[RISCV] Support ISD::STRICT_FADD/FSUB/FMUL/FDIV/FMA for scalable vector types.

[fakepaper56]

Summary

It makes RISC-V serve llvm.experimental.constrained.*.

[craig.topper]

llvm/test/CodeGen/RISCV/rvv/vfwadd-constrained-sdnode.ll

15

Shouldn't these be constrained.fpext?

17

Please use "fpexcept.strict"

[fakepaper56]

llvm/test/CodeGen/RISCV/rvv/vfwadd-constrained-sdnode.ll

15

I think you are right. I misunderstand it.

llvm/lib/Target/RISCV/RISCVISelLowering.cpp

const auto SetCommonVFPActions = [&](MVT VT) {
           VT, Custom);
 
       setOperationAction(ISD::VECTOR_DEINTERLEAVE, VT, Custom);
       setOperationAction(ISD::VECTOR_INTERLEAVE, VT, Custom);
 
       setOperationAction({ISD::VECTOR_REVERSE, ISD::VECTOR_SPLICE}, VT, Custom);
 
       setOperationAction(FloatingPointVPOps, VT, Custom);
+      setOperationAction({ISD::STRICT_FADD, ISD::STRICT_FSUB, ISD::STRICT_FMUL,
+                          ISD::STRICT_FDIV, ISD::STRICT_FMA},
+                         VT, Legal);
     };
 
     // Sets common extload/truncstore actions on RVV floating-point vector
     // types.
     const auto SetCommonVFPExtLoadTruncStoreActions =
         [&](MVT VT, ArrayRef<MVT::SimpleValueType> SmallerVTs) {
           for (auto SmallVT : SmallerVTs) {
             setTruncStoreAction(VT, SmallVT, Expand);

llvm/lib/Target/RISCV/RISCVInstrInfoVSDPatterns.td

multiclass VPatWidenBinaryFPSDNode_VV_VF_WV_WF<SDNode op, string instruction_name> {
   defm : VPatWidenBinaryFPSDNode_VV_VF<op, instruction_name>;
   defm : VPatWidenBinaryFPSDNode_WV_WF<op, instruction_name>;
 }
 
 multiclass VPatWidenFPMulAccSDNode_VV_VF<string instruction_name> {
   foreach vtiToWti = AllWidenableFloatVectors in {
     defvar vti = vtiToWti.Vti;
     defvar wti = vtiToWti.Wti;
-    def : Pat<(fma (wti.Vector (riscv_fpextend_vl_oneuse
+    def : Pat<(any_fma (wti.Vector (riscv_fpextend_vl_oneuse
                                        (vti.Vector vti.RegClass:$rs1),
-                                    (vti.Mask true_mask), (XLenVT srcvalue))),
+                                       (vti.Mask true_mask),
+                                       (XLenVT srcvalue))),
                        (wti.Vector (riscv_fpextend_vl_oneuse
                                        (vti.Vector vti.RegClass:$rs2),
-                                    (vti.Mask true_mask), (XLenVT srcvalue))),
+                                       (vti.Mask true_mask),
+                                       (XLenVT srcvalue))),
                        (wti.Vector wti.RegClass:$rd)),
               (!cast<Instruction>(instruction_name#"_VV_"#vti.LMul.MX)
                  wti.RegClass:$rd, vti.RegClass:$rs1, vti.RegClass:$rs2,
                  vti.AVL, vti.Log2SEW, TAIL_AGNOSTIC)>;
-    def : Pat<(fma (wti.Vector (SplatFPOp
+    def : Pat<(any_fma (wti.Vector (SplatFPOp
                                        (fpext_oneuse vti.ScalarRegClass:$rs1))),
                        (wti.Vector (riscv_fpextend_vl_oneuse
                                        (vti.Vector vti.RegClass:$rs2),
-                                    (vti.Mask true_mask), (XLenVT srcvalue))),
+                                       (vti.Mask true_mask),
+                                       (XLenVT srcvalue))),
                        (wti.Vector wti.RegClass:$rd)),
               (!cast<Instruction>(instruction_name#"_V"#vti.ScalarSuffix#"_"#vti.LMul.MX)
                  wti.RegClass:$rd, vti.ScalarRegClass:$rs1, vti.RegClass:$rs2,
                  vti.AVL, vti.Log2SEW, TAIL_AGNOSTIC)>;
   }
 }
 
 multiclass VPatWidenFPNegMulAccSDNode_VV_VF<string instruction_name> {
   foreach vtiToWti = AllWidenableFloatVectors in {
     defvar vti = vtiToWti.Vti;
     defvar wti = vtiToWti.Wti;
-    def : Pat<(fma (fneg (wti.Vector (riscv_fpextend_vl_oneuse
+    def : Pat<(any_fma (fneg (wti.Vector (riscv_fpextend_vl_oneuse
                                              (vti.Vector vti.RegClass:$rs1),
-                                          (vti.Mask true_mask), (XLenVT srcvalue)))),
+                                             (vti.Mask true_mask),
+                                             (XLenVT srcvalue)))),
                        (riscv_fpextend_vl_oneuse (vti.Vector vti.RegClass:$rs2),
-                                             (vti.Mask true_mask), (XLenVT srcvalue)),
+                                                 (vti.Mask true_mask),
+                                                 (XLenVT srcvalue)),
                        (fneg wti.RegClass:$rd)),
               (!cast<Instruction>(instruction_name#"_VV_"#vti.LMul.MX)
                  wti.RegClass:$rd, vti.RegClass:$rs1, vti.RegClass:$rs2,
                  vti.AVL, vti.Log2SEW, TAIL_AGNOSTIC)>;
-    def : Pat<(fma (SplatFPOp (fpext_oneuse vti.ScalarRegClass:$rs1)),
+    def : Pat<(any_fma (SplatFPOp (fpext_oneuse vti.ScalarRegClass:$rs1)),
                        (fneg (wti.Vector (riscv_fpextend_vl_oneuse
                                              (vti.Vector vti.RegClass:$rs2),
-                                          (vti.Mask true_mask), (XLenVT srcvalue)))),
+                                             (vti.Mask true_mask),
+                                             (XLenVT srcvalue)))),
                        (fneg wti.RegClass:$rd)),
               (!cast<Instruction>(instruction_name#"_V"#vti.ScalarSuffix#"_"#vti.LMul.MX)
                  wti.RegClass:$rd, vti.ScalarRegClass:$rs1, vti.RegClass:$rs2,
                  vti.AVL, vti.Log2SEW, TAIL_AGNOSTIC)>;
-    def : Pat<(fma (fneg (wti.Vector (SplatFPOp (fpext_oneuse vti.ScalarRegClass:$rs1)))),
+    def : Pat<(any_fma (fneg (wti.Vector (SplatFPOp (fpext_oneuse vti.ScalarRegClass:$rs1)))),
                        (riscv_fpextend_vl_oneuse (vti.Vector vti.RegClass:$rs2),
-                                             (vti.Mask true_mask), (XLenVT srcvalue)),
+                                                 (vti.Mask true_mask),
+                                                 (XLenVT srcvalue)),
                        (fneg wti.RegClass:$rd)),
               (!cast<Instruction>(instruction_name#"_V"#vti.ScalarSuffix#"_"#vti.LMul.MX)
                  wti.RegClass:$rd, vti.ScalarRegClass:$rs1, vti.RegClass:$rs2,
                  vti.AVL, vti.Log2SEW, TAIL_AGNOSTIC)>;
   }
 }
 
 multiclass VPatWidenFPMulSacSDNode_VV_VF<string instruction_name> {
   foreach vtiToWti = AllWidenableFloatVectors in {
     defvar vti = vtiToWti.Vti;
     defvar wti = vtiToWti.Wti;
-    def : Pat<(fma (wti.Vector (riscv_fpextend_vl_oneuse
+    def : Pat<(any_fma (wti.Vector (riscv_fpextend_vl_oneuse
                                     (vti.Vector vti.RegClass:$rs1),
                                     (vti.Mask true_mask), (XLenVT srcvalue))),
                        (riscv_fpextend_vl_oneuse (vti.Vector vti.RegClass:$rs2),
-                                             (vti.Mask true_mask), (XLenVT srcvalue)),
+                                                 (vti.Mask true_mask),
+                                                 (XLenVT srcvalue)),
                        (fneg wti.RegClass:$rd)),
               (!cast<Instruction>(instruction_name#"_VV_"#vti.LMul.MX)
                  wti.RegClass:$rd, vti.RegClass:$rs1, vti.RegClass:$rs2,
                  vti.AVL, vti.Log2SEW, TAIL_AGNOSTIC)>;
-    def : Pat<(fma (wti.Vector (SplatFPOp (fpext_oneuse vti.ScalarRegClass:$rs1))),
+    def : Pat<(any_fma (wti.Vector (SplatFPOp
+                                       (fpext_oneuse vti.ScalarRegClass:$rs1))),
                        (riscv_fpextend_vl_oneuse (vti.Vector vti.RegClass:$rs2),
-                                             (vti.Mask true_mask), (XLenVT srcvalue)),
+                                                 (vti.Mask true_mask),
+                                                 (XLenVT srcvalue)),
                        (fneg wti.RegClass:$rd)),
               (!cast<Instruction>(instruction_name#"_V"#vti.ScalarSuffix#"_"#vti.LMul.MX)
                  wti.RegClass:$rd, vti.ScalarRegClass:$rs1, vti.RegClass:$rs2,
                  vti.AVL, vti.Log2SEW, TAIL_AGNOSTIC)>;
   }
 }
 
 multiclass VPatWidenFPNegMulSacSDNode_VV_VF<string instruction_name> {
   foreach vtiToWti = AllWidenableFloatVectors in {
     defvar vti = vtiToWti.Vti;
     defvar wti = vtiToWti.Wti;
-    def : Pat<(fma (fneg (wti.Vector (riscv_fpextend_vl_oneuse
-                                          (vti.Vector vti.RegClass:$rs1),
-                                          (vti.Mask true_mask), (XLenVT srcvalue)))),
+    def : Pat<(any_fma (fneg (wti.Vector (riscv_fpextend_vl_oneuse
+                                         (vti.Mask true_mask),
+                                         (XLenVT srcvalue)))),
                        (riscv_fpextend_vl_oneuse (vti.Vector vti.RegClass:$rs2),
-                                             (vti.Mask true_mask), (XLenVT srcvalue)),
+                                                 (vti.Mask true_mask),
+                                                 (XLenVT srcvalue)),
                        wti.RegClass:$rd),
               (!cast<Instruction>(instruction_name#"_VV_"#vti.LMul.MX)
                  wti.RegClass:$rd, vti.RegClass:$rs1, vti.RegClass:$rs2,
                  vti.AVL, vti.Log2SEW, TAIL_AGNOSTIC)>;
-    def : Pat<(fma (wti.Vector (SplatFPOp (fpext_oneuse vti.ScalarRegClass:$rs1))),
+    def : Pat<(any_fma (wti.Vector (SplatFPOp
+                                       (fpext_oneuse vti.ScalarRegClass:$rs1))),
                        (fneg (wti.Vector (riscv_fpextend_vl_oneuse
-                                          (vti.Vector vti.RegClass:$rs2),
-                                          (vti.Mask true_mask), (XLenVT srcvalue)))),
+                                         (vti.Mask true_mask),
+                                         (XLenVT srcvalue)))),
                        wti.RegClass:$rd),
               (!cast<Instruction>(instruction_name#"_V"#vti.ScalarSuffix#"_"#vti.LMul.MX)
                  wti.RegClass:$rd, vti.ScalarRegClass:$rs1, vti.RegClass:$rs2,
                  vti.AVL, vti.Log2SEW, TAIL_AGNOSTIC)>;
-    def : Pat<(fma (fneg (wti.Vector (SplatFPOp (fpext_oneuse vti.ScalarRegClass:$rs1)))),
+    def : Pat<(any_fma (fneg (wti.Vector (SplatFPOp (fpext_oneuse vti.ScalarRegClass:$rs1)))),
                        (riscv_fpextend_vl_oneuse (vti.Vector vti.RegClass:$rs2),
-                                             (vti.Mask true_mask), (XLenVT srcvalue)),
+                                                 (vti.Mask true_mask),
+                                                 (XLenVT srcvalue)),
                        wti.RegClass:$rd),
               (!cast<Instruction>(instruction_name#"_V"#vti.ScalarSuffix#"_"#vti.LMul.MX)
                  wti.RegClass:$rd, vti.ScalarRegClass:$rs1, vti.RegClass:$rs2,
                  vti.AVL, vti.Log2SEW, TAIL_AGNOSTIC)>;
   }
 }
 
 multiclass VPatMultiplyAddSDNode_VV_VX<SDNode op, string instruction_name> {
   foreach vti = AllIntegerVectors in {

 } // Predicates = [HasVInstructions]
 
 // 13. Vector Floating-Point Instructions
 
 let Predicates = [HasVInstructionsAnyF] in {
 
 // 13.2. Vector Single-Width Floating-Point Add/Subtract Instructions
 defm : VPatBinaryFPSDNode_VV_VF<fadd, "PseudoVFADD">;
+defm : VPatBinaryFPSDNode_VV_VF<strict_fadd, "PseudoVFADD">;
 defm : VPatBinaryFPSDNode_VV_VF<fsub, "PseudoVFSUB">;
+defm : VPatBinaryFPSDNode_VV_VF<strict_fsub, "PseudoVFSUB">;
 defm : VPatBinaryFPSDNode_R_VF<fsub, "PseudoVFRSUB">;
+defm : VPatBinaryFPSDNode_R_VF<strict_fsub, "PseudoVFRSUB">;
 
 // 13.3. Vector Widening Floating-Point Add/Subtract Instructions
 defm : VPatWidenBinaryFPSDNode_VV_VF_WV_WF<fadd, "PseudoVFWADD">;
+defm : VPatWidenBinaryFPSDNode_VV_VF_WV_WF<strict_fadd, "PseudoVFWADD">;
 defm : VPatWidenBinaryFPSDNode_VV_VF_WV_WF<fsub, "PseudoVFWSUB">;
+defm : VPatWidenBinaryFPSDNode_VV_VF_WV_WF<strict_fsub, "PseudoVFWSUB">;
 
 // 13.4. Vector Single-Width Floating-Point Multiply/Divide Instructions
 defm : VPatBinaryFPSDNode_VV_VF<fmul, "PseudoVFMUL">;
+defm : VPatBinaryFPSDNode_VV_VF<strict_fmul, "PseudoVFMUL">;
 defm : VPatBinaryFPSDNode_VV_VF<fdiv, "PseudoVFDIV">;
+defm : VPatBinaryFPSDNode_VV_VF<strict_fdiv, "PseudoVFDIV">;
 defm : VPatBinaryFPSDNode_R_VF<fdiv, "PseudoVFRDIV">;
+defm : VPatBinaryFPSDNode_R_VF<strict_fdiv, "PseudoVFRDIV">;
 
 // 13.5. Vector Widening Floating-Point Multiply Instructions
 defm : VPatWidenBinaryFPSDNode_VV_VF<fmul, "PseudoVFWMUL">;
+defm : VPatWidenBinaryFPSDNode_VV_VF<strict_fmul, "PseudoVFWMUL">;
 
 // 13.6 Vector Single-Width Floating-Point Fused Multiply-Add Instructions.
 foreach fvti = AllFloatVectors in {
   // NOTE: We choose VFMADD because it has the most commuting freedom. So it
   // works best with how TwoAddressInstructionPass tries commuting.
   defvar suffix = fvti.LMul.MX;
-  def : Pat<(fvti.Vector (fma fvti.RegClass:$rs1, fvti.RegClass:$rd,
+  def : Pat<(fvti.Vector (any_fma fvti.RegClass:$rs1, fvti.RegClass:$rd,
                                   fvti.RegClass:$rs2)),
             (!cast<Instruction>("PseudoVFMADD_VV_"# suffix)
                  fvti.RegClass:$rd, fvti.RegClass:$rs1, fvti.RegClass:$rs2,
                  fvti.AVL, fvti.Log2SEW, TAIL_AGNOSTIC)>;
-  def : Pat<(fvti.Vector (fma fvti.RegClass:$rs1, fvti.RegClass:$rd,
+  def : Pat<(fvti.Vector (any_fma fvti.RegClass:$rs1, fvti.RegClass:$rd,
                                   (fneg fvti.RegClass:$rs2))),
             (!cast<Instruction>("PseudoVFMSUB_VV_"# suffix)
                  fvti.RegClass:$rd, fvti.RegClass:$rs1, fvti.RegClass:$rs2,
                  fvti.AVL, fvti.Log2SEW, TAIL_AGNOSTIC)>;
-  def : Pat<(fvti.Vector (fma (fneg fvti.RegClass:$rs1), fvti.RegClass:$rd,
+  def : Pat<(fvti.Vector (any_fma (fneg fvti.RegClass:$rs1), fvti.RegClass:$rd,
                                   (fneg fvti.RegClass:$rs2))),
             (!cast<Instruction>("PseudoVFNMADD_VV_"# suffix)
                  fvti.RegClass:$rd, fvti.RegClass:$rs1, fvti.RegClass:$rs2,
                  fvti.AVL, fvti.Log2SEW, TAIL_AGNOSTIC)>;
-  def : Pat<(fvti.Vector (fma (fneg fvti.RegClass:$rs1), fvti.RegClass:$rd,
+  def : Pat<(fvti.Vector (any_fma (fneg fvti.RegClass:$rs1), fvti.RegClass:$rd,
                                   fvti.RegClass:$rs2)),
             (!cast<Instruction>("PseudoVFNMSUB_VV_"# suffix)
                  fvti.RegClass:$rd, fvti.RegClass:$rs1, fvti.RegClass:$rs2,
                  fvti.AVL, fvti.Log2SEW, TAIL_AGNOSTIC)>;
 
   // The choice of VFMADD here is arbitrary, vfmadd.vf and vfmacc.vf are equally
   // commutable.
-  def : Pat<(fvti.Vector (fma (SplatFPOp fvti.ScalarRegClass:$rs1),
+  def : Pat<(fvti.Vector (any_fma (SplatFPOp fvti.ScalarRegClass:$rs1),
                                   fvti.RegClass:$rd, fvti.RegClass:$rs2)),
             (!cast<Instruction>("PseudoVFMADD_V" # fvti.ScalarSuffix # "_" # suffix)
                  fvti.RegClass:$rd, fvti.ScalarRegClass:$rs1, fvti.RegClass:$rs2,
                  fvti.AVL, fvti.Log2SEW, TAIL_AGNOSTIC)>;
-  def : Pat<(fvti.Vector (fma (SplatFPOp fvti.ScalarRegClass:$rs1),
+  def : Pat<(fvti.Vector (any_fma (SplatFPOp fvti.ScalarRegClass:$rs1),
                                   fvti.RegClass:$rd, (fneg fvti.RegClass:$rs2))),
             (!cast<Instruction>("PseudoVFMSUB_V" # fvti.ScalarSuffix # "_" # suffix)
                  fvti.RegClass:$rd, fvti.ScalarRegClass:$rs1, fvti.RegClass:$rs2,
                  fvti.AVL, fvti.Log2SEW, TAIL_AGNOSTIC)>;
 
-  def : Pat<(fvti.Vector (fma (SplatFPOp fvti.ScalarRegClass:$rs1),
+  def : Pat<(fvti.Vector (any_fma (SplatFPOp fvti.ScalarRegClass:$rs1),
                                   (fneg fvti.RegClass:$rd), (fneg fvti.RegClass:$rs2))),
             (!cast<Instruction>("PseudoVFNMADD_V" # fvti.ScalarSuffix # "_" # suffix)
                  fvti.RegClass:$rd, fvti.ScalarRegClass:$rs1, fvti.RegClass:$rs2,
                  fvti.AVL, fvti.Log2SEW, TAIL_AGNOSTIC)>;
-  def : Pat<(fvti.Vector (fma (SplatFPOp fvti.ScalarRegClass:$rs1),
+  def : Pat<(fvti.Vector (any_fma (SplatFPOp fvti.ScalarRegClass:$rs1),
                                   (fneg fvti.RegClass:$rd), fvti.RegClass:$rs2)),
             (!cast<Instruction>("PseudoVFNMSUB_V" # fvti.ScalarSuffix # "_" # suffix)
                  fvti.RegClass:$rd, fvti.ScalarRegClass:$rs1, fvti.RegClass:$rs2,
                  fvti.AVL, fvti.Log2SEW, TAIL_AGNOSTIC)>;
 
   // The splat might be negated.
-  def : Pat<(fvti.Vector (fma (fneg (SplatFPOp fvti.ScalarRegClass:$rs1)),
+  def : Pat<(fvti.Vector (any_fma (fneg (SplatFPOp fvti.ScalarRegClass:$rs1)),
                                   fvti.RegClass:$rd, (fneg fvti.RegClass:$rs2))),
             (!cast<Instruction>("PseudoVFNMADD_V" # fvti.ScalarSuffix # "_" # suffix)
                  fvti.RegClass:$rd, fvti.ScalarRegClass:$rs1, fvti.RegClass:$rs2,
                  fvti.AVL, fvti.Log2SEW, TAIL_AGNOSTIC)>;
-  def : Pat<(fvti.Vector (fma (fneg (SplatFPOp fvti.ScalarRegClass:$rs1)),
+  def : Pat<(fvti.Vector (any_fma (fneg (SplatFPOp fvti.ScalarRegClass:$rs1)),
                                   fvti.RegClass:$rd, fvti.RegClass:$rs2)),
             (!cast<Instruction>("PseudoVFNMSUB_V" # fvti.ScalarSuffix # "_" # suffix)
                  fvti.RegClass:$rd, fvti.ScalarRegClass:$rs1, fvti.RegClass:$rs2,
                  fvti.AVL, fvti.Log2SEW, TAIL_AGNOSTIC)>;
 }
 
