <vscale x 16 x half>,
   <vscale x 16 x half>,
   <vscale x 16 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 16 x float>  @intrinsic_vfwnmsac_mask_vv_nxv16f32_nxv16f16_nxv16f16(<vscale x 16 x float> %0, <vscale x 16 x half> %1, <vscale x 16 x half> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmsac_mask_vv_nxv16f32_nxv16f16_nxv16f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m4, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmsac.vv v8, v16, v20, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.nxv16f16(
@@ -233,7 +248,7 @@
     <vscale x 16 x half> %1,
     <vscale x 16 x half> %2,
     <vscale x 16 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 16 x float> %a
 }
@@ -242,21 +257,22 @@
   <vscale x 1 x double>,
   <vscale x 1 x float>,
   <vscale x 1 x float>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 1 x double>  @intrinsic_vfwnmsac_vv_nxv1f64_nxv1f32_nxv1f32(<vscale x 1 x double> %0, <vscale x 1 x float> %1, <vscale x 1 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmsac_vv_nxv1f64_nxv1f32_nxv1f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmsac.vv v8, v9, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.nxv1f64.nxv1f32(
     <vscale x 1 x double> %0,
     <vscale x 1 x float> %1,
     <vscale x 1 x float> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 1 x double> %a
 }
@@ -266,13 +282,15 @@
   <vscale x 1 x float>,
   <vscale x 1 x float>,
   <vscale x 1 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 1 x double>  @intrinsic_vfwnmsac_mask_vv_nxv1f64_nxv1f32_nxv1f32(<vscale x 1 x double> %0, <vscale x 1 x float> %1, <vscale x 1 x float> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmsac_mask_vv_nxv1f64_nxv1f32_nxv1f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmsac.vv v8, v9, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.nxv1f32(
@@ -280,7 +298,7 @@
     <vscale x 1 x float> %1,
     <vscale x 1 x float> %2,
     <vscale x 1 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 1 x double> %a
 }
@@ -289,21 +307,22 @@
   <vscale x 2 x double>,
   <vscale x 2 x float>,
   <vscale x 2 x float>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 2 x double>  @intrinsic_vfwnmsac_vv_nxv2f64_nxv2f32_nxv2f32(<vscale x 2 x double> %0, <vscale x 2 x float> %1, <vscale x 2 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmsac_vv_nxv2f64_nxv2f32_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m1, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmsac.vv v8, v10, v11
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.nxv2f64.nxv2f32(
     <vscale x 2 x double> %0,
     <vscale x 2 x float> %1,
     <vscale x 2 x float> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 2 x double> %a
 }
@@ -313,13 +332,15 @@
   <vscale x 2 x float>,
   <vscale x 2 x float>,
   <vscale x 2 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 2 x double>  @intrinsic_vfwnmsac_mask_vv_nxv2f64_nxv2f32_nxv2f32(<vscale x 2 x double> %0, <vscale x 2 x float> %1, <vscale x 2 x float> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmsac_mask_vv_nxv2f64_nxv2f32_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m1, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmsac.vv v8, v10, v11, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.nxv2f32(
@@ -327,7 +348,7 @@
     <vscale x 2 x float> %1,
     <vscale x 2 x float> %2,
     <vscale x 2 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 2 x double> %a
 }
@@ -336,21 +357,22 @@
   <vscale x 4 x double>,
   <vscale x 4 x float>,
   <vscale x 4 x float>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 4 x double>  @intrinsic_vfwnmsac_vv_nxv4f64_nxv4f32_nxv4f32(<vscale x 4 x double> %0, <vscale x 4 x float> %1, <vscale x 4 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmsac_vv_nxv4f64_nxv4f32_nxv4f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m2, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmsac.vv v8, v12, v14
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.nxv4f64.nxv4f32(
     <vscale x 4 x double> %0,
     <vscale x 4 x float> %1,
     <vscale x 4 x float> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 4 x double> %a
 }
@@ -360,13 +382,15 @@
   <vscale x 4 x float>,
   <vscale x 4 x float>,