 // 13.7. Vector Widening Floating-Point Fused Multiply-Add Instructions
 defm : VPatWidenFPMulAccSDNode_VV_VF<"PseudoVFWMACC">;
 defm : VPatWidenFPNegMulAccSDNode_VV_VF<"PseudoVFWNMACC">;

llvm/test/CodeGen/RISCV/rvv/vfadd-constrained-sdnode.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+experimental-zvfh,+v -target-abi=ilp32d \
+; RUN:     -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+experimental-zvfh,+v -target-abi=lp64d \
+; RUN:     -verify-machineinstrs < %s | FileCheck %s
+
+declare <vscale x 1 x half> @llvm.experimental.constrained.fadd.nxv1f16(<vscale x 1 x half>, <vscale x 1 x half>, metadata, metadata)
+define <vscale x 1 x half> @vfadd_vv_nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %vb) {
+; CHECK-LABEL: vfadd_vv_nxv1f16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vfadd.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 1 x half> @llvm.experimental.constrained.fadd.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x half> %vc
+}
+
+define <vscale x 1 x half> @vfadd_vf_nxv1f16(<vscale x 1 x half> %va, half %b) {
+; CHECK-LABEL: vfadd_vf_nxv1f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vfadd.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x half> poison, half %b, i32 0
+  %splat = shufflevector <vscale x 1 x half> %head, <vscale x 1 x half> poison, <vscale x 1 x i32> zeroinitializer
+  %vc = call <vscale x 1 x half> @llvm.experimental.constrained.fadd.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x half> %vc
+}
+
+declare <vscale x 2 x half> @llvm.experimental.constrained.fadd.nxv2f16(<vscale x 2 x half>, <vscale x 2 x half>, metadata, metadata)
+define <vscale x 2 x half> @vfadd_vv_nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %vb) {
+; CHECK-LABEL: vfadd_vv_nxv2f16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vfadd.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 2 x half> @llvm.experimental.constrained.fadd.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x half> %vc
+}
+
+define <vscale x 2 x half> @vfadd_vf_nxv2f16(<vscale x 2 x half> %va, half %b) {
+; CHECK-LABEL: vfadd_vf_nxv2f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vfadd.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x half> poison, half %b, i32 0
+  %splat = shufflevector <vscale x 2 x half> %head, <vscale x 2 x half> poison, <vscale x 2 x i32> zeroinitializer
+  %vc = call <vscale x 2 x half> @llvm.experimental.constrained.fadd.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x half> %vc
+}
+
+declare <vscale x 4 x half> @llvm.experimental.constrained.fadd.nxv4f16(<vscale x 4 x half>, <vscale x 4 x half>, metadata, metadata)
+define <vscale x 4 x half> @vfadd_vv_nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %vb) {
+; CHECK-LABEL: vfadd_vv_nxv4f16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vfadd.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 4 x half> @llvm.experimental.constrained.fadd.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x half> %vc
+}
+
+define <vscale x 4 x half> @vfadd_vf_nxv4f16(<vscale x 4 x half> %va, half %b) {
+; CHECK-LABEL: vfadd_vf_nxv4f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vfadd.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x half> poison, half %b, i32 0
+  %splat = shufflevector <vscale x 4 x half> %head, <vscale x 4 x half> poison, <vscale x 4 x i32> zeroinitializer
+  %vc = call <vscale x 4 x half> @llvm.experimental.constrained.fadd.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x half> %vc
+}
+
+declare <vscale x 8 x half> @llvm.experimental.constrained.fadd.nxv8f16(<vscale x 8 x half>, <vscale x 8 x half>, metadata, metadata)
+define <vscale x 8 x half> @vfadd_vv_nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %vb) {
+; CHECK-LABEL: vfadd_vv_nxv8f16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e16, m2, ta, ma
+; CHECK-NEXT:    vfadd.vv v8, v8, v10
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 8 x half> @llvm.experimental.constrained.fadd.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x half> %vc
+}
+
+define <vscale x 8 x half> @vfadd_vf_nxv8f16(<vscale x 8 x half> %va, half %b) {
+; CHECK-LABEL: vfadd_vf_nxv8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m2, ta, ma
+; CHECK-NEXT:    vfadd.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x half> poison, half %b, i32 0
+  %splat = shufflevector <vscale x 8 x half> %head, <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
+  %vc = call <vscale x 8 x half> @llvm.experimental.constrained.fadd.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x half> %vc
+}
+
+declare <vscale x 16 x half> @llvm.experimental.constrained.fadd.nxv16f16(<vscale x 16 x half>, <vscale x 16 x half>, metadata, metadata)
+define <vscale x 16 x half> @vfadd_vv_nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %vb) {
+; CHECK-LABEL: vfadd_vv_nxv16f16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e16, m4, ta, ma
+; CHECK-NEXT:    vfadd.vv v8, v8, v12
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 16 x half> @llvm.experimental.constrained.fadd.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x half> %vc
+}
+
+define <vscale x 16 x half> @vfadd_vf_nxv16f16(<vscale x 16 x half> %va, half %b) {
+; CHECK-LABEL: vfadd_vf_nxv16f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m4, ta, ma
+; CHECK-NEXT:    vfadd.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 16 x half> poison, half %b, i32 0
+  %splat = shufflevector <vscale x 16 x half> %head, <vscale x 16 x half> poison, <vscale x 16 x i32> zeroinitializer
+  %vc = call <vscale x 16 x half> @llvm.experimental.constrained.fadd.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x half> %vc
+}
+
+declare <vscale x 32 x half> @llvm.experimental.constrained.fadd.nxv32f16(<vscale x 32 x half>, <vscale x 32 x half>, metadata, metadata)
+define <vscale x 32 x half> @vfadd_vv_nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %vb) {
+; CHECK-LABEL: vfadd_vv_nxv32f16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e16, m8, ta, ma
+; CHECK-NEXT:    vfadd.vv v8, v8, v16
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 32 x half> @llvm.experimental.constrained.fadd.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 32 x half> %vc
+}
+
+define <vscale x 32 x half> @vfadd_vf_nxv32f16(<vscale x 32 x half> %va, half %b) {
+; CHECK-LABEL: vfadd_vf_nxv32f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m8, ta, ma
+; CHECK-NEXT:    vfadd.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 32 x half> poison, half %b, i32 0
+  %splat = shufflevector <vscale x 32 x half> %head, <vscale x 32 x half> poison, <vscale x 32 x i32> zeroinitializer
+  %vc = call <vscale x 32 x half> @llvm.experimental.constrained.fadd.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 32 x half> %vc
+}
+
+declare <vscale x 1 x float> @llvm.experimental.constrained.fadd.nxv1f32(<vscale x 1 x float>, <vscale x 1 x float>, metadata, metadata)
+define <vscale x 1 x float> @vfadd_vv_nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %vb) {
+; CHECK-LABEL: vfadd_vv_nxv1f32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfadd.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 1 x float> @llvm.experimental.constrained.fadd.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x float> %vc
+}
+
+define <vscale x 1 x float> @vfadd_vf_nxv1f32(<vscale x 1 x float> %va, float %b) {
+; CHECK-LABEL: vfadd_vf_nxv1f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfadd.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x float> poison, float %b, i32 0
+  %splat = shufflevector <vscale x 1 x float> %head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
+  %vc = call <vscale x 1 x float> @llvm.experimental.constrained.fadd.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x float> %vc
+}
+
+declare <vscale x 2 x float> @llvm.experimental.constrained.fadd.nxv2f32(<vscale x 2 x float>, <vscale x 2 x float>, metadata, metadata)
+define <vscale x 2 x float> @vfadd_vv_nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %vb) {
+; CHECK-LABEL: vfadd_vv_nxv2f32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vfadd.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 2 x float> @llvm.experimental.constrained.fadd.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x float> %vc
+}
+
+define <vscale x 2 x float> @vfadd_vf_nxv2f32(<vscale x 2 x float> %va, float %b) {
+; CHECK-LABEL: vfadd_vf_nxv2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vfadd.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x float> poison, float %b, i32 0
+  %splat = shufflevector <vscale x 2 x float> %head, <vscale x 2 x float> poison, <vscale x 2 x i32> zeroinitializer
+  %vc = call <vscale x 2 x float> @llvm.experimental.constrained.fadd.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x float> %vc
+}
+
+declare <vscale x 4 x float> @llvm.experimental.constrained.fadd.nxv4f32(<vscale x 4 x float>, <vscale x 4 x float>, metadata, metadata)
+define <vscale x 4 x float> @vfadd_vv_nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %vb) {
+; CHECK-LABEL: vfadd_vv_nxv4f32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
+; CHECK-NEXT:    vfadd.vv v8, v8, v10
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 4 x float> @llvm.experimental.constrained.fadd.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x float> %vc
+}
+
+define <vscale x 4 x float> @vfadd_vf_nxv4f32(<vscale x 4 x float> %va, float %b) {
+; CHECK-LABEL: vfadd_vf_nxv4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
+; CHECK-NEXT:    vfadd.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x float> poison, float %b, i32 0
+  %splat = shufflevector <vscale x 4 x float> %head, <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
+  %vc = call <vscale x 4 x float> @llvm.experimental.constrained.fadd.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x float> %vc
+}
+
+declare <vscale x 8 x float> @llvm.experimental.constrained.fadd.nxv8f32(<vscale x 8 x float>, <vscale x 8 x float>, metadata, metadata)
+define <vscale x 8 x float> @vfadd_vv_nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %vb) {
+; CHECK-LABEL: vfadd_vv_nxv8f32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e32, m4, ta, ma
+; CHECK-NEXT:    vfadd.vv v8, v8, v12
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 8 x float> @llvm.experimental.constrained.fadd.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x float> %vc
+}
+
+define <vscale x 8 x float> @vfadd_vf_nxv8f32(<vscale x 8 x float> %va, float %b) {
+; CHECK-LABEL: vfadd_vf_nxv8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m4, ta, ma
+; CHECK-NEXT:    vfadd.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x float> poison, float %b, i32 0
+  %splat = shufflevector <vscale x 8 x float> %head, <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer
+  %vc = call <vscale x 8 x float> @llvm.experimental.constrained.fadd.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x float> %vc
+}
+
+declare <vscale x 16 x float> @llvm.experimental.constrained.fadd.nxv16f32(<vscale x 16 x float>, <vscale x 16 x float>, metadata, metadata)
+define <vscale x 16 x float> @vfadd_vv_nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %vb) {
+; CHECK-LABEL: vfadd_vv_nxv16f32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e32, m8, ta, ma
+; CHECK-NEXT:    vfadd.vv v8, v8, v16
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 16 x float> @llvm.experimental.constrained.fadd.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x float> %vc
+}
+
+define <vscale x 16 x float> @vfadd_vf_nxv16f32(<vscale x 16 x float> %va, float %b) {
+; CHECK-LABEL: vfadd_vf_nxv16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m8, ta, ma
+; CHECK-NEXT:    vfadd.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 16 x float> poison, float %b, i32 0
+  %splat = shufflevector <vscale x 16 x float> %head, <vscale x 16 x float> poison, <vscale x 16 x i32> zeroinitializer
+  %vc = call <vscale x 16 x float> @llvm.experimental.constrained.fadd.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x float> %vc
+}
+
+declare <vscale x 1 x double> @llvm.experimental.constrained.fadd.nxv1f64(<vscale x 1 x double>, <vscale x 1 x double>, metadata, metadata)
+define <vscale x 1 x double> @vfadd_vv_nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %vb) {
+; CHECK-LABEL: vfadd_vv_nxv1f64:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
+; CHECK-NEXT:    vfadd.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 1 x double> @llvm.experimental.constrained.fadd.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x double> %vc
+}
+
+define <vscale x 1 x double> @vfadd_vf_nxv1f64(<vscale x 1 x double> %va, double %b) {
+; CHECK-LABEL: vfadd_vf_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
+; CHECK-NEXT:    vfadd.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x double> poison, double %b, i32 0
+  %splat = shufflevector <vscale x 1 x double> %head, <vscale x 1 x double> poison, <vscale x 1 x i32> zeroinitializer
+  %vc = call <vscale x 1 x double> @llvm.experimental.constrained.fadd.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x double> %vc
+}
+
+declare <vscale x 2 x double> @llvm.experimental.constrained.fadd.nxv2f64(<vscale x 2 x double>, <vscale x 2 x double>, metadata, metadata)
+define <vscale x 2 x double> @vfadd_vv_nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %vb) {
+; CHECK-LABEL: vfadd_vv_nxv2f64:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e64, m2, ta, ma
+; CHECK-NEXT:    vfadd.vv v8, v8, v10
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 2 x double> @llvm.experimental.constrained.fadd.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x double> %vc
+}
+
+define <vscale x 2 x double> @vfadd_vf_nxv2f64(<vscale x 2 x double> %va, double %b) {
+; CHECK-LABEL: vfadd_vf_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m2, ta, ma
+; CHECK-NEXT:    vfadd.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x double> poison, double %b, i32 0
+  %splat = shufflevector <vscale x 2 x double> %head, <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
+  %vc = call <vscale x 2 x double> @llvm.experimental.constrained.fadd.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x double> %vc
+}
+
+declare <vscale x 4 x double> @llvm.experimental.constrained.fadd.nxv4f64(<vscale x 4 x double>, <vscale x 4 x double>, metadata, metadata)
+define <vscale x 4 x double> @vfadd_vv_nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %vb) {
+; CHECK-LABEL: vfadd_vv_nxv4f64:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e64, m4, ta, ma
+; CHECK-NEXT:    vfadd.vv v8, v8, v12
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 4 x double> @llvm.experimental.constrained.fadd.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x double> %vc
+}
+
+define <vscale x 4 x double> @vfadd_vf_nxv4f64(<vscale x 4 x double> %va, double %b) {
+; CHECK-LABEL: vfadd_vf_nxv4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m4, ta, ma
+; CHECK-NEXT:    vfadd.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x double> poison, double %b, i32 0
+  %splat = shufflevector <vscale x 4 x double> %head, <vscale x 4 x double> poison, <vscale x 4 x i32> zeroinitializer
+  %vc = call <vscale x 4 x double> @llvm.experimental.constrained.fadd.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x double> %vc
+}
+
+declare <vscale x 8 x double> @llvm.experimental.constrained.fadd.nxv8f64(<vscale x 8 x double>, <vscale x 8 x double>, metadata, metadata)
+define <vscale x 8 x double> @vfadd_vv_nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %vb) {
+; CHECK-LABEL: vfadd_vv_nxv8f64:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e64, m8, ta, ma
+; CHECK-NEXT:    vfadd.vv v8, v8, v16
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 8 x double> @llvm.experimental.constrained.fadd.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x double> %vc
+}
+
+define <vscale x 8 x double> @vfadd_vf_nxv8f64(<vscale x 8 x double> %va, double %b) {
+; CHECK-LABEL: vfadd_vf_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m8, ta, ma
+; CHECK-NEXT:    vfadd.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x double> poison, double %b, i32 0
+  %splat = shufflevector <vscale x 8 x double> %head, <vscale x 8 x double> poison, <vscale x 8 x i32> zeroinitializer
+  %vc = call <vscale x 8 x double> @llvm.experimental.constrained.fadd.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x double> %vc
+}

llvm/test/CodeGen/RISCV/rvv/vfdiv-constrained-sdnode.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+experimental-zvfh,+v -target-abi=ilp32d \
+; RUN:     -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+experimental-zvfh,+v -target-abi=lp64d \
+; RUN:     -verify-machineinstrs < %s | FileCheck %s
+
+declare <vscale x 1 x half> @llvm.experimental.constrained.fdiv.nxv1f16(<vscale x 1 x half>, <vscale x 1 x half>, metadata, metadata)
+define <vscale x 1 x half> @vfdiv_vv_nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %vb) {
+; CHECK-LABEL: vfdiv_vv_nxv1f16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vfdiv.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 1 x half> @llvm.experimental.constrained.fdiv.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x half> %vc
+}
+
+define <vscale x 1 x half> @vfdiv_vf_nxv1f16(<vscale x 1 x half> %va, half %b) {
+; CHECK-LABEL: vfdiv_vf_nxv1f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vfdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x half> poison, half %b, i32 0
+  %splat = shufflevector <vscale x 1 x half> %head, <vscale x 1 x half> poison, <vscale x 1 x i32> zeroinitializer
+  %vc = call <vscale x 1 x half> @llvm.experimental.constrained.fdiv.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x half> %vc
+}
+
+declare <vscale x 2 x half> @llvm.experimental.constrained.fdiv.nxv2f16(<vscale x 2 x half>, <vscale x 2 x half>, metadata, metadata)
+define <vscale x 2 x half> @vfdiv_vv_nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %vb) {
+; CHECK-LABEL: vfdiv_vv_nxv2f16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vfdiv.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 2 x half> @llvm.experimental.constrained.fdiv.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x half> %vc
+}
+
+define <vscale x 2 x half> @vfdiv_vf_nxv2f16(<vscale x 2 x half> %va, half %b) {
+; CHECK-LABEL: vfdiv_vf_nxv2f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vfdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x half> poison, half %b, i32 0
+  %splat = shufflevector <vscale x 2 x half> %head, <vscale x 2 x half> poison, <vscale x 2 x i32> zeroinitializer
+  %vc = call <vscale x 2 x half> @llvm.experimental.constrained.fdiv.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x half> %vc
+}
+
+declare <vscale x 4 x half> @llvm.experimental.constrained.fdiv.nxv4f16(<vscale x 4 x half>, <vscale x 4 x half>, metadata, metadata)
+define <vscale x 4 x half> @vfdiv_vv_nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %vb) {
+; CHECK-LABEL: vfdiv_vv_nxv4f16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vfdiv.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 4 x half> @llvm.experimental.constrained.fdiv.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x half> %vc
+}
+
+define <vscale x 4 x half> @vfdiv_vf_nxv4f16(<vscale x 4 x half> %va, half %b) {
+; CHECK-LABEL: vfdiv_vf_nxv4f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vfdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x half> poison, half %b, i32 0
+  %splat = shufflevector <vscale x 4 x half> %head, <vscale x 4 x half> poison, <vscale x 4 x i32> zeroinitializer
+  %vc = call <vscale x 4 x half> @llvm.experimental.constrained.fdiv.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x half> %vc
+}
+
+declare <vscale x 8 x half> @llvm.experimental.constrained.fdiv.nxv8f16(<vscale x 8 x half>, <vscale x 8 x half>, metadata, metadata)
+define <vscale x 8 x half> @vfdiv_vv_nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %vb) {
+; CHECK-LABEL: vfdiv_vv_nxv8f16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e16, m2, ta, ma
+; CHECK-NEXT:    vfdiv.vv v8, v8, v10
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 8 x half> @llvm.experimental.constrained.fdiv.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x half> %vc
+}
+
+define <vscale x 8 x half> @vfdiv_vf_nxv8f16(<vscale x 8 x half> %va, half %b) {
+; CHECK-LABEL: vfdiv_vf_nxv8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m2, ta, ma
+; CHECK-NEXT:    vfdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x half> poison, half %b, i32 0
+  %splat = shufflevector <vscale x 8 x half> %head, <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
+  %vc = call <vscale x 8 x half> @llvm.experimental.constrained.fdiv.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x half> %vc
+}
+
+define <vscale x 8 x half> @vfdiv_fv_nxv8f16(<vscale x 8 x half> %va, half %b) {
+; CHECK-LABEL: vfdiv_fv_nxv8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m2, ta, ma
+; CHECK-NEXT:    vfrdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x half> poison, half %b, i32 0
+  %splat = shufflevector <vscale x 8 x half> %head, <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
+  %vc = call <vscale x 8 x half> @llvm.experimental.constrained.fdiv.nxv8f16(<vscale x 8 x half> %splat, <vscale x 8 x half> %va, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x half> %vc
+}
+
+declare <vscale x 16 x half> @llvm.experimental.constrained.fdiv.nxv16f16(<vscale x 16 x half>, <vscale x 16 x half>, metadata, metadata)
+define <vscale x 16 x half> @vfdiv_vv_nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %vb) {
+; CHECK-LABEL: vfdiv_vv_nxv16f16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e16, m4, ta, ma
+; CHECK-NEXT:    vfdiv.vv v8, v8, v12
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 16 x half> @llvm.experimental.constrained.fdiv.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x half> %vc
+}
+
+define <vscale x 16 x half> @vfdiv_vf_nxv16f16(<vscale x 16 x half> %va, half %b) {
+; CHECK-LABEL: vfdiv_vf_nxv16f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m4, ta, ma
+; CHECK-NEXT:    vfdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 16 x half> poison, half %b, i32 0
+  %splat = shufflevector <vscale x 16 x half> %head, <vscale x 16 x half> poison, <vscale x 16 x i32> zeroinitializer
+  %vc = call <vscale x 16 x half> @llvm.experimental.constrained.fdiv.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x half> %vc
+}
+
+declare <vscale x 32 x half> @llvm.experimental.constrained.fdiv.nxv32f16(<vscale x 32 x half>, <vscale x 32 x half>, metadata, metadata)
+define <vscale x 32 x half> @vfdiv_vv_nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %vb) {
+; CHECK-LABEL: vfdiv_vv_nxv32f16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e16, m8, ta, ma
+; CHECK-NEXT:    vfdiv.vv v8, v8, v16
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 32 x half> @llvm.experimental.constrained.fdiv.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 32 x half> %vc
+}
+
+define <vscale x 32 x half> @vfdiv_vf_nxv32f16(<vscale x 32 x half> %va, half %b) {
+; CHECK-LABEL: vfdiv_vf_nxv32f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m8, ta, ma
+; CHECK-NEXT:    vfdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 32 x half> poison, half %b, i32 0
+  %splat = shufflevector <vscale x 32 x half> %head, <vscale x 32 x half> poison, <vscale x 32 x i32> zeroinitializer
+  %vc = call <vscale x 32 x half> @llvm.experimental.constrained.fdiv.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 32 x half> %vc
+}
+
+declare <vscale x 1 x float> @llvm.experimental.constrained.fdiv.nxv1f32(<vscale x 1 x float>, <vscale x 1 x float>, metadata, metadata)
+define <vscale x 1 x float> @vfdiv_vv_nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %vb) {
+; CHECK-LABEL: vfdiv_vv_nxv1f32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfdiv.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 1 x float> @llvm.experimental.constrained.fdiv.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x float> %vc
+}
+
+define <vscale x 1 x float> @vfdiv_vf_nxv1f32(<vscale x 1 x float> %va, float %b) {
+; CHECK-LABEL: vfdiv_vf_nxv1f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x float> poison, float %b, i32 0
+  %splat = shufflevector <vscale x 1 x float> %head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
+  %vc = call <vscale x 1 x float> @llvm.experimental.constrained.fdiv.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x float> %vc
+}
+
+declare <vscale x 2 x float> @llvm.experimental.constrained.fdiv.nxv2f32(<vscale x 2 x float>, <vscale x 2 x float>, metadata, metadata)
+define <vscale x 2 x float> @vfdiv_vv_nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %vb) {
+; CHECK-LABEL: vfdiv_vv_nxv2f32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vfdiv.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 2 x float> @llvm.experimental.constrained.fdiv.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x float> %vc
+}
+
+define <vscale x 2 x float> @vfdiv_vf_nxv2f32(<vscale x 2 x float> %va, float %b) {
+; CHECK-LABEL: vfdiv_vf_nxv2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vfdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x float> poison, float %b, i32 0
+  %splat = shufflevector <vscale x 2 x float> %head, <vscale x 2 x float> poison, <vscale x 2 x i32> zeroinitializer
+  %vc = call <vscale x 2 x float> @llvm.experimental.constrained.fdiv.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x float> %vc
+}
+
+declare <vscale x 4 x float> @llvm.experimental.constrained.fdiv.nxv4f32(<vscale x 4 x float>, <vscale x 4 x float>, metadata, metadata)
+define <vscale x 4 x float> @vfdiv_vv_nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %vb) {
+; CHECK-LABEL: vfdiv_vv_nxv4f32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
+; CHECK-NEXT:    vfdiv.vv v8, v8, v10
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 4 x float> @llvm.experimental.constrained.fdiv.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x float> %vc
+}
+
+define <vscale x 4 x float> @vfdiv_vf_nxv4f32(<vscale x 4 x float> %va, float %b) {
+; CHECK-LABEL: vfdiv_vf_nxv4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
+; CHECK-NEXT:    vfdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x float> poison, float %b, i32 0
+  %splat = shufflevector <vscale x 4 x float> %head, <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
+  %vc = call <vscale x 4 x float> @llvm.experimental.constrained.fdiv.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x float> %vc
+}
+
+declare <vscale x 8 x float> @llvm.experimental.constrained.fdiv.nxv8f32(<vscale x 8 x float>, <vscale x 8 x float>, metadata, metadata)
+define <vscale x 8 x float> @vfdiv_vv_nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %vb) {
+; CHECK-LABEL: vfdiv_vv_nxv8f32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e32, m4, ta, ma
+; CHECK-NEXT:    vfdiv.vv v8, v8, v12
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 8 x float> @llvm.experimental.constrained.fdiv.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x float> %vc
+}
+
+define <vscale x 8 x float> @vfdiv_vf_nxv8f32(<vscale x 8 x float> %va, float %b) {
+; CHECK-LABEL: vfdiv_vf_nxv8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m4, ta, ma
+; CHECK-NEXT:    vfdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x float> poison, float %b, i32 0
+  %splat = shufflevector <vscale x 8 x float> %head, <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer
+  %vc = call <vscale x 8 x float> @llvm.experimental.constrained.fdiv.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x float> %vc
+}
+
+define <vscale x 8 x float> @vfdiv_fv_nxv8f32(<vscale x 8 x float> %va, float %b) {
+; CHECK-LABEL: vfdiv_fv_nxv8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m4, ta, ma
+; CHECK-NEXT:    vfrdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x float> poison, float %b, i32 0
+  %splat = shufflevector <vscale x 8 x float> %head, <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer
+  %vc = call <vscale x 8 x float> @llvm.experimental.constrained.fdiv.nxv8f32(<vscale x 8 x float> %splat, <vscale x 8 x float> %va, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x float> %vc
+}
+
+declare <vscale x 16 x float> @llvm.experimental.constrained.fdiv.nxv16f32(<vscale x 16 x float>, <vscale x 16 x float>, metadata, metadata)
+define <vscale x 16 x float> @vfdiv_vv_nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %vb) {
+; CHECK-LABEL: vfdiv_vv_nxv16f32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e32, m8, ta, ma
+; CHECK-NEXT:    vfdiv.vv v8, v8, v16
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 16 x float> @llvm.experimental.constrained.fdiv.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x float> %vc
+}
+
+define <vscale x 16 x float> @vfdiv_vf_nxv16f32(<vscale x 16 x float> %va, float %b) {
+; CHECK-LABEL: vfdiv_vf_nxv16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m8, ta, ma
+; CHECK-NEXT:    vfdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 16 x float> poison, float %b, i32 0
+  %splat = shufflevector <vscale x 16 x float> %head, <vscale x 16 x float> poison, <vscale x 16 x i32> zeroinitializer
+  %vc = call <vscale x 16 x float> @llvm.experimental.constrained.fdiv.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x float> %vc
+}
+
+declare <vscale x 1 x double> @llvm.experimental.constrained.fdiv.nxv1f64(<vscale x 1 x double>, <vscale x 1 x double>, metadata, metadata)
+define <vscale x 1 x double> @vfdiv_vv_nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %vb) {
+; CHECK-LABEL: vfdiv_vv_nxv1f64:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
+; CHECK-NEXT:    vfdiv.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 1 x double> @llvm.experimental.constrained.fdiv.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x double> %vc
+}
+
+define <vscale x 1 x double> @vfdiv_vf_nxv1f64(<vscale x 1 x double> %va, double %b) {
+; CHECK-LABEL: vfdiv_vf_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
+; CHECK-NEXT:    vfdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x double> poison, double %b, i32 0
+  %splat = shufflevector <vscale x 1 x double> %head, <vscale x 1 x double> poison, <vscale x 1 x i32> zeroinitializer
+  %vc = call <vscale x 1 x double> @llvm.experimental.constrained.fdiv.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x double> %vc
+}
+
+declare <vscale x 2 x double> @llvm.experimental.constrained.fdiv.nxv2f64(<vscale x 2 x double>, <vscale x 2 x double>, metadata, metadata)
+define <vscale x 2 x double> @vfdiv_vv_nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %vb) {
+; CHECK-LABEL: vfdiv_vv_nxv2f64:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e64, m2, ta, ma
+; CHECK-NEXT:    vfdiv.vv v8, v8, v10
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 2 x double> @llvm.experimental.constrained.fdiv.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x double> %vc
+}
+
+define <vscale x 2 x double> @vfdiv_vf_nxv2f64(<vscale x 2 x double> %va, double %b) {
+; CHECK-LABEL: vfdiv_vf_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m2, ta, ma
+; CHECK-NEXT:    vfdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x double> poison, double %b, i32 0
+  %splat = shufflevector <vscale x 2 x double> %head, <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
+  %vc = call <vscale x 2 x double> @llvm.experimental.constrained.fdiv.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x double> %vc
+}
+
+declare <vscale x 4 x double> @llvm.experimental.constrained.fdiv.nxv4f64(<vscale x 4 x double>, <vscale x 4 x double>, metadata, metadata)
+define <vscale x 4 x double> @vfdiv_vv_nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %vb) {
+; CHECK-LABEL: vfdiv_vv_nxv4f64:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e64, m4, ta, ma
+; CHECK-NEXT:    vfdiv.vv v8, v8, v12
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 4 x double> @llvm.experimental.constrained.fdiv.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x double> %vc
+}
+
+define <vscale x 4 x double> @vfdiv_vf_nxv4f64(<vscale x 4 x double> %va, double %b) {
+; CHECK-LABEL: vfdiv_vf_nxv4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m4, ta, ma
+; CHECK-NEXT:    vfdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x double> poison, double %b, i32 0
+  %splat = shufflevector <vscale x 4 x double> %head, <vscale x 4 x double> poison, <vscale x 4 x i32> zeroinitializer
+  %vc = call <vscale x 4 x double> @llvm.experimental.constrained.fdiv.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x double> %vc
+}
+
+declare <vscale x 8 x double> @llvm.experimental.constrained.fdiv.nxv8f64(<vscale x 8 x double>, <vscale x 8 x double>, metadata, metadata)
+define <vscale x 8 x double> @vfdiv_vv_nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %vb) {
+; CHECK-LABEL: vfdiv_vv_nxv8f64:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e64, m8, ta, ma
+; CHECK-NEXT:    vfdiv.vv v8, v8, v16
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 8 x double> @llvm.experimental.constrained.fdiv.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x double> %vc
+}
+
+define <vscale x 8 x double> @vfdiv_vf_nxv8f64(<vscale x 8 x double> %va, double %b) {
+; CHECK-LABEL: vfdiv_vf_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m8, ta, ma
+; CHECK-NEXT:    vfdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x double> poison, double %b, i32 0
+  %splat = shufflevector <vscale x 8 x double> %head, <vscale x 8 x double> poison, <vscale x 8 x i32> zeroinitializer
+  %vc = call <vscale x 8 x double> @llvm.experimental.constrained.fdiv.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x double> %vc
+}
+
+define <vscale x 8 x double> @vfdiv_fv_nxv8f64(<vscale x 8 x double> %va, double %b) {
+; CHECK-LABEL: vfdiv_fv_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m8, ta, ma
+; CHECK-NEXT:    vfrdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x double> poison, double %b, i32 0
+  %splat = shufflevector <vscale x 8 x double> %head, <vscale x 8 x double> poison, <vscale x 8 x i32> zeroinitializer
+  %vc = call <vscale x 8 x double> @llvm.experimental.constrained.fdiv.nxv8f64(<vscale x 8 x double> %splat, <vscale x 8 x double> %va, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x double> %vc
+}

llvm/test/CodeGen/RISCV/rvv/vfmadd-constrained-sdnode.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+experimental-zvfh,+v -target-abi=ilp32d \
+; RUN:     -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+experimental-zvfh,+v -target-abi=lp64d \
+; RUN:     -verify-machineinstrs < %s | FileCheck %s
+
+; This tests a mix of vfmacc and vfmadd by using different operand orders to
+; trigger commuting in TwoAddressInstructionPass.
+
+declare <vscale x 1 x half> @llvm.experimental.constrained.fma.v1f16(<vscale x 1 x half>, <vscale x 1 x half>, <vscale x 1 x half>, metadata, metadata)
+
+define <vscale x 1 x half> @vfmadd_vv_nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %vb, <vscale x 1 x half> %vc) {
+; CHECK-LABEL: vfmadd_vv_nxv1f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vfmadd.vv v8, v9, v10
+; CHECK-NEXT:    ret
+  %vd = call <vscale x 1 x half> @llvm.experimental.constrained.fma.v1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %vb, <vscale x 1 x half> %vc,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x half> %vd
+}
+
+define <vscale x 1 x half> @vfmadd_vf_nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %vb, half %c) {
+; CHECK-LABEL: vfmadd_vf_nxv1f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vfmadd.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 1 x half> %head, <vscale x 1 x half> poison, <vscale x 1 x i32> zeroinitializer
+  %vd = call <vscale x 1 x half> @llvm.experimental.constrained.fma.v1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %splat, <vscale x 1 x half> %vb,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x half> %vd
+}
+
+declare <vscale x 2 x half> @llvm.experimental.constrained.fma.v2f16(<vscale x 2 x half>, <vscale x 2 x half>, <vscale x 2 x half>, metadata, metadata)
+
+define <vscale x 2 x half> @vfmadd_vv_nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %vb, <vscale x 2 x half> %vc) {
+; CHECK-LABEL: vfmadd_vv_nxv2f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vfmadd.vv v8, v10, v9
+; CHECK-NEXT:    ret
+  %vd = call <vscale x 2 x half> @llvm.experimental.constrained.fma.v2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %vc, <vscale x 2 x half> %vb,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x half> %vd
+}
+
+define <vscale x 2 x half> @vfmadd_vf_nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %vb, half %c) {
+; CHECK-LABEL: vfmadd_vf_nxv2f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vfmacc.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 2 x half> %head, <vscale x 2 x half> poison, <vscale x 2 x i32> zeroinitializer
+  %vd = call <vscale x 2 x half> @llvm.experimental.constrained.fma.v2f16(<vscale x 2 x half> %vb, <vscale x 2 x half> %splat, <vscale x 2 x half> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x half> %vd
+}
+
+declare <vscale x 4 x half> @llvm.experimental.constrained.fma.v4f16(<vscale x 4 x half>, <vscale x 4 x half>, <vscale x 4 x half>, metadata, metadata)
+
+define <vscale x 4 x half> @vfmadd_vv_nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %vb, <vscale x 4 x half> %vc) {
+; CHECK-LABEL: vfmadd_vv_nxv4f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vfmadd.vv v8, v9, v10
+; CHECK-NEXT:    ret
+  %vd = call <vscale x 4 x half> @llvm.experimental.constrained.fma.v4f16(<vscale x 4 x half> %vb, <vscale x 4 x half> %va, <vscale x 4 x half> %vc,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x half> %vd
+}
+
+define <vscale x 4 x half> @vfmadd_vf_nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %vb, half %c) {
+; CHECK-LABEL: vfmadd_vf_nxv4f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vfmadd.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 4 x half> %head, <vscale x 4 x half> poison, <vscale x 4 x i32> zeroinitializer
+  %vd = call <vscale x 4 x half> @llvm.experimental.constrained.fma.v4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %splat, <vscale x 4 x half> %vb,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x half> %vd
+}
+
+declare <vscale x 8 x half> @llvm.experimental.constrained.fma.v8f16(<vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, metadata, metadata)
+
+define <vscale x 8 x half> @vfmadd_vv_nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %vb, <vscale x 8 x half> %vc) {
+; CHECK-LABEL: vfmadd_vv_nxv8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m2, ta, ma
+; CHECK-NEXT:    vfmacc.vv v8, v12, v10
+; CHECK-NEXT:    ret
+  %vd = call <vscale x 8 x half> @llvm.experimental.constrained.fma.v8f16(<vscale x 8 x half> %vb, <vscale x 8 x half> %vc, <vscale x 8 x half> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x half> %vd
+}
+
+define <vscale x 8 x half> @vfmadd_vf_nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %vb, half %c) {
+; CHECK-LABEL: vfmadd_vf_nxv8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m2, ta, ma
+; CHECK-NEXT:    vfmacc.vf v8, fa0, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 8 x half> %head, <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
+  %vd = call <vscale x 8 x half> @llvm.experimental.constrained.fma.v8f16(<vscale x 8 x half> %vb, <vscale x 8 x half> %splat, <vscale x 8 x half> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x half> %vd
+}
+
+declare <vscale x 16 x half> @llvm.experimental.constrained.fma.v16f16(<vscale x 16 x half>, <vscale x 16 x half>, <vscale x 16 x half>, metadata, metadata)
+
+define <vscale x 16 x half> @vfmadd_vv_nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %vb, <vscale x 16 x half> %vc) {
+; CHECK-LABEL: vfmadd_vv_nxv16f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m4, ta, ma
+; CHECK-NEXT:    vfmadd.vv v8, v16, v12
+; CHECK-NEXT:    ret
+  %vd = call <vscale x 16 x half> @llvm.experimental.constrained.fma.v16f16(<vscale x 16 x half> %vc, <vscale x 16 x half> %va, <vscale x 16 x half> %vb,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x half> %vd
+}
+
+define <vscale x 16 x half> @vfmadd_vf_nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %vb, half %c) {
+; CHECK-LABEL: vfmadd_vf_nxv16f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m4, ta, ma
+; CHECK-NEXT:    vfmadd.vf v8, fa0, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 16 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 16 x half> %head, <vscale x 16 x half> poison, <vscale x 16 x i32> zeroinitializer
+  %vd = call <vscale x 16 x half> @llvm.experimental.constrained.fma.v16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %splat, <vscale x 16 x half> %vb,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x half> %vd
+}
+
+declare <vscale x 32 x half> @llvm.experimental.constrained.fma.v32f16(<vscale x 32 x half>, <vscale x 32 x half>, <vscale x 32 x half>, metadata, metadata)
+
+define <vscale x 32 x half> @vfmadd_vv_nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %vb, <vscale x 32 x half> %vc) {
+; CHECK-LABEL: vfmadd_vv_nxv32f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl8re16.v v24, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e16, m8, ta, ma
+; CHECK-NEXT:    vfmacc.vv v8, v16, v24
+; CHECK-NEXT:    ret
+  %vd = call <vscale x 32 x half> @llvm.experimental.constrained.fma.v32f16(<vscale x 32 x half> %vc, <vscale x 32 x half> %vb, <vscale x 32 x half> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 32 x half> %vd
+}
+
+define <vscale x 32 x half> @vfmadd_vf_nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %vb, half %c) {
+; CHECK-LABEL: vfmadd_vf_nxv32f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m8, ta, ma
+; CHECK-NEXT:    vfmacc.vf v8, fa0, v16
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 32 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 32 x half> %head, <vscale x 32 x half> poison, <vscale x 32 x i32> zeroinitializer
+  %vd = call <vscale x 32 x half> @llvm.experimental.constrained.fma.v32f16(<vscale x 32 x half> %vb, <vscale x 32 x half> %splat, <vscale x 32 x half> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 32 x half> %vd
+}
+
+declare <vscale x 1 x float> @llvm.experimental.constrained.fma.v1f32(<vscale x 1 x float>, <vscale x 1 x float>, <vscale x 1 x float>, metadata, metadata)
+
+define <vscale x 1 x float> @vfmadd_vv_nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %vb, <vscale x 1 x float> %vc) {
+; CHECK-LABEL: vfmadd_vv_nxv1f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfmadd.vv v8, v9, v10
+; CHECK-NEXT:    ret
+  %vd = call <vscale x 1 x float> @llvm.experimental.constrained.fma.v1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %vb, <vscale x 1 x float> %vc,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x float> %vd
+}
+
+define <vscale x 1 x float> @vfmadd_vf_nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %vb, float %c) {
+; CHECK-LABEL: vfmadd_vf_nxv1f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfmadd.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 1 x float> %head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
+  %vd = call <vscale x 1 x float> @llvm.experimental.constrained.fma.v1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %splat, <vscale x 1 x float> %vb,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x float> %vd
+}
+
+declare <vscale x 2 x float> @llvm.experimental.constrained.fma.v2f32(<vscale x 2 x float>, <vscale x 2 x float>, <vscale x 2 x float>, metadata, metadata)
+
+define <vscale x 2 x float> @vfmadd_vv_nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %vb, <vscale x 2 x float> %vc) {
+; CHECK-LABEL: vfmadd_vv_nxv2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vfmadd.vv v8, v10, v9
+; CHECK-NEXT:    ret
+  %vd = call <vscale x 2 x float> @llvm.experimental.constrained.fma.v2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %vc, <vscale x 2 x float> %vb,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x float> %vd
+}
+
+define <vscale x 2 x float> @vfmadd_vf_nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %vb, float %c) {
+; CHECK-LABEL: vfmadd_vf_nxv2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vfmacc.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 2 x float> %head, <vscale x 2 x float> poison, <vscale x 2 x i32> zeroinitializer
+  %vd = call <vscale x 2 x float> @llvm.experimental.constrained.fma.v2f32(<vscale x 2 x float> %vb, <vscale x 2 x float> %splat, <vscale x 2 x float> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x float> %vd
+}
+
+declare <vscale x 4 x float> @llvm.experimental.constrained.fma.v4f32(<vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, metadata, metadata)
+
+define <vscale x 4 x float> @vfmadd_vv_nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %vb, <vscale x 4 x float> %vc) {
+; CHECK-LABEL: vfmadd_vv_nxv4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
+; CHECK-NEXT:    vfmadd.vv v8, v10, v12
+; CHECK-NEXT:    ret
+  %vd = call <vscale x 4 x float> @llvm.experimental.constrained.fma.v4f32(<vscale x 4 x float> %vb, <vscale x 4 x float> %va, <vscale x 4 x float> %vc,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x float> %vd
+}
+
+define <vscale x 4 x float> @vfmadd_vf_nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %vb, float %c) {
+; CHECK-LABEL: vfmadd_vf_nxv4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
+; CHECK-NEXT:    vfmadd.vf v8, fa0, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 4 x float> %head, <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
+  %vd = call <vscale x 4 x float> @llvm.experimental.constrained.fma.v4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %splat, <vscale x 4 x float> %vb,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x float> %vd
+}
+
+declare <vscale x 8 x float> @llvm.experimental.constrained.fma.v8f32(<vscale x 8 x float>, <vscale x 8 x float>, <vscale x 8 x float>, metadata, metadata)
+
+define <vscale x 8 x float> @vfmadd_vv_nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %vb, <vscale x 8 x float> %vc) {
+; CHECK-LABEL: vfmadd_vv_nxv8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m4, ta, ma
+; CHECK-NEXT:    vfmacc.vv v8, v16, v12
+; CHECK-NEXT:    ret
+  %vd = call <vscale x 8 x float> @llvm.experimental.constrained.fma.v8f32(<vscale x 8 x float> %vb, <vscale x 8 x float> %vc, <vscale x 8 x float> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x float> %vd
+}
+
+define <vscale x 8 x float> @vfmadd_vf_nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %vb, float %c) {
+; CHECK-LABEL: vfmadd_vf_nxv8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m4, ta, ma
+; CHECK-NEXT:    vfmacc.vf v8, fa0, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 8 x float> %head, <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer
+  %vd = call <vscale x 8 x float> @llvm.experimental.constrained.fma.v8f32(<vscale x 8 x float> %vb, <vscale x 8 x float> %splat, <vscale x 8 x float> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x float> %vd
+}
+
+declare <vscale x 16 x float> @llvm.experimental.constrained.fma.v16f32(<vscale x 16 x float>, <vscale x 16 x float>, <vscale x 16 x float>, metadata, metadata)
+
+define <vscale x 16 x float> @vfmadd_vv_nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %vb, <vscale x 16 x float> %vc) {
+; CHECK-LABEL: vfmadd_vv_nxv16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl8re32.v v24, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e32, m8, ta, ma
+; CHECK-NEXT:    vfmadd.vv v8, v24, v16
+; CHECK-NEXT:    ret
+  %vd = call <vscale x 16 x float> @llvm.experimental.constrained.fma.v16f32(<vscale x 16 x float> %vc, <vscale x 16 x float> %va, <vscale x 16 x float> %vb,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x float> %vd
+}
+
+define <vscale x 16 x float> @vfmadd_vf_nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %vb, float %c) {
+; CHECK-LABEL: vfmadd_vf_nxv16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m8, ta, ma
+; CHECK-NEXT:    vfmadd.vf v8, fa0, v16
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 16 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 16 x float> %head, <vscale x 16 x float> poison, <vscale x 16 x i32> zeroinitializer
+  %vd = call <vscale x 16 x float> @llvm.experimental.constrained.fma.v16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %splat, <vscale x 16 x float> %vb,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x float> %vd
+}
+
+declare <vscale x 1 x double> @llvm.experimental.constrained.fma.v1f64(<vscale x 1 x double>, <vscale x 1 x double>, <vscale x 1 x double>, metadata, metadata)
+
+define <vscale x 1 x double> @vfmadd_vv_nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %vb, <vscale x 1 x double> %vc) {
+; CHECK-LABEL: vfmadd_vv_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
+; CHECK-NEXT:    vfmadd.vv v8, v9, v10
+; CHECK-NEXT:    ret
+  %vd = call <vscale x 1 x double> @llvm.experimental.constrained.fma.v1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %vb, <vscale x 1 x double> %vc,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x double> %vd
+}
+
+define <vscale x 1 x double> @vfmadd_vf_nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %vb, double %c) {
+; CHECK-LABEL: vfmadd_vf_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
+; CHECK-NEXT:    vfmadd.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x double> poison, double %c, i32 0
+  %splat = shufflevector <vscale x 1 x double> %head, <vscale x 1 x double> poison, <vscale x 1 x i32> zeroinitializer
+  %vd = call <vscale x 1 x double> @llvm.experimental.constrained.fma.v1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %splat, <vscale x 1 x double> %vb,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x double> %vd
+}
+
+declare <vscale x 2 x double> @llvm.experimental.constrained.fma.v2f64(<vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, metadata, metadata)
+
+define <vscale x 2 x double> @vfmadd_vv_nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %vb, <vscale x 2 x double> %vc) {
+; CHECK-LABEL: vfmadd_vv_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m2, ta, ma
+; CHECK-NEXT:    vfmadd.vv v8, v12, v10
+; CHECK-NEXT:    ret
+  %vd = call <vscale x 2 x double> @llvm.experimental.constrained.fma.v2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %vc, <vscale x 2 x double> %vb,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x double> %vd
+}
+
+define <vscale x 2 x double> @vfmadd_vf_nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %vb, double %c) {
+; CHECK-LABEL: vfmadd_vf_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m2, ta, ma
+; CHECK-NEXT:    vfmacc.vf v8, fa0, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x double> poison, double %c, i32 0
+  %splat = shufflevector <vscale x 2 x double> %head, <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
+  %vd = call <vscale x 2 x double> @llvm.experimental.constrained.fma.v2f64(<vscale x 2 x double> %vb, <vscale x 2 x double> %splat, <vscale x 2 x double> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x double> %vd
+}
+
+declare <vscale x 4 x double> @llvm.experimental.constrained.fma.v4f64(<vscale x 4 x double>, <vscale x 4 x double>, <vscale x 4 x double>, metadata, metadata)
+
+define <vscale x 4 x double> @vfmadd_vv_nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %vb, <vscale x 4 x double> %vc) {
+; CHECK-LABEL: vfmadd_vv_nxv4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m4, ta, ma
+; CHECK-NEXT:    vfmadd.vv v8, v12, v16
+; CHECK-NEXT:    ret
+  %vd = call <vscale x 4 x double> @llvm.experimental.constrained.fma.v4f64(<vscale x 4 x double> %vb, <vscale x 4 x double> %va, <vscale x 4 x double> %vc,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x double> %vd
+}
+
+define <vscale x 4 x double> @vfmadd_vf_nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %vb, double %c) {
+; CHECK-LABEL: vfmadd_vf_nxv4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m4, ta, ma
+; CHECK-NEXT:    vfmadd.vf v8, fa0, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x double> poison, double %c, i32 0
+  %splat = shufflevector <vscale x 4 x double> %head, <vscale x 4 x double> poison, <vscale x 4 x i32> zeroinitializer
+  %vd = call <vscale x 4 x double> @llvm.experimental.constrained.fma.v4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %splat, <vscale x 4 x double> %vb,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x double> %vd
+}
+
+declare <vscale x 8 x double> @llvm.experimental.constrained.fma.v8f64(<vscale x 8 x double>, <vscale x 8 x double>, <vscale x 8 x double>, metadata, metadata)
+
+define <vscale x 8 x double> @vfmadd_vv_nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %vb, <vscale x 8 x double> %vc) {
+; CHECK-LABEL: vfmadd_vv_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl8re64.v v24, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e64, m8, ta, ma
+; CHECK-NEXT:    vfmacc.vv v8, v16, v24
+; CHECK-NEXT:    ret
+  %vd = call <vscale x 8 x double> @llvm.experimental.constrained.fma.v8f64(<vscale x 8 x double> %vb, <vscale x 8 x double> %vc, <vscale x 8 x double> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x double> %vd
+}
+
+define <vscale x 8 x double> @vfmadd_vf_nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %vb, double %c) {
+; CHECK-LABEL: vfmadd_vf_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m8, ta, ma
+; CHECK-NEXT:    vfmacc.vf v8, fa0, v16
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x double> poison, double %c, i32 0
+  %splat = shufflevector <vscale x 8 x double> %head, <vscale x 8 x double> poison, <vscale x 8 x i32> zeroinitializer
+  %vd = call <vscale x 8 x double> @llvm.experimental.constrained.fma.v8f64(<vscale x 8 x double> %vb, <vscale x 8 x double> %splat, <vscale x 8 x double> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x double> %vd
+}