   <vscale x 4 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 4 x double>  @intrinsic_vfwnmsac_mask_vv_nxv4f64_nxv4f32_nxv4f32(<vscale x 4 x double> %0, <vscale x 4 x float> %1, <vscale x 4 x float> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmsac_mask_vv_nxv4f64_nxv4f32_nxv4f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m2, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmsac.vv v8, v12, v14, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.nxv4f32(
@@ -374,7 +398,7 @@
     <vscale x 4 x float> %1,
     <vscale x 4 x float> %2,
     <vscale x 4 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 4 x double> %a
 }
@@ -383,21 +407,22 @@
   <vscale x 8 x double>,
   <vscale x 8 x float>,
   <vscale x 8 x float>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 8 x double>  @intrinsic_vfwnmsac_vv_nxv8f64_nxv8f32_nxv8f32(<vscale x 8 x double> %0, <vscale x 8 x float> %1, <vscale x 8 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmsac_vv_nxv8f64_nxv8f32_nxv8f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m4, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmsac.vv v8, v16, v20
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.nxv8f64.nxv8f32(
     <vscale x 8 x double> %0,
     <vscale x 8 x float> %1,
     <vscale x 8 x float> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 8 x double> %a
 }
@@ -407,13 +432,15 @@
   <vscale x 8 x float>,
   <vscale x 8 x float>,
   <vscale x 8 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 8 x double>  @intrinsic_vfwnmsac_mask_vv_nxv8f64_nxv8f32_nxv8f32(<vscale x 8 x double> %0, <vscale x 8 x float> %1, <vscale x 8 x float> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmsac_mask_vv_nxv8f64_nxv8f32_nxv8f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m4, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmsac.vv v8, v16, v20, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.nxv8f32(
@@ -421,7 +448,7 @@
     <vscale x 8 x float> %1,
     <vscale x 8 x float> %2,
     <vscale x 8 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 8 x double> %a
 }
@@ -430,21 +457,22 @@
   <vscale x 1 x float>,
   half,
   <vscale x 1 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 1 x float>  @intrinsic_vfwnmsac_vf_nxv1f32_f16_nxv1f16(<vscale x 1 x float> %0, half %1, <vscale x 1 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmsac_vf_nxv1f32_f16_nxv1f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.nxv1f32.f16(
     <vscale x 1 x float> %0,
     half %1,
     <vscale x 1 x half> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 1 x float> %a
 }
@@ -454,13 +482,15 @@
   half,
   <vscale x 1 x half>,
   <vscale x 1 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 1 x float> @intrinsic_vfwnmsac_mask_vf_nxv1f32_f16_nxv1f16(<vscale x 1 x float> %0, half %1, <vscale x 1 x half> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmsac_mask_vf_nxv1f32_f16_nxv1f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x float> @llvm.riscv.vfwnmsac.mask.nxv1f32.f16(
@@ -468,7 +498,7 @@
     half %1,
     <vscale x 1 x half> %2,
     <vscale x 1 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 1 x float> %a
 }
@@ -477,21 +507,22 @@
   <vscale x 2 x float>,
   half,
   <vscale x 2 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 2 x float>  @intrinsic_vfwnmsac_vf_nxv2f32_f16_nxv2f16(<vscale x 2 x float> %0, half %1, <vscale x 2 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmsac_vf_nxv2f32_f16_nxv2f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.nxv2f32.f16(
     <vscale x 2 x float> %0,
     half %1,
     <vscale x 2 x half> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 2 x float> %a
 }
@@ -501,13 +532,15 @@
   half,
   <vscale x 2 x half>,
   <vscale x 2 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 2 x float> @intrinsic_vfwnmsac_mask_vf_nxv2f32_f16_nxv2f16(<vscale x 2 x float> %0, half %1, <vscale x 2 x half> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmsac_mask_vf_nxv2f32_f16_nxv2f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x float> @llvm.riscv.vfwnmsac.mask.nxv2f32.f16(
@@ -515,7 +548,7 @@
     half %1,
     <vscale x 2 x half> %2,
     <vscale x 2 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 2 x float> %a
 }
@@ -524,21 +557,22 @@
   <vscale x 4 x float>,
   half,
   <vscale x 4 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 4 x float>  @intrinsic_vfwnmsac_vf_nxv4f32_f16_nxv4f16(<vscale x 4 x float> %0, half %1, <vscale x 4 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmsac_vf_nxv4f32_f16_nxv4f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m1, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.nxv4f32.f16(