llvm/test/CodeGen/RISCV/rvv/vfmsub-constrained-sdnode.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+experimental-zvfh,+v -target-abi=ilp32d \
+; RUN:     -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+experimental-zvfh,+v -target-abi=lp64d \
+; RUN:     -verify-machineinstrs < %s | FileCheck %s
+
+; This tests a mix of vfmsac and vfmsub by using different operand orders to
+; trigger commuting in TwoAddressInstructionPass.
+
+declare <vscale x 1 x half> @llvm.experimental.constrained.fma.v1f16(<vscale x 1 x half>, <vscale x 1 x half>, <vscale x 1 x half>, metadata, metadata)
+
+define <vscale x 1 x half> @vfmsub_vv_nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %vb, <vscale x 1 x half> %vc) {
+; CHECK-LABEL: vfmsub_vv_nxv1f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vfmsub.vv v8, v9, v10
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 1 x half> %vc
+  %vd = call <vscale x 1 x half> @llvm.experimental.constrained.fma.v1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %vb, <vscale x 1 x half> %neg,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x half> %vd
+}
+
+define <vscale x 1 x half> @vfmsub_vf_nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %vb, half %c) {
+; CHECK-LABEL: vfmsub_vf_nxv1f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vfmsub.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 1 x half> %head, <vscale x 1 x half> poison, <vscale x 1 x i32> zeroinitializer
+  %neg = fneg <vscale x 1 x half> %vb
+  %vd = call <vscale x 1 x half> @llvm.experimental.constrained.fma.v1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %splat, <vscale x 1 x half> %neg,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x half> %vd
+}
+
+declare <vscale x 2 x half> @llvm.experimental.constrained.fma.v2f16(<vscale x 2 x half>, <vscale x 2 x half>, <vscale x 2 x half>, metadata, metadata)
+
+define <vscale x 2 x half> @vfmsub_vv_nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %vb, <vscale x 2 x half> %vc) {
+; CHECK-LABEL: vfmsub_vv_nxv2f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vfmsub.vv v8, v10, v9
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 2 x half> %vb
+  %vd = call <vscale x 2 x half> @llvm.experimental.constrained.fma.v2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %vc, <vscale x 2 x half> %neg,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x half> %vd
+}
+
+define <vscale x 2 x half> @vfmsub_vf_nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %vb, half %c) {
+; CHECK-LABEL: vfmsub_vf_nxv2f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vfmsac.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 2 x half> %head, <vscale x 2 x half> poison, <vscale x 2 x i32> zeroinitializer
+  %neg = fneg <vscale x 2 x half> %va
+  %vd = call <vscale x 2 x half> @llvm.experimental.constrained.fma.v2f16(<vscale x 2 x half> %vb, <vscale x 2 x half> %splat, <vscale x 2 x half> %neg,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x half> %vd
+}
+
+declare <vscale x 4 x half> @llvm.experimental.constrained.fma.v4f16(<vscale x 4 x half>, <vscale x 4 x half>, <vscale x 4 x half>, metadata, metadata)
+
+define <vscale x 4 x half> @vfmsub_vv_nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %vb, <vscale x 4 x half> %vc) {
+; CHECK-LABEL: vfmsub_vv_nxv4f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vfmsub.vv v8, v9, v10
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 4 x half> %vc
+  %vd = call <vscale x 4 x half> @llvm.experimental.constrained.fma.v4f16(<vscale x 4 x half> %vb, <vscale x 4 x half> %va, <vscale x 4 x half> %neg,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x half> %vd
+}
+
+define <vscale x 4 x half> @vfmsub_vf_nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %vb, half %c) {
+; CHECK-LABEL: vfmsub_vf_nxv4f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vfmsub.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 4 x half> %head, <vscale x 4 x half> poison, <vscale x 4 x i32> zeroinitializer
+  %neg = fneg <vscale x 4 x half> %vb
+  %vd = call <vscale x 4 x half> @llvm.experimental.constrained.fma.v4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %splat, <vscale x 4 x half> %neg,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x half> %vd
+}
+
+declare <vscale x 8 x half> @llvm.experimental.constrained.fma.v8f16(<vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, metadata, metadata)
+
+define <vscale x 8 x half> @vfmsub_vv_nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %vb, <vscale x 8 x half> %vc) {
+; CHECK-LABEL: vfmsub_vv_nxv8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m2, ta, ma
+; CHECK-NEXT:    vfmsac.vv v8, v12, v10
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 8 x half> %va
+  %vd = call <vscale x 8 x half> @llvm.experimental.constrained.fma.v8f16(<vscale x 8 x half> %vb, <vscale x 8 x half> %vc, <vscale x 8 x half> %neg,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x half> %vd
+}
+
+define <vscale x 8 x half> @vfmsub_vf_nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %vb, half %c) {
+; CHECK-LABEL: vfmsub_vf_nxv8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m2, ta, ma
+; CHECK-NEXT:    vfmsac.vf v8, fa0, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 8 x half> %head, <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
+  %neg = fneg <vscale x 8 x half> %va
+  %vd = call <vscale x 8 x half> @llvm.experimental.constrained.fma.v8f16(<vscale x 8 x half> %vb, <vscale x 8 x half> %splat, <vscale x 8 x half> %neg,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x half> %vd
+}
+
+declare <vscale x 16 x half> @llvm.experimental.constrained.fma.v16f16(<vscale x 16 x half>, <vscale x 16 x half>, <vscale x 16 x half>, metadata, metadata)
+
+define <vscale x 16 x half> @vfmsub_vv_nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %vb, <vscale x 16 x half> %vc) {
+; CHECK-LABEL: vfmsub_vv_nxv16f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m4, ta, ma
+; CHECK-NEXT:    vfmsub.vv v8, v16, v12
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 16 x half> %vb
+  %vd = call <vscale x 16 x half> @llvm.experimental.constrained.fma.v16f16(<vscale x 16 x half> %vc, <vscale x 16 x half> %va, <vscale x 16 x half> %neg,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x half> %vd
+}
+
+define <vscale x 16 x half> @vfmsub_vf_nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %vb, half %c) {
+; CHECK-LABEL: vfmsub_vf_nxv16f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m4, ta, ma
+; CHECK-NEXT:    vfmsub.vf v8, fa0, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 16 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 16 x half> %head, <vscale x 16 x half> poison, <vscale x 16 x i32> zeroinitializer
+  %neg = fneg <vscale x 16 x half> %vb
+  %vd = call <vscale x 16 x half> @llvm.experimental.constrained.fma.v16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %splat, <vscale x 16 x half> %neg,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x half> %vd
+}
+
+declare <vscale x 32 x half> @llvm.experimental.constrained.fma.v32f16(<vscale x 32 x half>, <vscale x 32 x half>, <vscale x 32 x half>, metadata, metadata)
+
+define <vscale x 32 x half> @vfmsub_vv_nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %vb, <vscale x 32 x half> %vc) {
+; CHECK-LABEL: vfmsub_vv_nxv32f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl8re16.v v24, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e16, m8, ta, ma
+; CHECK-NEXT:    vfmsac.vv v8, v16, v24
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 32 x half> %va
+  %vd = call <vscale x 32 x half> @llvm.experimental.constrained.fma.v32f16(<vscale x 32 x half> %vc, <vscale x 32 x half> %vb, <vscale x 32 x half> %neg,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 32 x half> %vd
+}
+
+define <vscale x 32 x half> @vfmsub_vf_nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %vb, half %c) {
+; CHECK-LABEL: vfmsub_vf_nxv32f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m8, ta, ma
+; CHECK-NEXT:    vfmsac.vf v8, fa0, v16
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 32 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 32 x half> %head, <vscale x 32 x half> poison, <vscale x 32 x i32> zeroinitializer
+  %neg = fneg <vscale x 32 x half> %va
+  %vd = call <vscale x 32 x half> @llvm.experimental.constrained.fma.v32f16(<vscale x 32 x half> %vb, <vscale x 32 x half> %splat, <vscale x 32 x half> %neg,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 32 x half> %vd
+}
+
+declare <vscale x 1 x float> @llvm.experimental.constrained.fma.v1f32(<vscale x 1 x float>, <vscale x 1 x float>, <vscale x 1 x float>, metadata, metadata)
+
+define <vscale x 1 x float> @vfmsub_vv_nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %vb, <vscale x 1 x float> %vc) {
+; CHECK-LABEL: vfmsub_vv_nxv1f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfmsub.vv v8, v9, v10
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 1 x float> %vc
+  %vd = call <vscale x 1 x float> @llvm.experimental.constrained.fma.v1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %vb, <vscale x 1 x float> %neg,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x float> %vd
+}
+
+define <vscale x 1 x float> @vfmsub_vf_nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %vb, float %c) {
+; CHECK-LABEL: vfmsub_vf_nxv1f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfmsub.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 1 x float> %head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
+  %neg = fneg <vscale x 1 x float> %vb
+  %vd = call <vscale x 1 x float> @llvm.experimental.constrained.fma.v1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %splat, <vscale x 1 x float> %neg,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x float> %vd
+}
+
+declare <vscale x 2 x float> @llvm.experimental.constrained.fma.v2f32(<vscale x 2 x float>, <vscale x 2 x float>, <vscale x 2 x float>, metadata, metadata)
+
+define <vscale x 2 x float> @vfmsub_vv_nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %vb, <vscale x 2 x float> %vc) {
+; CHECK-LABEL: vfmsub_vv_nxv2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vfmsub.vv v8, v10, v9
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 2 x float> %vb
+  %vd = call <vscale x 2 x float> @llvm.experimental.constrained.fma.v2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %vc, <vscale x 2 x float> %neg,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x float> %vd
+}
+
+define <vscale x 2 x float> @vfmsub_vf_nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %vb, float %c) {
+; CHECK-LABEL: vfmsub_vf_nxv2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vfmsac.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 2 x float> %head, <vscale x 2 x float> poison, <vscale x 2 x i32> zeroinitializer
+  %neg = fneg <vscale x 2 x float> %va
+  %vd = call <vscale x 2 x float> @llvm.experimental.constrained.fma.v2f32(<vscale x 2 x float> %vb, <vscale x 2 x float> %splat, <vscale x 2 x float> %neg,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x float> %vd
+}
+
+declare <vscale x 4 x float> @llvm.experimental.constrained.fma.v4f32(<vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, metadata, metadata)
+
+define <vscale x 4 x float> @vfmsub_vv_nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %vb, <vscale x 4 x float> %vc) {
+; CHECK-LABEL: vfmsub_vv_nxv4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
+; CHECK-NEXT:    vfmsub.vv v8, v10, v12
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 4 x float> %vc
+  %vd = call <vscale x 4 x float> @llvm.experimental.constrained.fma.v4f32(<vscale x 4 x float> %vb, <vscale x 4 x float> %va, <vscale x 4 x float> %neg,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x float> %vd
+}
+
+define <vscale x 4 x float> @vfmsub_vf_nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %vb, float %c) {
+; CHECK-LABEL: vfmsub_vf_nxv4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
+; CHECK-NEXT:    vfmsub.vf v8, fa0, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 4 x float> %head, <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
+  %neg = fneg <vscale x 4 x float> %vb
+  %vd = call <vscale x 4 x float> @llvm.experimental.constrained.fma.v4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %splat, <vscale x 4 x float> %neg,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x float> %vd
+}
+
+declare <vscale x 8 x float> @llvm.experimental.constrained.fma.v8f32(<vscale x 8 x float>, <vscale x 8 x float>, <vscale x 8 x float>, metadata, metadata)
+
+define <vscale x 8 x float> @vfmsub_vv_nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %vb, <vscale x 8 x float> %vc) {
+; CHECK-LABEL: vfmsub_vv_nxv8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m4, ta, ma
+; CHECK-NEXT:    vfmsac.vv v8, v16, v12
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 8 x float> %va
+  %vd = call <vscale x 8 x float> @llvm.experimental.constrained.fma.v8f32(<vscale x 8 x float> %vb, <vscale x 8 x float> %vc, <vscale x 8 x float> %neg,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x float> %vd
+}
+
+define <vscale x 8 x float> @vfmsub_vf_nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %vb, float %c) {
+; CHECK-LABEL: vfmsub_vf_nxv8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m4, ta, ma
+; CHECK-NEXT:    vfmsac.vf v8, fa0, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 8 x float> %head, <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer
+  %neg = fneg <vscale x 8 x float> %va
+  %vd = call <vscale x 8 x float> @llvm.experimental.constrained.fma.v8f32(<vscale x 8 x float> %vb, <vscale x 8 x float> %splat, <vscale x 8 x float> %neg,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x float> %vd
+}
+
+declare <vscale x 16 x float> @llvm.experimental.constrained.fma.v16f32(<vscale x 16 x float>, <vscale x 16 x float>, <vscale x 16 x float>, metadata, metadata)
+
+define <vscale x 16 x float> @vfmsub_vv_nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %vb, <vscale x 16 x float> %vc) {
+; CHECK-LABEL: vfmsub_vv_nxv16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl8re32.v v24, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e32, m8, ta, ma
+; CHECK-NEXT:    vfmsub.vv v8, v24, v16
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 16 x float> %vb
+  %vd = call <vscale x 16 x float> @llvm.experimental.constrained.fma.v16f32(<vscale x 16 x float> %vc, <vscale x 16 x float> %va, <vscale x 16 x float> %neg,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x float> %vd
+}
+
+define <vscale x 16 x float> @vfmsub_vf_nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %vb, float %c) {
+; CHECK-LABEL: vfmsub_vf_nxv16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m8, ta, ma
+; CHECK-NEXT:    vfmsub.vf v8, fa0, v16
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 16 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 16 x float> %head, <vscale x 16 x float> poison, <vscale x 16 x i32> zeroinitializer
+  %neg = fneg <vscale x 16 x float> %vb
+  %vd = call <vscale x 16 x float> @llvm.experimental.constrained.fma.v16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %splat, <vscale x 16 x float> %neg,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x float> %vd
+}
+
+declare <vscale x 1 x double> @llvm.experimental.constrained.fma.v1f64(<vscale x 1 x double>, <vscale x 1 x double>, <vscale x 1 x double>, metadata, metadata)
+
+define <vscale x 1 x double> @vfmsub_vv_nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %vb, <vscale x 1 x double> %vc) {
+; CHECK-LABEL: vfmsub_vv_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
+; CHECK-NEXT:    vfmsub.vv v8, v9, v10
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 1 x double> %vc
+  %vd = call <vscale x 1 x double> @llvm.experimental.constrained.fma.v1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %vb, <vscale x 1 x double> %neg,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x double> %vd
+}
+
+define <vscale x 1 x double> @vfmsub_vf_nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %vb, double %c) {
+; CHECK-LABEL: vfmsub_vf_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
+; CHECK-NEXT:    vfmsub.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x double> poison, double %c, i32 0
+  %splat = shufflevector <vscale x 1 x double> %head, <vscale x 1 x double> poison, <vscale x 1 x i32> zeroinitializer
+  %neg = fneg <vscale x 1 x double> %vb
+  %vd = call <vscale x 1 x double> @llvm.experimental.constrained.fma.v1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %splat, <vscale x 1 x double> %neg,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x double> %vd
+}
+
+declare <vscale x 2 x double> @llvm.experimental.constrained.fma.v2f64(<vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, metadata, metadata)
+
+define <vscale x 2 x double> @vfmsub_vv_nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %vb, <vscale x 2 x double> %vc) {
+; CHECK-LABEL: vfmsub_vv_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m2, ta, ma
+; CHECK-NEXT:    vfmsub.vv v8, v12, v10
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 2 x double> %vb
+  %vd = call <vscale x 2 x double> @llvm.experimental.constrained.fma.v2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %vc, <vscale x 2 x double> %neg,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x double> %vd
+}
+
+define <vscale x 2 x double> @vfmsub_vf_nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %vb, double %c) {
+; CHECK-LABEL: vfmsub_vf_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m2, ta, ma
+; CHECK-NEXT:    vfmsac.vf v8, fa0, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x double> poison, double %c, i32 0
+  %splat = shufflevector <vscale x 2 x double> %head, <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
+  %neg = fneg <vscale x 2 x double> %va
+  %vd = call <vscale x 2 x double> @llvm.experimental.constrained.fma.v2f64(<vscale x 2 x double> %vb, <vscale x 2 x double> %splat, <vscale x 2 x double> %neg,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x double> %vd
+}
+
+declare <vscale x 4 x double> @llvm.experimental.constrained.fma.v4f64(<vscale x 4 x double>, <vscale x 4 x double>, <vscale x 4 x double>, metadata, metadata)
+
+define <vscale x 4 x double> @vfmsub_vv_nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %vb, <vscale x 4 x double> %vc) {
+; CHECK-LABEL: vfmsub_vv_nxv4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m4, ta, ma
+; CHECK-NEXT:    vfmsub.vv v8, v12, v16
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 4 x double> %vc
+  %vd = call <vscale x 4 x double> @llvm.experimental.constrained.fma.v4f64(<vscale x 4 x double> %vb, <vscale x 4 x double> %va, <vscale x 4 x double> %neg,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x double> %vd
+}
+
+define <vscale x 4 x double> @vfmsub_vf_nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %vb, double %c) {
+; CHECK-LABEL: vfmsub_vf_nxv4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m4, ta, ma
+; CHECK-NEXT:    vfmsub.vf v8, fa0, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x double> poison, double %c, i32 0
+  %splat = shufflevector <vscale x 4 x double> %head, <vscale x 4 x double> poison, <vscale x 4 x i32> zeroinitializer
+  %neg = fneg <vscale x 4 x double> %vb
+  %vd = call <vscale x 4 x double> @llvm.experimental.constrained.fma.v4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %splat, <vscale x 4 x double> %neg,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x double> %vd
+}
+
+declare <vscale x 8 x double> @llvm.experimental.constrained.fma.v8f64(<vscale x 8 x double>, <vscale x 8 x double>, <vscale x 8 x double>, metadata, metadata)
+
+define <vscale x 8 x double> @vfmsub_vv_nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %vb, <vscale x 8 x double> %vc) {
+; CHECK-LABEL: vfmsub_vv_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl8re64.v v24, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e64, m8, ta, ma
+; CHECK-NEXT:    vfmsac.vv v8, v16, v24
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 8 x double> %va
+  %vd = call <vscale x 8 x double> @llvm.experimental.constrained.fma.v8f64(<vscale x 8 x double> %vb, <vscale x 8 x double> %vc, <vscale x 8 x double> %neg,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x double> %vd
+}
+
+define <vscale x 8 x double> @vfmsub_vf_nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %vb, double %c) {
+; CHECK-LABEL: vfmsub_vf_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m8, ta, ma
+; CHECK-NEXT:    vfmsac.vf v8, fa0, v16
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x double> poison, double %c, i32 0
+  %splat = shufflevector <vscale x 8 x double> %head, <vscale x 8 x double> poison, <vscale x 8 x i32> zeroinitializer
+  %neg = fneg <vscale x 8 x double> %va
+  %vd = call <vscale x 8 x double> @llvm.experimental.constrained.fma.v8f64(<vscale x 8 x double> %vb, <vscale x 8 x double> %splat, <vscale x 8 x double> %neg,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x double> %vd
+}