     <vscale x 4 x float> %0,
     half %1,
     <vscale x 4 x half> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 4 x float> %a
 }
@@ -548,13 +582,15 @@
   half,
   <vscale x 4 x half>,
   <vscale x 4 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 4 x float> @intrinsic_vfwnmsac_mask_vf_nxv4f32_f16_nxv4f16(<vscale x 4 x float> %0, half %1, <vscale x 4 x half> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmsac_mask_vf_nxv4f32_f16_nxv4f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m1, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x float> @llvm.riscv.vfwnmsac.mask.nxv4f32.f16(
@@ -562,7 +598,7 @@
     half %1,
     <vscale x 4 x half> %2,
     <vscale x 4 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 4 x float> %a
 }
@@ -571,21 +607,22 @@
   <vscale x 8 x float>,
   half,
   <vscale x 8 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 8 x float>  @intrinsic_vfwnmsac_vf_nxv8f32_f16_nxv8f16(<vscale x 8 x float> %0, half %1, <vscale x 8 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmsac_vf_nxv8f32_f16_nxv8f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m2, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v12
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.nxv8f32.f16(
     <vscale x 8 x float> %0,
     half %1,
     <vscale x 8 x half> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 8 x float> %a
 }
@@ -595,13 +632,15 @@
   half,
   <vscale x 8 x half>,
   <vscale x 8 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 8 x float> @intrinsic_vfwnmsac_mask_vf_nxv8f32_f16_nxv8f16(<vscale x 8 x float> %0, half %1, <vscale x 8 x half> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmsac_mask_vf_nxv8f32_f16_nxv8f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m2, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v12, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 8 x float> @llvm.riscv.vfwnmsac.mask.nxv8f32.f16(
@@ -609,7 +648,7 @@
     half %1,
     <vscale x 8 x half> %2,
     <vscale x 8 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 8 x float> %a
 }
@@ -618,21 +657,22 @@
   <vscale x 16 x float>,
   half,
   <vscale x 16 x half>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 16 x float>  @intrinsic_vfwnmsac_vf_nxv16f32_f16_nxv16f16(<vscale x 16 x float> %0, half %1, <vscale x 16 x half> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmsac_vf_nxv16f32_f16_nxv16f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m4, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v16
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.nxv16f32.f16(
     <vscale x 16 x float> %0,
     half %1,
     <vscale x 16 x half> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 16 x float> %a
 }
@@ -642,13 +682,15 @@
   half,
   <vscale x 16 x half>,
   <vscale x 16 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 16 x float> @intrinsic_vfwnmsac_mask_vf_nxv16f32_f16_nxv16f16(<vscale x 16 x float> %0, half %1, <vscale x 16 x half> %2, <vscale x 16 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmsac_mask_vf_nxv16f32_f16_nxv16f16:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m4, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v16, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 16 x float> @llvm.riscv.vfwnmsac.mask.nxv16f32.f16(
@@ -656,7 +698,7 @@
     half %1,
     <vscale x 16 x half> %2,
     <vscale x 16 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 16 x float> %a
 }
@@ -665,21 +707,22 @@
   <vscale x 1 x double>,
   float,
   <vscale x 1 x float>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 1 x double>  @intrinsic_vfwnmsac_vf_nxv1f64_f32_nxv1f32(<vscale x 1 x double> %0, float %1, <vscale x 1 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmsac_vf_nxv1f64_f32_nxv1f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v9
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.nxv1f64.f32(
     <vscale x 1 x double> %0,
     float %1,
     <vscale x 1 x float> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 1 x double> %a
 }
@@ -689,13 +732,15 @@
   float,
   <vscale x 1 x float>,
   <vscale x 1 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 1 x double> @intrinsic_vfwnmsac_mask_vf_nxv1f64_f32_nxv1f32(<vscale x 1 x double> %0, float %1, <vscale x 1 x float> %2, <vscale x 1 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmsac_mask_vf_nxv1f64_f32_nxv1f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v9, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 1 x double> @llvm.riscv.vfwnmsac.mask.nxv1f64.f32(