llvm/test/CodeGen/RISCV/rvv/vfmul-constrained-sdnode.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+experimental-zvfh,+v -target-abi=ilp32d \
+; RUN:     -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+experimental-zvfh,+v -target-abi=lp64d \
+; RUN:     -verify-machineinstrs < %s | FileCheck %s
+
+declare <vscale x 1 x half> @llvm.experimental.constrained.fmul.nxv1f16(<vscale x 1 x half>, <vscale x 1 x half>, metadata, metadata)
+define <vscale x 1 x half> @vfmul_vv_nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %vb) {
+; CHECK-LABEL: vfmul_vv_nxv1f16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vfmul.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 1 x half> @llvm.experimental.constrained.fmul.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x half> %vc
+}
+
+define <vscale x 1 x half> @vfmul_vf_nxv1f16(<vscale x 1 x half> %va, half %b) {
+; CHECK-LABEL: vfmul_vf_nxv1f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vfmul.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x half> poison, half %b, i32 0
+  %splat = shufflevector <vscale x 1 x half> %head, <vscale x 1 x half> poison, <vscale x 1 x i32> zeroinitializer
+  %vc = call <vscale x 1 x half> @llvm.experimental.constrained.fmul.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x half> %vc
+}
+
+declare <vscale x 2 x half> @llvm.experimental.constrained.fmul.nxv2f16(<vscale x 2 x half>, <vscale x 2 x half>, metadata, metadata)
+define <vscale x 2 x half> @vfmul_vv_nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %vb) {
+; CHECK-LABEL: vfmul_vv_nxv2f16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vfmul.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 2 x half> @llvm.experimental.constrained.fmul.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x half> %vc
+}
+
+define <vscale x 2 x half> @vfmul_vf_nxv2f16(<vscale x 2 x half> %va, half %b) {
+; CHECK-LABEL: vfmul_vf_nxv2f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vfmul.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x half> poison, half %b, i32 0
+  %splat = shufflevector <vscale x 2 x half> %head, <vscale x 2 x half> poison, <vscale x 2 x i32> zeroinitializer
+  %vc = call <vscale x 2 x half> @llvm.experimental.constrained.fmul.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x half> %vc
+}
+
+declare <vscale x 4 x half> @llvm.experimental.constrained.fmul.nxv4f16(<vscale x 4 x half>, <vscale x 4 x half>, metadata, metadata)
+define <vscale x 4 x half> @vfmul_vv_nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %vb) {
+; CHECK-LABEL: vfmul_vv_nxv4f16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vfmul.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 4 x half> @llvm.experimental.constrained.fmul.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x half> %vc
+}
+
+define <vscale x 4 x half> @vfmul_vf_nxv4f16(<vscale x 4 x half> %va, half %b) {
+; CHECK-LABEL: vfmul_vf_nxv4f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vfmul.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x half> poison, half %b, i32 0
+  %splat = shufflevector <vscale x 4 x half> %head, <vscale x 4 x half> poison, <vscale x 4 x i32> zeroinitializer
+  %vc = call <vscale x 4 x half> @llvm.experimental.constrained.fmul.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x half> %vc
+}
+
+declare <vscale x 8 x half> @llvm.experimental.constrained.fmul.nxv8f16(<vscale x 8 x half>, <vscale x 8 x half>, metadata, metadata)
+define <vscale x 8 x half> @vfmul_vv_nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %vb) {
+; CHECK-LABEL: vfmul_vv_nxv8f16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e16, m2, ta, ma
+; CHECK-NEXT:    vfmul.vv v8, v8, v10
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 8 x half> @llvm.experimental.constrained.fmul.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x half> %vc
+}
+
+define <vscale x 8 x half> @vfmul_vf_nxv8f16(<vscale x 8 x half> %va, half %b) {
+; CHECK-LABEL: vfmul_vf_nxv8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m2, ta, ma
+; CHECK-NEXT:    vfmul.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x half> poison, half %b, i32 0
+  %splat = shufflevector <vscale x 8 x half> %head, <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
+  %vc = call <vscale x 8 x half> @llvm.experimental.constrained.fmul.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x half> %vc
+}
+
+declare <vscale x 16 x half> @llvm.experimental.constrained.fmul.nxv16f16(<vscale x 16 x half>, <vscale x 16 x half>, metadata, metadata)
+define <vscale x 16 x half> @vfmul_vv_nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %vb) {
+; CHECK-LABEL: vfmul_vv_nxv16f16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e16, m4, ta, ma
+; CHECK-NEXT:    vfmul.vv v8, v8, v12
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 16 x half> @llvm.experimental.constrained.fmul.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x half> %vc
+}
+
+define <vscale x 16 x half> @vfmul_vf_nxv16f16(<vscale x 16 x half> %va, half %b) {
+; CHECK-LABEL: vfmul_vf_nxv16f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m4, ta, ma
+; CHECK-NEXT:    vfmul.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 16 x half> poison, half %b, i32 0
+  %splat = shufflevector <vscale x 16 x half> %head, <vscale x 16 x half> poison, <vscale x 16 x i32> zeroinitializer
+  %vc = call <vscale x 16 x half> @llvm.experimental.constrained.fmul.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x half> %vc
+}
+
+declare <vscale x 32 x half> @llvm.experimental.constrained.fmul.nxv32f16(<vscale x 32 x half>, <vscale x 32 x half>, metadata, metadata)
+define <vscale x 32 x half> @vfmul_vv_nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %vb) {
+; CHECK-LABEL: vfmul_vv_nxv32f16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e16, m8, ta, ma
+; CHECK-NEXT:    vfmul.vv v8, v8, v16
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 32 x half> @llvm.experimental.constrained.fmul.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 32 x half> %vc
+}
+
+define <vscale x 32 x half> @vfmul_vf_nxv32f16(<vscale x 32 x half> %va, half %b) {
+; CHECK-LABEL: vfmul_vf_nxv32f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m8, ta, ma
+; CHECK-NEXT:    vfmul.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 32 x half> poison, half %b, i32 0
+  %splat = shufflevector <vscale x 32 x half> %head, <vscale x 32 x half> poison, <vscale x 32 x i32> zeroinitializer
+  %vc = call <vscale x 32 x half> @llvm.experimental.constrained.fmul.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 32 x half> %vc
+}
+
+declare <vscale x 1 x float> @llvm.experimental.constrained.fmul.nxv1f32(<vscale x 1 x float>, <vscale x 1 x float>, metadata, metadata)
+define <vscale x 1 x float> @vfmul_vv_nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %vb) {
+; CHECK-LABEL: vfmul_vv_nxv1f32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfmul.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 1 x float> @llvm.experimental.constrained.fmul.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x float> %vc
+}
+
+define <vscale x 1 x float> @vfmul_vf_nxv1f32(<vscale x 1 x float> %va, float %b) {
+; CHECK-LABEL: vfmul_vf_nxv1f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfmul.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x float> poison, float %b, i32 0
+  %splat = shufflevector <vscale x 1 x float> %head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
+  %vc = call <vscale x 1 x float> @llvm.experimental.constrained.fmul.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x float> %vc
+}
+
+declare <vscale x 2 x float> @llvm.experimental.constrained.fmul.nxv2f32(<vscale x 2 x float>, <vscale x 2 x float>, metadata, metadata)
+define <vscale x 2 x float> @vfmul_vv_nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %vb) {
+; CHECK-LABEL: vfmul_vv_nxv2f32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vfmul.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 2 x float> @llvm.experimental.constrained.fmul.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x float> %vc
+}
+
+define <vscale x 2 x float> @vfmul_vf_nxv2f32(<vscale x 2 x float> %va, float %b) {
+; CHECK-LABEL: vfmul_vf_nxv2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vfmul.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x float> poison, float %b, i32 0
+  %splat = shufflevector <vscale x 2 x float> %head, <vscale x 2 x float> poison, <vscale x 2 x i32> zeroinitializer
+  %vc = call <vscale x 2 x float> @llvm.experimental.constrained.fmul.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x float> %vc
+}
+
+declare <vscale x 4 x float> @llvm.experimental.constrained.fmul.nxv4f32(<vscale x 4 x float>, <vscale x 4 x float>, metadata, metadata)
+define <vscale x 4 x float> @vfmul_vv_nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %vb) {
+; CHECK-LABEL: vfmul_vv_nxv4f32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
+; CHECK-NEXT:    vfmul.vv v8, v8, v10
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 4 x float> @llvm.experimental.constrained.fmul.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x float> %vc
+}
+
+define <vscale x 4 x float> @vfmul_vf_nxv4f32(<vscale x 4 x float> %va, float %b) {
+; CHECK-LABEL: vfmul_vf_nxv4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
+; CHECK-NEXT:    vfmul.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x float> poison, float %b, i32 0
+  %splat = shufflevector <vscale x 4 x float> %head, <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
+  %vc = call <vscale x 4 x float> @llvm.experimental.constrained.fmul.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x float> %vc
+}
+
+declare <vscale x 8 x float> @llvm.experimental.constrained.fmul.nxv8f32(<vscale x 8 x float>, <vscale x 8 x float>, metadata, metadata)
+define <vscale x 8 x float> @vfmul_vv_nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %vb) {
+; CHECK-LABEL: vfmul_vv_nxv8f32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e32, m4, ta, ma
+; CHECK-NEXT:    vfmul.vv v8, v8, v12
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 8 x float> @llvm.experimental.constrained.fmul.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x float> %vc
+}
+
+define <vscale x 8 x float> @vfmul_vf_nxv8f32(<vscale x 8 x float> %va, float %b) {
+; CHECK-LABEL: vfmul_vf_nxv8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m4, ta, ma
+; CHECK-NEXT:    vfmul.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x float> poison, float %b, i32 0
+  %splat = shufflevector <vscale x 8 x float> %head, <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer
+  %vc = call <vscale x 8 x float> @llvm.experimental.constrained.fmul.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x float> %vc
+}
+
+declare <vscale x 16 x float> @llvm.experimental.constrained.fmul.nxv16f32(<vscale x 16 x float>, <vscale x 16 x float>, metadata, metadata)
+define <vscale x 16 x float> @vfmul_vv_nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %vb) {
+; CHECK-LABEL: vfmul_vv_nxv16f32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e32, m8, ta, ma
+; CHECK-NEXT:    vfmul.vv v8, v8, v16
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 16 x float> @llvm.experimental.constrained.fmul.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x float> %vc
+}
+
+define <vscale x 16 x float> @vfmul_vf_nxv16f32(<vscale x 16 x float> %va, float %b) {
+; CHECK-LABEL: vfmul_vf_nxv16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m8, ta, ma
+; CHECK-NEXT:    vfmul.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 16 x float> poison, float %b, i32 0
+  %splat = shufflevector <vscale x 16 x float> %head, <vscale x 16 x float> poison, <vscale x 16 x i32> zeroinitializer
+  %vc = call <vscale x 16 x float> @llvm.experimental.constrained.fmul.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x float> %vc
+}
+
+declare <vscale x 1 x double> @llvm.experimental.constrained.fmul.nxv1f64(<vscale x 1 x double>, <vscale x 1 x double>, metadata, metadata)
+define <vscale x 1 x double> @vfmul_vv_nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %vb) {
+; CHECK-LABEL: vfmul_vv_nxv1f64:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
+; CHECK-NEXT:    vfmul.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 1 x double> @llvm.experimental.constrained.fmul.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x double> %vc
+}
+
+define <vscale x 1 x double> @vfmul_vf_nxv1f64(<vscale x 1 x double> %va, double %b) {
+; CHECK-LABEL: vfmul_vf_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
+; CHECK-NEXT:    vfmul.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x double> poison, double %b, i32 0
+  %splat = shufflevector <vscale x 1 x double> %head, <vscale x 1 x double> poison, <vscale x 1 x i32> zeroinitializer
+  %vc = call <vscale x 1 x double> @llvm.experimental.constrained.fmul.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x double> %vc
+}
+
+declare <vscale x 2 x double> @llvm.experimental.constrained.fmul.nxv2f64(<vscale x 2 x double>, <vscale x 2 x double>, metadata, metadata)
+define <vscale x 2 x double> @vfmul_vv_nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %vb) {
+; CHECK-LABEL: vfmul_vv_nxv2f64:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e64, m2, ta, ma
+; CHECK-NEXT:    vfmul.vv v8, v8, v10
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 2 x double> @llvm.experimental.constrained.fmul.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x double> %vc
+}
+
+define <vscale x 2 x double> @vfmul_vf_nxv2f64(<vscale x 2 x double> %va, double %b) {
+; CHECK-LABEL: vfmul_vf_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m2, ta, ma
+; CHECK-NEXT:    vfmul.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x double> poison, double %b, i32 0
+  %splat = shufflevector <vscale x 2 x double> %head, <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
+  %vc = call <vscale x 2 x double> @llvm.experimental.constrained.fmul.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x double> %vc
+}
+
+declare <vscale x 4 x double> @llvm.experimental.constrained.fmul.nxv4f64(<vscale x 4 x double>, <vscale x 4 x double>, metadata, metadata)
+define <vscale x 4 x double> @vfmul_vv_nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %vb) {
+; CHECK-LABEL: vfmul_vv_nxv4f64:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e64, m4, ta, ma
+; CHECK-NEXT:    vfmul.vv v8, v8, v12
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 4 x double> @llvm.experimental.constrained.fmul.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x double> %vc
+}
+
+define <vscale x 4 x double> @vfmul_vf_nxv4f64(<vscale x 4 x double> %va, double %b) {
+; CHECK-LABEL: vfmul_vf_nxv4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m4, ta, ma
+; CHECK-NEXT:    vfmul.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x double> poison, double %b, i32 0
+  %splat = shufflevector <vscale x 4 x double> %head, <vscale x 4 x double> poison, <vscale x 4 x i32> zeroinitializer
+  %vc = call <vscale x 4 x double> @llvm.experimental.constrained.fmul.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x double> %vc
+}
+
+declare <vscale x 8 x double> @llvm.experimental.constrained.fmul.nxv8f64(<vscale x 8 x double>, <vscale x 8 x double>, metadata, metadata)
+define <vscale x 8 x double> @vfmul_vv_nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %vb) {
+; CHECK-LABEL: vfmul_vv_nxv8f64:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e64, m8, ta, ma
+; CHECK-NEXT:    vfmul.vv v8, v8, v16
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 8 x double> @llvm.experimental.constrained.fmul.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x double> %vc
+}
+
+define <vscale x 8 x double> @vfmul_vf_nxv8f64(<vscale x 8 x double> %va, double %b) {
+; CHECK-LABEL: vfmul_vf_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m8, ta, ma
+; CHECK-NEXT:    vfmul.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x double> poison, double %b, i32 0
+  %splat = shufflevector <vscale x 8 x double> %head, <vscale x 8 x double> poison, <vscale x 8 x i32> zeroinitializer
+  %vc = call <vscale x 8 x double> @llvm.experimental.constrained.fmul.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x double> %vc
+}

llvm/test/CodeGen/RISCV/rvv/vfnmadd-constrained-sdnode.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+experimental-zvfh,+v -target-abi=ilp32d \
+; RUN:     -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+experimental-zvfh,+v -target-abi=lp64d \
+; RUN:     -verify-machineinstrs < %s | FileCheck %s
+
+; This tests a mix of vfnmacc and vfnmadd by using different operand orders to
+; trigger commuting in TwoAddressInstructionPass.
+
+declare <vscale x 1 x half> @llvm.experimental.constrained.fma.v1f16(<vscale x 1 x half>, <vscale x 1 x half>, <vscale x 1 x half>, metadata, metadata)
+
+define <vscale x 1 x half> @vfnmsub_vv_nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %vb, <vscale x 1 x half> %vc) {
+; CHECK-LABEL: vfnmsub_vv_nxv1f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vfnmadd.vv v8, v9, v10
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 1 x half> %va
+  %neg2 = fneg <vscale x 1 x half> %vc
+  %vd = call <vscale x 1 x half> @llvm.experimental.constrained.fma.v1f16(<vscale x 1 x half> %neg, <vscale x 1 x half> %vb, <vscale x 1 x half> %neg2,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x half> %vd
+}
+
+define <vscale x 1 x half> @vfnmsub_vf_nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %vb, half %c) {
+; CHECK-LABEL: vfnmsub_vf_nxv1f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vfnmadd.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 1 x half> %head, <vscale x 1 x half> poison, <vscale x 1 x i32> zeroinitializer
+  %neg = fneg <vscale x 1 x half> %va
+  %neg2 = fneg <vscale x 1 x half> %vb
+  %vd = call <vscale x 1 x half> @llvm.experimental.constrained.fma.v1f16(<vscale x 1 x half> %neg, <vscale x 1 x half> %splat, <vscale x 1 x half> %neg2,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x half> %vd
+}
+
+declare <vscale x 2 x half> @llvm.experimental.constrained.fma.v2f16(<vscale x 2 x half>, <vscale x 2 x half>, <vscale x 2 x half>, metadata, metadata)
+
+define <vscale x 2 x half> @vfnmsub_vv_nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %vb, <vscale x 2 x half> %vc) {
+; CHECK-LABEL: vfnmsub_vv_nxv2f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vfnmadd.vv v8, v10, v9
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 2 x half> %va
+  %neg2 = fneg <vscale x 2 x half> %vb
+  %vd = call <vscale x 2 x half> @llvm.experimental.constrained.fma.v2f16(<vscale x 2 x half> %neg, <vscale x 2 x half> %vc, <vscale x 2 x half> %neg2,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x half> %vd
+}
+
+define <vscale x 2 x half> @vfnmsub_vf_nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %vb, half %c) {
+; CHECK-LABEL: vfnmsub_vf_nxv2f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vfnmadd.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 2 x half> %head, <vscale x 2 x half> poison, <vscale x 2 x i32> zeroinitializer
+  %neg = fneg <vscale x 2 x half> %va
+  %neg2 = fneg <vscale x 2 x half> %vb
+  %vd = call <vscale x 2 x half> @llvm.experimental.constrained.fma.v2f16(<vscale x 2 x half> %splat, <vscale x 2 x half> %neg, <vscale x 2 x half> %neg2,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x half> %vd
+}
+
+declare <vscale x 4 x half> @llvm.experimental.constrained.fma.v4f16(<vscale x 4 x half>, <vscale x 4 x half>, <vscale x 4 x half>, metadata, metadata)
+
+define <vscale x 4 x half> @vfnmsub_vv_nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %vb, <vscale x 4 x half> %vc) {
+; CHECK-LABEL: vfnmsub_vv_nxv4f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vfnmadd.vv v8, v9, v10
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 4 x half> %vb
+  %neg2 = fneg <vscale x 4 x half> %vc
+  %vd = call <vscale x 4 x half> @llvm.experimental.constrained.fma.v4f16(<vscale x 4 x half> %neg, <vscale x 4 x half> %va, <vscale x 4 x half> %neg2,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x half> %vd
+}
+
+define <vscale x 4 x half> @vfnmsub_vf_nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %vb, half %c) {
+; CHECK-LABEL: vfnmsub_vf_nxv4f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vfnmadd.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 4 x half> %head, <vscale x 4 x half> poison, <vscale x 4 x i32> zeroinitializer
+  %neg = fneg <vscale x 4 x half> %splat
+  %neg2 = fneg <vscale x 4 x half> %vb
+  %vd = call <vscale x 4 x half> @llvm.experimental.constrained.fma.v4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %neg, <vscale x 4 x half> %neg2,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x half> %vd
+}
+
+declare <vscale x 8 x half> @llvm.experimental.constrained.fma.v8f16(<vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, metadata, metadata)
+
+define <vscale x 8 x half> @vfnmsub_vv_nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %vb, <vscale x 8 x half> %vc) {
+; CHECK-LABEL: vfnmsub_vv_nxv8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m2, ta, ma
+; CHECK-NEXT:    vfnmacc.vv v8, v12, v10
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 8 x half> %vb
+  %neg2 = fneg <vscale x 8 x half> %va
+  %vd = call <vscale x 8 x half> @llvm.experimental.constrained.fma.v8f16(<vscale x 8 x half> %neg, <vscale x 8 x half> %vc, <vscale x 8 x half> %neg2,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x half> %vd
+}
+
+define <vscale x 8 x half> @vfnmsub_vf_nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %vb, half %c) {
+; CHECK-LABEL: vfnmsub_vf_nxv8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m2, ta, ma
+; CHECK-NEXT:    vfnmacc.vf v8, fa0, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 8 x half> %head, <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
+  %neg = fneg <vscale x 8 x half> %splat
+  %neg2 = fneg <vscale x 8 x half> %va
+  %vd = call <vscale x 8 x half> @llvm.experimental.constrained.fma.v8f16(<vscale x 8 x half> %vb, <vscale x 8 x half> %neg, <vscale x 8 x half> %neg2,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x half> %vd
+}
+
+declare <vscale x 16 x half> @llvm.experimental.constrained.fma.v16f16(<vscale x 16 x half>, <vscale x 16 x half>, <vscale x 16 x half>, metadata, metadata)
+
+define <vscale x 16 x half> @vfnmsub_vv_nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %vb, <vscale x 16 x half> %vc) {
+; CHECK-LABEL: vfnmsub_vv_nxv16f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m4, ta, ma
+; CHECK-NEXT:    vfnmadd.vv v8, v16, v12
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 16 x half> %vc
+  %neg2 = fneg <vscale x 16 x half> %vb
+  %vd = call <vscale x 16 x half> @llvm.experimental.constrained.fma.v16f16(<vscale x 16 x half> %neg, <vscale x 16 x half> %va, <vscale x 16 x half> %neg2,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x half> %vd
+}
+
+define <vscale x 16 x half> @vfnmsub_vf_nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %vb, half %c) {
+; CHECK-LABEL: vfnmsub_vf_nxv16f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m4, ta, ma
+; CHECK-NEXT:    vfnmadd.vf v8, fa0, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 16 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 16 x half> %head, <vscale x 16 x half> poison, <vscale x 16 x i32> zeroinitializer
+  %neg = fneg <vscale x 16 x half> %splat
+  %neg2 = fneg <vscale x 16 x half> %vb
+  %vd = call <vscale x 16 x half> @llvm.experimental.constrained.fma.v16f16(<vscale x 16 x half> %neg, <vscale x 16 x half> %va, <vscale x 16 x half> %neg2,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x half> %vd
+}
+
+declare <vscale x 32 x half> @llvm.experimental.constrained.fma.v32f16(<vscale x 32 x half>, <vscale x 32 x half>, <vscale x 32 x half>, metadata, metadata)
+
+define <vscale x 32 x half> @vfnmsub_vv_nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %vb, <vscale x 32 x half> %vc) {
+; CHECK-LABEL: vfnmsub_vv_nxv32f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl8re16.v v24, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e16, m8, ta, ma
+; CHECK-NEXT:    vfnmadd.vv v8, v24, v16
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 32 x half> %vc
+  %neg2 = fneg <vscale x 32 x half> %vb
+  %vd = call <vscale x 32 x half> @llvm.experimental.constrained.fma.v32f16(<vscale x 32 x half> %neg, <vscale x 32 x half> %va, <vscale x 32 x half> %neg2,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 32 x half> %vd
+}
+
+define <vscale x 32 x half> @vfnmsub_vf_nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %vb, half %c) {
+; CHECK-LABEL: vfnmsub_vf_nxv32f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m8, ta, ma
+; CHECK-NEXT:    vfnmacc.vf v8, fa0, v16
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 32 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 32 x half> %head, <vscale x 32 x half> poison, <vscale x 32 x i32> zeroinitializer
+  %neg = fneg <vscale x 32 x half> %splat
+  %neg2 = fneg <vscale x 32 x half> %va
+  %vd = call <vscale x 32 x half> @llvm.experimental.constrained.fma.v32f16(<vscale x 32 x half> %neg, <vscale x 32 x half> %vb, <vscale x 32 x half> %neg2,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 32 x half> %vd
+}
+
+declare <vscale x 1 x float> @llvm.experimental.constrained.fma.v1f32(<vscale x 1 x float>, <vscale x 1 x float>, <vscale x 1 x float>, metadata, metadata)
+
+define <vscale x 1 x float> @vfnmsub_vv_nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %vb, <vscale x 1 x float> %vc) {
+; CHECK-LABEL: vfnmsub_vv_nxv1f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfnmadd.vv v8, v9, v10
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 1 x float> %vb
+  %neg2 = fneg <vscale x 1 x float> %vc
+  %vd = call <vscale x 1 x float> @llvm.experimental.constrained.fma.v1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %neg, <vscale x 1 x float> %neg2,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x float> %vd
+}
+
+define <vscale x 1 x float> @vfnmsub_vf_nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %vb, float %c) {
+; CHECK-LABEL: vfnmsub_vf_nxv1f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfnmadd.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 1 x float> %head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
+  %neg = fneg <vscale x 1 x float> %va
+  %neg2 = fneg <vscale x 1 x float> %vb
+  %vd = call <vscale x 1 x float> @llvm.experimental.constrained.fma.v1f32(<vscale x 1 x float> %neg, <vscale x 1 x float> %splat, <vscale x 1 x float> %neg2,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x float> %vd
+}
+
+declare <vscale x 2 x float> @llvm.experimental.constrained.fma.v2f32(<vscale x 2 x float>, <vscale x 2 x float>, <vscale x 2 x float>, metadata, metadata)
+
+define <vscale x 2 x float> @vfnmsub_vv_nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %vb, <vscale x 2 x float> %vc) {
+; CHECK-LABEL: vfnmsub_vv_nxv2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vfnmadd.vv v8, v10, v9
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 2 x float> %vc
+  %neg2 = fneg <vscale x 2 x float> %vb
+  %vd = call <vscale x 2 x float> @llvm.experimental.constrained.fma.v2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %neg, <vscale x 2 x float> %neg2,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x float> %vd
+}
+
+define <vscale x 2 x float> @vfnmsub_vf_nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %vb, float %c) {
+; CHECK-LABEL: vfnmsub_vf_nxv2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vfnmadd.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 2 x float> %head, <vscale x 2 x float> poison, <vscale x 2 x i32> zeroinitializer
+  %neg = fneg <vscale x 2 x float> %va
+  %neg2 = fneg <vscale x 2 x float> %vb
+  %vd = call <vscale x 2 x float> @llvm.experimental.constrained.fma.v2f32(<vscale x 2 x float> %splat, <vscale x 2 x float> %neg, <vscale x 2 x float> %neg2,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x float> %vd
+}
+
+declare <vscale x 4 x float> @llvm.experimental.constrained.fma.v4f32(<vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, metadata, metadata)
+
+define <vscale x 4 x float> @vfnmsub_vv_nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %vb, <vscale x 4 x float> %vc) {
+; CHECK-LABEL: vfnmsub_vv_nxv4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
+; CHECK-NEXT:    vfnmadd.vv v8, v10, v12
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 4 x float> %va
+  %neg2 = fneg <vscale x 4 x float> %vc
+  %vd = call <vscale x 4 x float> @llvm.experimental.constrained.fma.v4f32(<vscale x 4 x float> %vb, <vscale x 4 x float> %neg, <vscale x 4 x float> %neg2,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x float> %vd
+}
+
+define <vscale x 4 x float> @vfnmsub_vf_nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %vb, float %c) {
+; CHECK-LABEL: vfnmsub_vf_nxv4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
+; CHECK-NEXT:    vfnmadd.vf v8, fa0, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 4 x float> %head, <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
+  %neg = fneg <vscale x 4 x float> %splat
+  %neg2 = fneg <vscale x 4 x float> %vb
+  %vd = call <vscale x 4 x float> @llvm.experimental.constrained.fma.v4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %neg, <vscale x 4 x float> %neg2,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x float> %vd
+}
+
+declare <vscale x 8 x float> @llvm.experimental.constrained.fma.v8f32(<vscale x 8 x float>, <vscale x 8 x float>, <vscale x 8 x float>, metadata, metadata)
+
+define <vscale x 8 x float> @vfnmsub_vv_nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %vb, <vscale x 8 x float> %vc) {
+; CHECK-LABEL: vfnmsub_vv_nxv8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m4, ta, ma
+; CHECK-NEXT:    vfnmacc.vv v8, v16, v12
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 8 x float> %vc
+  %neg2 = fneg <vscale x 8 x float> %va
+  %vd = call <vscale x 8 x float> @llvm.experimental.constrained.fma.v8f32(<vscale x 8 x float> %vb, <vscale x 8 x float> %neg, <vscale x 8 x float> %neg2,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x float> %vd
+}
+
+define <vscale x 8 x float> @vfnmsub_vf_nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %vb, float %c) {
+; CHECK-LABEL: vfnmsub_vf_nxv8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m4, ta, ma
+; CHECK-NEXT:    vfnmacc.vf v8, fa0, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 8 x float> %head, <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer
+  %neg = fneg <vscale x 8 x float> %splat
+  %neg2 = fneg <vscale x 8 x float> %va
+  %vd = call <vscale x 8 x float> @llvm.experimental.constrained.fma.v8f32(<vscale x 8 x float> %vb, <vscale x 8 x float> %neg, <vscale x 8 x float> %neg2,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x float> %vd
+}
+
+declare <vscale x 16 x float> @llvm.experimental.constrained.fma.v16f32(<vscale x 16 x float>, <vscale x 16 x float>, <vscale x 16 x float>, metadata, metadata)
+
+define <vscale x 16 x float> @vfnmsub_vv_nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %vb, <vscale x 16 x float> %vc) {
+; CHECK-LABEL: vfnmsub_vv_nxv16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl8re32.v v24, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e32, m8, ta, ma
+; CHECK-NEXT:    vfnmadd.vv v8, v24, v16
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 16 x float> %va
+  %neg2 = fneg <vscale x 16 x float> %vb
+  %vd = call <vscale x 16 x float> @llvm.experimental.constrained.fma.v16f32(<vscale x 16 x float> %vc, <vscale x 16 x float> %neg, <vscale x 16 x float> %neg2,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x float> %vd
+}
+
+define <vscale x 16 x float> @vfnmsub_vf_nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %vb, float %c) {
+; CHECK-LABEL: vfnmsub_vf_nxv16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m8, ta, ma
+; CHECK-NEXT:    vfnmadd.vf v8, fa0, v16
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 16 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 16 x float> %head, <vscale x 16 x float> poison, <vscale x 16 x i32> zeroinitializer
+  %neg = fneg <vscale x 16 x float> %splat
+  %neg2 = fneg <vscale x 16 x float> %vb
+  %vd = call <vscale x 16 x float> @llvm.experimental.constrained.fma.v16f32(<vscale x 16 x float> %neg, <vscale x 16 x float> %va, <vscale x 16 x float> %neg2,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x float> %vd
+}
+
+declare <vscale x 1 x double> @llvm.experimental.constrained.fma.v1f64(<vscale x 1 x double>, <vscale x 1 x double>, <vscale x 1 x double>, metadata, metadata)
+
+define <vscale x 1 x double> @vfnmsub_vv_nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %vb, <vscale x 1 x double> %vc) {
+; CHECK-LABEL: vfnmsub_vv_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
+; CHECK-NEXT:    vfnmacc.vv v8, v10, v9
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 1 x double> %vb
+  %neg2 = fneg <vscale x 1 x double> %va
+  %vd = call <vscale x 1 x double> @llvm.experimental.constrained.fma.v1f64(<vscale x 1 x double> %vc, <vscale x 1 x double> %neg, <vscale x 1 x double> %neg2,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x double> %vd
+}
+
+define <vscale x 1 x double> @vfnmsub_vf_nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %vb, double %c) {
+; CHECK-LABEL: vfnmsub_vf_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
+; CHECK-NEXT:    vfnmadd.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x double> poison, double %c, i32 0
+  %splat = shufflevector <vscale x 1 x double> %head, <vscale x 1 x double> poison, <vscale x 1 x i32> zeroinitializer
+  %neg = fneg <vscale x 1 x double> %va
+  %neg2 = fneg <vscale x 1 x double> %vb
+  %vd = call <vscale x 1 x double> @llvm.experimental.constrained.fma.v1f64(<vscale x 1 x double> %neg, <vscale x 1 x double> %splat, <vscale x 1 x double> %neg2,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x double> %vd
+}
+
+declare <vscale x 2 x double> @llvm.experimental.constrained.fma.v2f64(<vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, metadata, metadata)
+
+define <vscale x 2 x double> @vfnmsub_vv_nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %vb, <vscale x 2 x double> %vc) {
+; CHECK-LABEL: vfnmsub_vv_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m2, ta, ma
+; CHECK-NEXT:    vfnmadd.vv v8, v12, v10
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 2 x double> %va
+  %neg2 = fneg <vscale x 2 x double> %vb
+  %vd = call <vscale x 2 x double> @llvm.experimental.constrained.fma.v2f64(<vscale x 2 x double> %neg, <vscale x 2 x double> %vc, <vscale x 2 x double> %neg2,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x double> %vd
+}
+
+define <vscale x 2 x double> @vfnmsub_vf_nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %vb, double %c) {
+; CHECK-LABEL: vfnmsub_vf_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m2, ta, ma
+; CHECK-NEXT:    vfnmadd.vf v8, fa0, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x double> poison, double %c, i32 0
+  %splat = shufflevector <vscale x 2 x double> %head, <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
+  %neg = fneg <vscale x 2 x double> %va
+  %neg2 = fneg <vscale x 2 x double> %vb
+  %vd = call <vscale x 2 x double> @llvm.experimental.constrained.fma.v2f64(<vscale x 2 x double> %splat, <vscale x 2 x double> %neg, <vscale x 2 x double> %neg2,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x double> %vd
+}
+
+declare <vscale x 4 x double> @llvm.experimental.constrained.fma.v4f64(<vscale x 4 x double>, <vscale x 4 x double>, <vscale x 4 x double>, metadata, metadata)
+
+define <vscale x 4 x double> @vfnmsub_vv_nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %vb, <vscale x 4 x double> %vc) {
+; CHECK-LABEL: vfnmsub_vv_nxv4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m4, ta, ma
+; CHECK-NEXT:    vfnmadd.vv v8, v12, v16
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 4 x double> %vb
+  %neg2 = fneg <vscale x 4 x double> %vc
+  %vd = call <vscale x 4 x double> @llvm.experimental.constrained.fma.v4f64(<vscale x 4 x double> %neg, <vscale x 4 x double> %va, <vscale x 4 x double> %neg2,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x double> %vd
+}
+
+define <vscale x 4 x double> @vfnmsub_vf_nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %vb, double %c) {
+; CHECK-LABEL: vfnmsub_vf_nxv4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m4, ta, ma
+; CHECK-NEXT:    vfnmadd.vf v8, fa0, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x double> poison, double %c, i32 0
+  %splat = shufflevector <vscale x 4 x double> %head, <vscale x 4 x double> poison, <vscale x 4 x i32> zeroinitializer
+  %neg = fneg <vscale x 4 x double> %splat
+  %neg2 = fneg <vscale x 4 x double> %vb
+  %vd = call <vscale x 4 x double> @llvm.experimental.constrained.fma.v4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %neg, <vscale x 4 x double> %neg2,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x double> %vd
+}
+
+declare <vscale x 8 x double> @llvm.experimental.constrained.fma.v8f64(<vscale x 8 x double>, <vscale x 8 x double>, <vscale x 8 x double>, metadata, metadata)
+
+define <vscale x 8 x double> @vfnmsub_vv_nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %vb, <vscale x 8 x double> %vc) {
+; CHECK-LABEL: vfnmsub_vv_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl8re64.v v24, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e64, m8, ta, ma
+; CHECK-NEXT:    vfnmacc.vv v8, v16, v24
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 8 x double> %vb
+  %neg2 = fneg <vscale x 8 x double> %va
+  %vd = call <vscale x 8 x double> @llvm.experimental.constrained.fma.v8f64(<vscale x 8 x double> %neg, <vscale x 8 x double> %vc, <vscale x 8 x double> %neg2,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x double> %vd
+}
+
+define <vscale x 8 x double> @vfnmsub_vf_nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %vb, double %c) {
+; CHECK-LABEL: vfnmsub_vf_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m8, ta, ma
+; CHECK-NEXT:    vfnmacc.vf v8, fa0, v16
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x double> poison, double %c, i32 0
+  %splat = shufflevector <vscale x 8 x double> %head, <vscale x 8 x double> poison, <vscale x 8 x i32> zeroinitializer
+  %neg = fneg <vscale x 8 x double> %splat
+  %neg2 = fneg <vscale x 8 x double> %va
+  %vd = call <vscale x 8 x double> @llvm.experimental.constrained.fma.v8f64(<vscale x 8 x double> %vb, <vscale x 8 x double> %neg, <vscale x 8 x double> %neg2,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x double> %vd
+}