@@ -703,7 +748,7 @@
     float %1,
     <vscale x 1 x float> %2,
     <vscale x 1 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 1 x double> %a
 }
@@ -712,21 +757,22 @@
   <vscale x 2 x double>,
   float,
   <vscale x 2 x float>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 2 x double>  @intrinsic_vfwnmsac_vf_nxv2f64_f32_nxv2f32(<vscale x 2 x double> %0, float %1, <vscale x 2 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmsac_vf_nxv2f64_f32_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m1, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v10
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.nxv2f64.f32(
     <vscale x 2 x double> %0,
     float %1,
     <vscale x 2 x float> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 2 x double> %a
 }
@@ -736,13 +782,15 @@
   float,
   <vscale x 2 x float>,
   <vscale x 2 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 2 x double> @intrinsic_vfwnmsac_mask_vf_nxv2f64_f32_nxv2f32(<vscale x 2 x double> %0, float %1, <vscale x 2 x float> %2, <vscale x 2 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmsac_mask_vf_nxv2f64_f32_nxv2f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m1, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v10, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 2 x double> @llvm.riscv.vfwnmsac.mask.nxv2f64.f32(
@@ -750,7 +798,7 @@
     float %1,
     <vscale x 2 x float> %2,
     <vscale x 2 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 2 x double> %a
 }
@@ -759,21 +807,22 @@
   <vscale x 4 x double>,
   float,
   <vscale x 4 x float>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 4 x double>  @intrinsic_vfwnmsac_vf_nxv4f64_f32_nxv4f32(<vscale x 4 x double> %0, float %1, <vscale x 4 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmsac_vf_nxv4f64_f32_nxv4f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m2, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v12
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.nxv4f64.f32(
     <vscale x 4 x double> %0,
     float %1,
     <vscale x 4 x float> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 4 x double> %a
 }
@@ -783,13 +832,15 @@
   float,
   <vscale x 4 x float>,
   <vscale x 4 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 4 x double> @intrinsic_vfwnmsac_mask_vf_nxv4f64_f32_nxv4f32(<vscale x 4 x double> %0, float %1, <vscale x 4 x float> %2, <vscale x 4 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmsac_mask_vf_nxv4f64_f32_nxv4f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m2, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v12, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 4 x double> @llvm.riscv.vfwnmsac.mask.nxv4f64.f32(
@@ -797,7 +848,7 @@
     float %1,
     <vscale x 4 x float> %2,
     <vscale x 4 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 4 x double> %a
 }
@@ -806,21 +857,22 @@
   <vscale x 8 x double>,
   float,
   <vscale x 8 x float>,
-  iXLen,
-  iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 8 x double>  @intrinsic_vfwnmsac_vf_nxv8f64_f32_nxv8f32(<vscale x 8 x double> %0, float %1, <vscale x 8 x float> %2, iXLen %3) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmsac_vf_nxv8f64_f32_nxv8f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m4, tu, ma
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v16
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.nxv8f64.f32(
     <vscale x 8 x double> %0,
     float %1,
     <vscale x 8 x float> %2,
-    iXLen %3, iXLen 0)
+    iXLen 0, iXLen %3, iXLen 0)
 
   ret <vscale x 8 x double> %a
 }
@@ -830,13 +882,15 @@
   float,
   <vscale x 8 x float>,
   <vscale x 8 x i1>,
-  iXLen, iXLen);
+  iXLen, iXLen, iXLen);
 
 define <vscale x 8 x double> @intrinsic_vfwnmsac_mask_vf_nxv8f64_f32_nxv8f32(<vscale x 8 x double> %0, float %1, <vscale x 8 x float> %2, <vscale x 8 x i1> %3, iXLen %4) nounwind {
 ; CHECK-LABEL: intrinsic_vfwnmsac_mask_vf_nxv8f64_f32_nxv8f32:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m4, tu, mu
+; CHECK-NEXT:    fsrmi a0, 0
 ; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v16, v0.t
+; CHECK-NEXT:    fsrm a0
 ; CHECK-NEXT:    ret
 entry:
   %a = call <vscale x 8 x double> @llvm.riscv.vfwnmsac.mask.nxv8f64.f32(
@@ -844,7 +898,7 @@
     float %1,
     <vscale x 8 x float> %2,
     <vscale x 8 x i1> %3,
-    iXLen %4, iXLen 0);
+    iXLen 0, iXLen %4, iXLen 0)
 
   ret <vscale x 8 x double> %a
 }