llvm/test/CodeGen/RISCV/rvv/vfnmsub-constrained-sdnode.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+experimental-zvfh,+v -target-abi=ilp32d \
+; RUN:     -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+experimental-zvfh,+v -target-abi=lp64d \
+; RUN:     -verify-machineinstrs < %s | FileCheck %s
+
+; This tests a mix of vfnmsac and vfnmsub by using different operand orders to
+; trigger commuting in TwoAddressInstructionPass.
+
+declare <vscale x 1 x half> @llvm.experimental.constrained.fma.v1f16(<vscale x 1 x half>, <vscale x 1 x half>, <vscale x 1 x half>, metadata, metadata)
+
+define <vscale x 1 x half> @vfnmsub_vv_nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %vb, <vscale x 1 x half> %vc) {
+; CHECK-LABEL: vfnmsub_vv_nxv1f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vfnmsub.vv v8, v9, v10
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 1 x half> %va
+  %vd = call <vscale x 1 x half> @llvm.experimental.constrained.fma.v1f16(<vscale x 1 x half> %neg, <vscale x 1 x half> %vb, <vscale x 1 x half> %vc,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x half> %vd
+}
+
+define <vscale x 1 x half> @vfnmsub_vf_nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %vb, half %c) {
+; CHECK-LABEL: vfnmsub_vf_nxv1f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vfnmsub.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 1 x half> %head, <vscale x 1 x half> poison, <vscale x 1 x i32> zeroinitializer
+  %neg = fneg <vscale x 1 x half> %va
+  %vd = call <vscale x 1 x half> @llvm.experimental.constrained.fma.v1f16(<vscale x 1 x half> %neg, <vscale x 1 x half> %splat, <vscale x 1 x half> %vb,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x half> %vd
+}
+
+declare <vscale x 2 x half> @llvm.experimental.constrained.fma.v2f16(<vscale x 2 x half>, <vscale x 2 x half>, <vscale x 2 x half>, metadata, metadata)
+
+define <vscale x 2 x half> @vfnmsub_vv_nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %vb, <vscale x 2 x half> %vc) {
+; CHECK-LABEL: vfnmsub_vv_nxv2f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vfnmsub.vv v8, v10, v9
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 2 x half> %va
+  %vd = call <vscale x 2 x half> @llvm.experimental.constrained.fma.v2f16(<vscale x 2 x half> %neg, <vscale x 2 x half> %vc, <vscale x 2 x half> %vb,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x half> %vd
+}
+
+define <vscale x 2 x half> @vfnmsub_vf_nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %vb, half %c) {
+; CHECK-LABEL: vfnmsub_vf_nxv2f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vfnmsub.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 2 x half> %head, <vscale x 2 x half> poison, <vscale x 2 x i32> zeroinitializer
+  %neg = fneg <vscale x 2 x half> %va
+  %vd = call <vscale x 2 x half> @llvm.experimental.constrained.fma.v2f16(<vscale x 2 x half> %splat, <vscale x 2 x half> %neg, <vscale x 2 x half> %vb,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x half> %vd
+}
+
+declare <vscale x 4 x half> @llvm.experimental.constrained.fma.v4f16(<vscale x 4 x half>, <vscale x 4 x half>, <vscale x 4 x half>, metadata, metadata)
+
+define <vscale x 4 x half> @vfnmsub_vv_nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %vb, <vscale x 4 x half> %vc) {
+; CHECK-LABEL: vfnmsub_vv_nxv4f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vfnmsub.vv v8, v9, v10
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 4 x half> %vb
+  %vd = call <vscale x 4 x half> @llvm.experimental.constrained.fma.v4f16(<vscale x 4 x half> %neg, <vscale x 4 x half> %va, <vscale x 4 x half> %vc,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x half> %vd
+}
+
+define <vscale x 4 x half> @vfnmsub_vf_nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %vb, half %c) {
+; CHECK-LABEL: vfnmsub_vf_nxv4f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vfnmsub.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 4 x half> %head, <vscale x 4 x half> poison, <vscale x 4 x i32> zeroinitializer
+  %neg = fneg <vscale x 4 x half> %splat
+  %vd = call <vscale x 4 x half> @llvm.experimental.constrained.fma.v4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %neg, <vscale x 4 x half> %vb,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x half> %vd
+}
+
+declare <vscale x 8 x half> @llvm.experimental.constrained.fma.v8f16(<vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, metadata, metadata)
+
+define <vscale x 8 x half> @vfnmsub_vv_nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %vb, <vscale x 8 x half> %vc) {
+; CHECK-LABEL: vfnmsub_vv_nxv8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m2, ta, ma
+; CHECK-NEXT:    vfnmsac.vv v8, v12, v10
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 8 x half> %vb
+  %vd = call <vscale x 8 x half> @llvm.experimental.constrained.fma.v8f16(<vscale x 8 x half> %neg, <vscale x 8 x half> %vc, <vscale x 8 x half> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x half> %vd
+}
+
+define <vscale x 8 x half> @vfnmsub_vf_nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %vb, half %c) {
+; CHECK-LABEL: vfnmsub_vf_nxv8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m2, ta, ma
+; CHECK-NEXT:    vfnmsac.vf v8, fa0, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 8 x half> %head, <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
+  %neg = fneg <vscale x 8 x half> %splat
+  %vd = call <vscale x 8 x half> @llvm.experimental.constrained.fma.v8f16(<vscale x 8 x half> %vb, <vscale x 8 x half> %neg, <vscale x 8 x half> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x half> %vd
+}
+
+declare <vscale x 16 x half> @llvm.experimental.constrained.fma.v16f16(<vscale x 16 x half>, <vscale x 16 x half>, <vscale x 16 x half>, metadata, metadata)
+
+define <vscale x 16 x half> @vfnmsub_vv_nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %vb, <vscale x 16 x half> %vc) {
+; CHECK-LABEL: vfnmsub_vv_nxv16f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m4, ta, ma
+; CHECK-NEXT:    vfnmsub.vv v8, v16, v12
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 16 x half> %vc
+  %vd = call <vscale x 16 x half> @llvm.experimental.constrained.fma.v16f16(<vscale x 16 x half> %neg, <vscale x 16 x half> %va, <vscale x 16 x half> %vb,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x half> %vd
+}
+
+define <vscale x 16 x half> @vfnmsub_vf_nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %vb, half %c) {
+; CHECK-LABEL: vfnmsub_vf_nxv16f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m4, ta, ma
+; CHECK-NEXT:    vfnmsub.vf v8, fa0, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 16 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 16 x half> %head, <vscale x 16 x half> poison, <vscale x 16 x i32> zeroinitializer
+  %neg = fneg <vscale x 16 x half> %splat
+  %vd = call <vscale x 16 x half> @llvm.experimental.constrained.fma.v16f16(<vscale x 16 x half> %neg, <vscale x 16 x half> %va, <vscale x 16 x half> %vb,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x half> %vd
+}
+
+declare <vscale x 32 x half> @llvm.experimental.constrained.fma.v32f16(<vscale x 32 x half>, <vscale x 32 x half>, <vscale x 32 x half>, metadata, metadata)
+
+define <vscale x 32 x half> @vfnmsub_vv_nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %vb, <vscale x 32 x half> %vc) {
+; CHECK-LABEL: vfnmsub_vv_nxv32f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl8re16.v v24, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e16, m8, ta, ma
+; CHECK-NEXT:    vfnmsub.vv v8, v24, v16
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 32 x half> %vc
+  %vd = call <vscale x 32 x half> @llvm.experimental.constrained.fma.v32f16(<vscale x 32 x half> %neg, <vscale x 32 x half> %va, <vscale x 32 x half> %vb,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 32 x half> %vd
+}
+
+define <vscale x 32 x half> @vfnmsub_vf_nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %vb, half %c) {
+; CHECK-LABEL: vfnmsub_vf_nxv32f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m8, ta, ma
+; CHECK-NEXT:    vfnmsac.vf v8, fa0, v16
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 32 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 32 x half> %head, <vscale x 32 x half> poison, <vscale x 32 x i32> zeroinitializer
+  %neg = fneg <vscale x 32 x half> %splat
+  %vd = call <vscale x 32 x half> @llvm.experimental.constrained.fma.v32f16(<vscale x 32 x half> %neg, <vscale x 32 x half> %vb, <vscale x 32 x half> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 32 x half> %vd
+}
+
+declare <vscale x 1 x float> @llvm.experimental.constrained.fma.v1f32(<vscale x 1 x float>, <vscale x 1 x float>, <vscale x 1 x float>, metadata, metadata)
+
+define <vscale x 1 x float> @vfnmsub_vv_nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %vb, <vscale x 1 x float> %vc) {
+; CHECK-LABEL: vfnmsub_vv_nxv1f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfnmsub.vv v8, v9, v10
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 1 x float> %vb
+  %vd = call <vscale x 1 x float> @llvm.experimental.constrained.fma.v1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %neg, <vscale x 1 x float> %vc,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x float> %vd
+}
+
+define <vscale x 1 x float> @vfnmsub_vf_nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %vb, float %c) {
+; CHECK-LABEL: vfnmsub_vf_nxv1f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfnmsub.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 1 x float> %head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
+  %neg = fneg <vscale x 1 x float> %va
+  %vd = call <vscale x 1 x float> @llvm.experimental.constrained.fma.v1f32(<vscale x 1 x float> %neg, <vscale x 1 x float> %splat, <vscale x 1 x float> %vb,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x float> %vd
+}
+
+declare <vscale x 2 x float> @llvm.experimental.constrained.fma.v2f32(<vscale x 2 x float>, <vscale x 2 x float>, <vscale x 2 x float>, metadata, metadata)
+
+define <vscale x 2 x float> @vfnmsub_vv_nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %vb, <vscale x 2 x float> %vc) {
+; CHECK-LABEL: vfnmsub_vv_nxv2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vfnmsub.vv v8, v10, v9
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 2 x float> %vc
+  %vd = call <vscale x 2 x float> @llvm.experimental.constrained.fma.v2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %neg, <vscale x 2 x float> %vb,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x float> %vd
+}
+
+define <vscale x 2 x float> @vfnmsub_vf_nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %vb, float %c) {
+; CHECK-LABEL: vfnmsub_vf_nxv2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vfnmsub.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 2 x float> %head, <vscale x 2 x float> poison, <vscale x 2 x i32> zeroinitializer
+  %neg = fneg <vscale x 2 x float> %va
+  %vd = call <vscale x 2 x float> @llvm.experimental.constrained.fma.v2f32(<vscale x 2 x float> %splat, <vscale x 2 x float> %neg, <vscale x 2 x float> %vb,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x float> %vd
+}
+
+declare <vscale x 4 x float> @llvm.experimental.constrained.fma.v4f32(<vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, metadata, metadata)
+
+define <vscale x 4 x float> @vfnmsub_vv_nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %vb, <vscale x 4 x float> %vc) {
+; CHECK-LABEL: vfnmsub_vv_nxv4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
+; CHECK-NEXT:    vfnmsub.vv v8, v10, v12
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 4 x float> %va
+  %vd = call <vscale x 4 x float> @llvm.experimental.constrained.fma.v4f32(<vscale x 4 x float> %vb, <vscale x 4 x float> %neg, <vscale x 4 x float> %vc,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x float> %vd
+}
+
+define <vscale x 4 x float> @vfnmsub_vf_nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %vb, float %c) {
+; CHECK-LABEL: vfnmsub_vf_nxv4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
+; CHECK-NEXT:    vfnmsub.vf v8, fa0, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 4 x float> %head, <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
+  %neg = fneg <vscale x 4 x float> %splat
+  %vd = call <vscale x 4 x float> @llvm.experimental.constrained.fma.v4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %neg, <vscale x 4 x float> %vb,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x float> %vd
+}
+
+declare <vscale x 8 x float> @llvm.experimental.constrained.fma.v8f32(<vscale x 8 x float>, <vscale x 8 x float>, <vscale x 8 x float>, metadata, metadata)
+
+define <vscale x 8 x float> @vfnmsub_vv_nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %vb, <vscale x 8 x float> %vc) {
+; CHECK-LABEL: vfnmsub_vv_nxv8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m4, ta, ma
+; CHECK-NEXT:    vfnmsac.vv v8, v16, v12
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 8 x float> %vc
+  %vd = call <vscale x 8 x float> @llvm.experimental.constrained.fma.v8f32(<vscale x 8 x float> %vb, <vscale x 8 x float> %neg, <vscale x 8 x float> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x float> %vd
+}
+
+define <vscale x 8 x float> @vfnmsub_vf_nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %vb, float %c) {
+; CHECK-LABEL: vfnmsub_vf_nxv8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m4, ta, ma
+; CHECK-NEXT:    vfnmsac.vf v8, fa0, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 8 x float> %head, <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer
+  %neg = fneg <vscale x 8 x float> %splat
+  %vd = call <vscale x 8 x float> @llvm.experimental.constrained.fma.v8f32(<vscale x 8 x float> %vb, <vscale x 8 x float> %neg, <vscale x 8 x float> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x float> %vd
+}
+
+declare <vscale x 16 x float> @llvm.experimental.constrained.fma.v16f32(<vscale x 16 x float>, <vscale x 16 x float>, <vscale x 16 x float>, metadata, metadata)
+
+define <vscale x 16 x float> @vfnmsub_vv_nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %vb, <vscale x 16 x float> %vc) {
+; CHECK-LABEL: vfnmsub_vv_nxv16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl8re32.v v24, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e32, m8, ta, ma
+; CHECK-NEXT:    vfnmsub.vv v8, v24, v16
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 16 x float> %va
+  %vd = call <vscale x 16 x float> @llvm.experimental.constrained.fma.v16f32(<vscale x 16 x float> %vc, <vscale x 16 x float> %neg, <vscale x 16 x float> %vb,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x float> %vd
+}
+
+define <vscale x 16 x float> @vfnmsub_vf_nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %vb, float %c) {
+; CHECK-LABEL: vfnmsub_vf_nxv16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m8, ta, ma
+; CHECK-NEXT:    vfnmsub.vf v8, fa0, v16
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 16 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 16 x float> %head, <vscale x 16 x float> poison, <vscale x 16 x i32> zeroinitializer
+  %neg = fneg <vscale x 16 x float> %splat
+  %vd = call <vscale x 16 x float> @llvm.experimental.constrained.fma.v16f32(<vscale x 16 x float> %neg, <vscale x 16 x float> %va, <vscale x 16 x float> %vb,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x float> %vd
+}
+
+declare <vscale x 1 x double> @llvm.experimental.constrained.fma.v1f64(<vscale x 1 x double>, <vscale x 1 x double>, <vscale x 1 x double>, metadata, metadata)
+
+define <vscale x 1 x double> @vfnmsub_vv_nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %vb, <vscale x 1 x double> %vc) {
+; CHECK-LABEL: vfnmsub_vv_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
+; CHECK-NEXT:    vfnmsac.vv v8, v10, v9
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 1 x double> %vb
+  %vd = call <vscale x 1 x double> @llvm.experimental.constrained.fma.v1f64(<vscale x 1 x double> %vc, <vscale x 1 x double> %neg, <vscale x 1 x double> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x double> %vd
+}
+
+define <vscale x 1 x double> @vfnmsub_vf_nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %vb, double %c) {
+; CHECK-LABEL: vfnmsub_vf_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
+; CHECK-NEXT:    vfnmsub.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x double> poison, double %c, i32 0
+  %splat = shufflevector <vscale x 1 x double> %head, <vscale x 1 x double> poison, <vscale x 1 x i32> zeroinitializer
+  %neg = fneg <vscale x 1 x double> %va
+  %vd = call <vscale x 1 x double> @llvm.experimental.constrained.fma.v1f64(<vscale x 1 x double> %neg, <vscale x 1 x double> %splat, <vscale x 1 x double> %vb,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x double> %vd
+}
+
+declare <vscale x 2 x double> @llvm.experimental.constrained.fma.v2f64(<vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, metadata, metadata)
+
+define <vscale x 2 x double> @vfnmsub_vv_nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %vb, <vscale x 2 x double> %vc) {
+; CHECK-LABEL: vfnmsub_vv_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m2, ta, ma
+; CHECK-NEXT:    vfnmsub.vv v8, v12, v10
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 2 x double> %va
+  %vd = call <vscale x 2 x double> @llvm.experimental.constrained.fma.v2f64(<vscale x 2 x double> %neg, <vscale x 2 x double> %vc, <vscale x 2 x double> %vb,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x double> %vd
+}
+
+define <vscale x 2 x double> @vfnmsub_vf_nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %vb, double %c) {
+; CHECK-LABEL: vfnmsub_vf_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m2, ta, ma
+; CHECK-NEXT:    vfnmsub.vf v8, fa0, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x double> poison, double %c, i32 0
+  %splat = shufflevector <vscale x 2 x double> %head, <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
+  %neg = fneg <vscale x 2 x double> %va
+  %vd = call <vscale x 2 x double> @llvm.experimental.constrained.fma.v2f64(<vscale x 2 x double> %splat, <vscale x 2 x double> %neg, <vscale x 2 x double> %vb,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x double> %vd
+}
+
+declare <vscale x 4 x double> @llvm.experimental.constrained.fma.v4f64(<vscale x 4 x double>, <vscale x 4 x double>, <vscale x 4 x double>, metadata, metadata)
+
+define <vscale x 4 x double> @vfnmsub_vv_nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %vb, <vscale x 4 x double> %vc) {
+; CHECK-LABEL: vfnmsub_vv_nxv4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m4, ta, ma
+; CHECK-NEXT:    vfnmsub.vv v8, v12, v16
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 4 x double> %vb
+  %vd = call <vscale x 4 x double> @llvm.experimental.constrained.fma.v4f64(<vscale x 4 x double> %neg, <vscale x 4 x double> %va, <vscale x 4 x double> %vc,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x double> %vd
+}
+
+define <vscale x 4 x double> @vfnmsub_vf_nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %vb, double %c) {
+; CHECK-LABEL: vfnmsub_vf_nxv4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m4, ta, ma
+; CHECK-NEXT:    vfnmsub.vf v8, fa0, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x double> poison, double %c, i32 0
+  %splat = shufflevector <vscale x 4 x double> %head, <vscale x 4 x double> poison, <vscale x 4 x i32> zeroinitializer
+  %neg = fneg <vscale x 4 x double> %splat
+  %vd = call <vscale x 4 x double> @llvm.experimental.constrained.fma.v4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %neg, <vscale x 4 x double> %vb,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x double> %vd
+}
+
+declare <vscale x 8 x double> @llvm.experimental.constrained.fma.v8f64(<vscale x 8 x double>, <vscale x 8 x double>, <vscale x 8 x double>, metadata, metadata)
+
+define <vscale x 8 x double> @vfnmsub_vv_nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %vb, <vscale x 8 x double> %vc) {
+; CHECK-LABEL: vfnmsub_vv_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vl8re64.v v24, (a0)
+; CHECK-NEXT:    vsetvli a0, zero, e64, m8, ta, ma
+; CHECK-NEXT:    vfnmsac.vv v8, v16, v24
+; CHECK-NEXT:    ret
+  %neg = fneg <vscale x 8 x double> %vb
+  %vd = call <vscale x 8 x double> @llvm.experimental.constrained.fma.v8f64(<vscale x 8 x double> %neg, <vscale x 8 x double> %vc, <vscale x 8 x double> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x double> %vd
+}
+
+define <vscale x 8 x double> @vfnmsub_vf_nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %vb, double %c) {
+; CHECK-LABEL: vfnmsub_vf_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m8, ta, ma
+; CHECK-NEXT:    vfnmsac.vf v8, fa0, v16
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x double> poison, double %c, i32 0
+  %splat = shufflevector <vscale x 8 x double> %head, <vscale x 8 x double> poison, <vscale x 8 x i32> zeroinitializer
+  %neg = fneg <vscale x 8 x double> %splat
+  %vd = call <vscale x 8 x double> @llvm.experimental.constrained.fma.v8f64(<vscale x 8 x double> %vb, <vscale x 8 x double> %neg, <vscale x 8 x double> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x double> %vd
+}

llvm/test/CodeGen/RISCV/rvv/vfsub-constrained-sdnode.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+experimental-zvfh,+v -target-abi=ilp32d \
+; RUN:     -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+experimental-zvfh,+v -target-abi=lp64d \
+; RUN:     -verify-machineinstrs < %s | FileCheck %s
+
+declare <vscale x 1 x half> @llvm.experimental.constrained.fsub.nxv1f16(<vscale x 1 x half>, <vscale x 1 x half>, metadata, metadata)
+define <vscale x 1 x half> @vfsub_vv_nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %vb) {
+; CHECK-LABEL: vfsub_vv_nxv1f16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vfsub.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 1 x half> @llvm.experimental.constrained.fsub.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x half> %vc
+}
+
+define <vscale x 1 x half> @vfsub_vf_nxv1f16(<vscale x 1 x half> %va, half %b) {
+; CHECK-LABEL: vfsub_vf_nxv1f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vfsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x half> poison, half %b, i32 0
+  %splat = shufflevector <vscale x 1 x half> %head, <vscale x 1 x half> poison, <vscale x 1 x i32> zeroinitializer
+  %vc = call <vscale x 1 x half> @llvm.experimental.constrained.fsub.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x half> %vc
+}
+
+declare <vscale x 2 x half> @llvm.experimental.constrained.fsub.nxv2f16(<vscale x 2 x half>, <vscale x 2 x half>, metadata, metadata)
+define <vscale x 2 x half> @vfsub_vv_nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %vb) {
+; CHECK-LABEL: vfsub_vv_nxv2f16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vfsub.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 2 x half> @llvm.experimental.constrained.fsub.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x half> %vc
+}
+
+define <vscale x 2 x half> @vfsub_vf_nxv2f16(<vscale x 2 x half> %va, half %b) {
+; CHECK-LABEL: vfsub_vf_nxv2f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vfsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x half> poison, half %b, i32 0
+  %splat = shufflevector <vscale x 2 x half> %head, <vscale x 2 x half> poison, <vscale x 2 x i32> zeroinitializer
+  %vc = call <vscale x 2 x half> @llvm.experimental.constrained.fsub.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x half> %vc
+}
+
+declare <vscale x 4 x half> @llvm.experimental.constrained.fsub.nxv4f16(<vscale x 4 x half>, <vscale x 4 x half>, metadata, metadata)
+define <vscale x 4 x half> @vfsub_vv_nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %vb) {
+; CHECK-LABEL: vfsub_vv_nxv4f16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vfsub.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 4 x half> @llvm.experimental.constrained.fsub.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x half> %vc
+}
+
+define <vscale x 4 x half> @vfsub_vf_nxv4f16(<vscale x 4 x half> %va, half %b) {
+; CHECK-LABEL: vfsub_vf_nxv4f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vfsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x half> poison, half %b, i32 0
+  %splat = shufflevector <vscale x 4 x half> %head, <vscale x 4 x half> poison, <vscale x 4 x i32> zeroinitializer
+  %vc = call <vscale x 4 x half> @llvm.experimental.constrained.fsub.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x half> %vc
+}
+
+declare <vscale x 8 x half> @llvm.experimental.constrained.fsub.nxv8f16(<vscale x 8 x half>, <vscale x 8 x half>, metadata, metadata)
+define <vscale x 8 x half> @vfsub_vv_nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %vb) {
+; CHECK-LABEL: vfsub_vv_nxv8f16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e16, m2, ta, ma
+; CHECK-NEXT:    vfsub.vv v8, v8, v10
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 8 x half> @llvm.experimental.constrained.fsub.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x half> %vc
+}
+
+define <vscale x 8 x half> @vfsub_vf_nxv8f16(<vscale x 8 x half> %va, half %b) {
+; CHECK-LABEL: vfsub_vf_nxv8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m2, ta, ma
+; CHECK-NEXT:    vfsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x half> poison, half %b, i32 0
+  %splat = shufflevector <vscale x 8 x half> %head, <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
+  %vc = call <vscale x 8 x half> @llvm.experimental.constrained.fsub.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x half> %vc
+}
+
+define <vscale x 8 x half> @vfsub_fv_nxv8f16(<vscale x 8 x half> %va, half %b) {
+; CHECK-LABEL: vfsub_fv_nxv8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m2, ta, ma
+; CHECK-NEXT:    vfrsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x half> poison, half %b, i32 0
+  %splat = shufflevector <vscale x 8 x half> %head, <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
+  %vc = call <vscale x 8 x half> @llvm.experimental.constrained.fsub.nxv8f16(<vscale x 8 x half> %splat, <vscale x 8 x half> %va, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x half> %vc
+}
+
+declare <vscale x 16 x half> @llvm.experimental.constrained.fsub.nxv16f16(<vscale x 16 x half>, <vscale x 16 x half>, metadata, metadata)
+define <vscale x 16 x half> @vfsub_vv_nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %vb) {
+; CHECK-LABEL: vfsub_vv_nxv16f16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e16, m4, ta, ma
+; CHECK-NEXT:    vfsub.vv v8, v8, v12
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 16 x half> @llvm.experimental.constrained.fsub.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x half> %vc
+}
+
+define <vscale x 16 x half> @vfsub_vf_nxv16f16(<vscale x 16 x half> %va, half %b) {
+; CHECK-LABEL: vfsub_vf_nxv16f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m4, ta, ma
+; CHECK-NEXT:    vfsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 16 x half> poison, half %b, i32 0
+  %splat = shufflevector <vscale x 16 x half> %head, <vscale x 16 x half> poison, <vscale x 16 x i32> zeroinitializer
+  %vc = call <vscale x 16 x half> @llvm.experimental.constrained.fsub.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x half> %vc
+}
+
+declare <vscale x 32 x half> @llvm.experimental.constrained.fsub.nxv32f16(<vscale x 32 x half>, <vscale x 32 x half>, metadata, metadata)
+define <vscale x 32 x half> @vfsub_vv_nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %vb) {
+; CHECK-LABEL: vfsub_vv_nxv32f16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e16, m8, ta, ma
+; CHECK-NEXT:    vfsub.vv v8, v8, v16
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 32 x half> @llvm.experimental.constrained.fsub.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 32 x half> %vc
+}
+
+define <vscale x 32 x half> @vfsub_vf_nxv32f16(<vscale x 32 x half> %va, half %b) {
+; CHECK-LABEL: vfsub_vf_nxv32f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m8, ta, ma
+; CHECK-NEXT:    vfsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 32 x half> poison, half %b, i32 0
+  %splat = shufflevector <vscale x 32 x half> %head, <vscale x 32 x half> poison, <vscale x 32 x i32> zeroinitializer
+  %vc = call <vscale x 32 x half> @llvm.experimental.constrained.fsub.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 32 x half> %vc
+}
+
+declare <vscale x 1 x float> @llvm.experimental.constrained.fsub.nxv1f32(<vscale x 1 x float>, <vscale x 1 x float>, metadata, metadata)
+define <vscale x 1 x float> @vfsub_vv_nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %vb) {
+; CHECK-LABEL: vfsub_vv_nxv1f32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfsub.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 1 x float> @llvm.experimental.constrained.fsub.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x float> %vc
+}
+
+define <vscale x 1 x float> @vfsub_vf_nxv1f32(<vscale x 1 x float> %va, float %b) {
+; CHECK-LABEL: vfsub_vf_nxv1f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x float> poison, float %b, i32 0
+  %splat = shufflevector <vscale x 1 x float> %head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
+  %vc = call <vscale x 1 x float> @llvm.experimental.constrained.fsub.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x float> %vc
+}
+
+declare <vscale x 2 x float> @llvm.experimental.constrained.fsub.nxv2f32(<vscale x 2 x float>, <vscale x 2 x float>, metadata, metadata)
+define <vscale x 2 x float> @vfsub_vv_nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %vb) {
+; CHECK-LABEL: vfsub_vv_nxv2f32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vfsub.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 2 x float> @llvm.experimental.constrained.fsub.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x float> %vc
+}
+
+define <vscale x 2 x float> @vfsub_vf_nxv2f32(<vscale x 2 x float> %va, float %b) {
+; CHECK-LABEL: vfsub_vf_nxv2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vfsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x float> poison, float %b, i32 0
+  %splat = shufflevector <vscale x 2 x float> %head, <vscale x 2 x float> poison, <vscale x 2 x i32> zeroinitializer
+  %vc = call <vscale x 2 x float> @llvm.experimental.constrained.fsub.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x float> %vc
+}
+
+declare <vscale x 4 x float> @llvm.experimental.constrained.fsub.nxv4f32(<vscale x 4 x float>, <vscale x 4 x float>, metadata, metadata)
+define <vscale x 4 x float> @vfsub_vv_nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %vb) {
+; CHECK-LABEL: vfsub_vv_nxv4f32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
+; CHECK-NEXT:    vfsub.vv v8, v8, v10
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 4 x float> @llvm.experimental.constrained.fsub.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x float> %vc
+}
+
+define <vscale x 4 x float> @vfsub_vf_nxv4f32(<vscale x 4 x float> %va, float %b) {
+; CHECK-LABEL: vfsub_vf_nxv4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
+; CHECK-NEXT:    vfsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x float> poison, float %b, i32 0
+  %splat = shufflevector <vscale x 4 x float> %head, <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
+  %vc = call <vscale x 4 x float> @llvm.experimental.constrained.fsub.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x float> %vc
+}
+
+declare <vscale x 8 x float> @llvm.experimental.constrained.fsub.nxv8f32(<vscale x 8 x float>, <vscale x 8 x float>, metadata, metadata)
+define <vscale x 8 x float> @vfsub_vv_nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %vb) {
+; CHECK-LABEL: vfsub_vv_nxv8f32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e32, m4, ta, ma
+; CHECK-NEXT:    vfsub.vv v8, v8, v12
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 8 x float> @llvm.experimental.constrained.fsub.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x float> %vc
+}
+
+define <vscale x 8 x float> @vfsub_vf_nxv8f32(<vscale x 8 x float> %va, float %b) {
+; CHECK-LABEL: vfsub_vf_nxv8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m4, ta, ma
+; CHECK-NEXT:    vfsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x float> poison, float %b, i32 0
+  %splat = shufflevector <vscale x 8 x float> %head, <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer
+  %vc = call <vscale x 8 x float> @llvm.experimental.constrained.fsub.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x float> %vc
+}
+
+define <vscale x 8 x float> @vfsub_fv_nxv8f32(<vscale x 8 x float> %va, float %b) {
+; CHECK-LABEL: vfsub_fv_nxv8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m4, ta, ma
+; CHECK-NEXT:    vfrsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x float> poison, float %b, i32 0
+  %splat = shufflevector <vscale x 8 x float> %head, <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer
+  %vc = call <vscale x 8 x float> @llvm.experimental.constrained.fsub.nxv8f32(<vscale x 8 x float> %splat, <vscale x 8 x float> %va, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x float> %vc
+}
+
+declare <vscale x 16 x float> @llvm.experimental.constrained.fsub.nxv16f32(<vscale x 16 x float>, <vscale x 16 x float>, metadata, metadata)
+define <vscale x 16 x float> @vfsub_vv_nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %vb) {
+; CHECK-LABEL: vfsub_vv_nxv16f32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e32, m8, ta, ma
+; CHECK-NEXT:    vfsub.vv v8, v8, v16
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 16 x float> @llvm.experimental.constrained.fsub.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x float> %vc
+}
+
+define <vscale x 16 x float> @vfsub_vf_nxv16f32(<vscale x 16 x float> %va, float %b) {
+; CHECK-LABEL: vfsub_vf_nxv16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m8, ta, ma
+; CHECK-NEXT:    vfsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 16 x float> poison, float %b, i32 0
+  %splat = shufflevector <vscale x 16 x float> %head, <vscale x 16 x float> poison, <vscale x 16 x i32> zeroinitializer
+  %vc = call <vscale x 16 x float> @llvm.experimental.constrained.fsub.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x float> %vc
+}
+
+declare <vscale x 1 x double> @llvm.experimental.constrained.fsub.nxv1f64(<vscale x 1 x double>, <vscale x 1 x double>, metadata, metadata)
+define <vscale x 1 x double> @vfsub_vv_nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %vb) {
+; CHECK-LABEL: vfsub_vv_nxv1f64:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
+; CHECK-NEXT:    vfsub.vv v8, v8, v9
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 1 x double> @llvm.experimental.constrained.fsub.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x double> %vc
+}
+
+define <vscale x 1 x double> @vfsub_vf_nxv1f64(<vscale x 1 x double> %va, double %b) {
+; CHECK-LABEL: vfsub_vf_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
+; CHECK-NEXT:    vfsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x double> poison, double %b, i32 0
+  %splat = shufflevector <vscale x 1 x double> %head, <vscale x 1 x double> poison, <vscale x 1 x i32> zeroinitializer
+  %vc = call <vscale x 1 x double> @llvm.experimental.constrained.fsub.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x double> %vc
+}
+
+declare <vscale x 2 x double> @llvm.experimental.constrained.fsub.nxv2f64(<vscale x 2 x double>, <vscale x 2 x double>, metadata, metadata)
+define <vscale x 2 x double> @vfsub_vv_nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %vb) {
+; CHECK-LABEL: vfsub_vv_nxv2f64:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e64, m2, ta, ma
+; CHECK-NEXT:    vfsub.vv v8, v8, v10
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 2 x double> @llvm.experimental.constrained.fsub.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x double> %vc
+}
+
+define <vscale x 2 x double> @vfsub_vf_nxv2f64(<vscale x 2 x double> %va, double %b) {
+; CHECK-LABEL: vfsub_vf_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m2, ta, ma
+; CHECK-NEXT:    vfsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x double> poison, double %b, i32 0
+  %splat = shufflevector <vscale x 2 x double> %head, <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
+  %vc = call <vscale x 2 x double> @llvm.experimental.constrained.fsub.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x double> %vc
+}
+
+declare <vscale x 4 x double> @llvm.experimental.constrained.fsub.nxv4f64(<vscale x 4 x double>, <vscale x 4 x double>, metadata, metadata)
+define <vscale x 4 x double> @vfsub_vv_nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %vb) {
+; CHECK-LABEL: vfsub_vv_nxv4f64:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e64, m4, ta, ma
+; CHECK-NEXT:    vfsub.vv v8, v8, v12
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 4 x double> @llvm.experimental.constrained.fsub.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x double> %vc
+}
+
+define <vscale x 4 x double> @vfsub_vf_nxv4f64(<vscale x 4 x double> %va, double %b) {
+; CHECK-LABEL: vfsub_vf_nxv4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m4, ta, ma
+; CHECK-NEXT:    vfsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x double> poison, double %b, i32 0
+  %splat = shufflevector <vscale x 4 x double> %head, <vscale x 4 x double> poison, <vscale x 4 x i32> zeroinitializer
+  %vc = call <vscale x 4 x double> @llvm.experimental.constrained.fsub.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x double> %vc
+}
+
+declare <vscale x 8 x double> @llvm.experimental.constrained.fsub.nxv8f64(<vscale x 8 x double>, <vscale x 8 x double>, metadata, metadata)
+define <vscale x 8 x double> @vfsub_vv_nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %vb) {
+; CHECK-LABEL: vfsub_vv_nxv8f64:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a0, zero, e64, m8, ta, ma
+; CHECK-NEXT:    vfsub.vv v8, v8, v16
+; CHECK-NEXT:    ret
+entry:
+  %vc = call <vscale x 8 x double> @llvm.experimental.constrained.fsub.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %vb, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x double> %vc
+}
+
+define <vscale x 8 x double> @vfsub_vf_nxv8f64(<vscale x 8 x double> %va, double %b) {
+; CHECK-LABEL: vfsub_vf_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m8, ta, ma
+; CHECK-NEXT:    vfsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x double> poison, double %b, i32 0
+  %splat = shufflevector <vscale x 8 x double> %head, <vscale x 8 x double> poison, <vscale x 8 x i32> zeroinitializer
+  %vc = call <vscale x 8 x double> @llvm.experimental.constrained.fsub.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x double> %vc
+}
+
+define <vscale x 8 x double> @vfsub_fv_nxv8f64(<vscale x 8 x double> %va, double %b) {
+; CHECK-LABEL: vfsub_fv_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e64, m8, ta, ma
+; CHECK-NEXT:    vfrsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x double> poison, double %b, i32 0
+  %splat = shufflevector <vscale x 8 x double> %head, <vscale x 8 x double> poison, <vscale x 8 x i32> zeroinitializer
+  %vc = call <vscale x 8 x double> @llvm.experimental.constrained.fsub.nxv8f64(<vscale x 8 x double> %splat, <vscale x 8 x double> %va, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x double> %vc
+}

llvm/test/CodeGen/RISCV/rvv/vfwadd-constrained-sdnode.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+experimental-zvfh,+v -target-abi=ilp32d \
+; RUN:     -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+experimental-zvfh,+v -target-abi=lp64d \
+; RUN:     -verify-machineinstrs < %s | FileCheck %s
+
+declare <vscale x 1 x double> @llvm.experimental.constrained.fadd.nxv1f64(<vscale x 1 x double>, <vscale x 1 x double>, metadata, metadata)
+define <vscale x 1 x double> @vfwadd_vv_nxv1f64(<vscale x 1 x float> %va, <vscale x 1 x float> %vb) {
+; CHECK-LABEL: vfwadd_vv_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfwadd.vv v10, v8, v9
+; CHECK-NEXT:    vmv1r.v v8, v10
+; CHECK-NEXT:    ret
+  %vc = fpext <vscale x 1 x float> %va to <vscale x 1 x double>
+  %vd = fpext <vscale x 1 x float> %vb to <vscale x 1 x double>
+  %ve = call <vscale x 1 x double> @llvm.experimental.constrained.fadd.nxv1f64(<vscale x 1 x double> %vc, <vscale x 1 x double> %vd, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x double> %ve
+}
+
+define <vscale x 1 x double> @vfwadd_vf_nxv1f64(<vscale x 1 x float> %va, float %b) {
+; CHECK-LABEL: vfwadd_vf_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfwadd.vf v9, v8, fa0
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x float> poison, float %b, i32 0
+  %splat = shufflevector <vscale x 1 x float> %head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
+  %vc = fpext <vscale x 1 x float> %va to <vscale x 1 x double>
+  %vd = fpext <vscale x 1 x float> %splat to <vscale x 1 x double>
+  %ve = call <vscale x 1 x double> @llvm.experimental.constrained.fadd.nxv1f64(<vscale x 1 x double> %vc, <vscale x 1 x double> %vd, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x double> %ve
+}
+
+define <vscale x 1 x double> @vfwadd_vf_nxv1f64_2(<vscale x 1 x float> %va, float %b) {
+; CHECK-LABEL: vfwadd_vf_nxv1f64_2:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfwadd.vf v9, v8, fa0
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+  %fpext = fpext float %b to double
+  %head = insertelement <vscale x 1 x double> poison, double %fpext, i32 0
+  %splat = shufflevector <vscale x 1 x double> %head, <vscale x 1 x double> poison, <vscale x 1 x i32> zeroinitializer
+  %vc = fpext <vscale x 1 x float> %va to <vscale x 1 x double>
+  %ve = call <vscale x 1 x double> @llvm.experimental.constrained.fadd.nxv1f64(<vscale x 1 x double> %vc, <vscale x 1 x double> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x double> %ve
+}
+
+define <vscale x 1 x double> @vfwadd_wv_nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x float> %vb) {
+; CHECK-LABEL: vfwadd_wv_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfwadd.wv v8, v8, v9
+; CHECK-NEXT:    ret
+  %vc = fpext <vscale x 1 x float> %vb to <vscale x 1 x double>
+  %vd = call <vscale x 1 x double> @llvm.experimental.constrained.fadd.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %vc, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x double> %vd
+}
+
+define <vscale x 1 x double> @vfwadd_wf_nxv1f64(<vscale x 1 x double> %va, float %b) {
+; CHECK-LABEL: vfwadd_wf_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfwadd.wf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x float> poison, float %b, i32 0
+  %splat = shufflevector <vscale x 1 x float> %head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
+  %vc = fpext <vscale x 1 x float> %splat to <vscale x 1 x double>
+  %vd = call <vscale x 1 x double> @llvm.experimental.constrained.fadd.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %vc, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x double> %vd
+}
+
+define <vscale x 1 x double> @vfwadd_wf_nxv1f64_2(<vscale x 1 x double> %va, float %b) {
+; CHECK-LABEL: vfwadd_wf_nxv1f64_2:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfwadd.wf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %fpext = fpext float %b to double
+  %head = insertelement <vscale x 1 x double> poison, double %fpext, i32 0
+  %splat = shufflevector <vscale x 1 x double> %head, <vscale x 1 x double> poison, <vscale x 1 x i32> zeroinitializer
+  %vd = call <vscale x 1 x double> @llvm.experimental.constrained.fadd.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x double> %vd
+}

llvm/test/CodeGen/RISCV/rvv/vfwmacc-constrained-sdnode.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+experimental-zvfh,+v -target-abi=ilp32d \
+; RUN:     -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+experimental-zvfh,+v -target-abi=lp64d \
+; RUN:     -verify-machineinstrs < %s | FileCheck %s
+
+declare <vscale x 1 x float> @llvm.experimental.constrained.fma.v1f32(<vscale x 1 x float>, <vscale x 1 x float>, <vscale x 1 x float>, metadata, metadata)
+
+define <vscale x 1 x float> @vfwmacc_vv_nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x half> %vb, <vscale x 1 x half> %vc) {
+; CHECK-LABEL: vfwmacc_vv_nxv1f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vfwmacc.vv v8, v9, v10
+; CHECK-NEXT:    ret
+  %vd = fpext <vscale x 1 x half> %vb to <vscale x 1 x float>
+  %ve = fpext <vscale x 1 x half> %vc to <vscale x 1 x float>
+  %vf = call <vscale x 1 x float> @llvm.experimental.constrained.fma.v1f32(<vscale x 1 x float> %vd, <vscale x 1 x float> %ve, <vscale x 1 x float> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x float> %vf
+}
+
+define <vscale x 1 x float> @vfwmacc_vf_nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x half> %vb, half %c) {
+; CHECK-LABEL: vfwmacc_vf_nxv1f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vfwmacc.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 1 x half> %head, <vscale x 1 x half> poison, <vscale x 1 x i32> zeroinitializer
+  %vd = fpext <vscale x 1 x half> %vb to <vscale x 1 x float>
+  %ve = fpext <vscale x 1 x half> %splat to <vscale x 1 x float>
+  %vf = call <vscale x 1 x float> @llvm.experimental.constrained.fma.v1f32(<vscale x 1 x float> %vd, <vscale x 1 x float> %ve, <vscale x 1 x float> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x float> %vf
+}
+
+define <vscale x 1 x float> @vfwnmacc_vv_nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x half> %vb, <vscale x 1 x half> %vc) {
+; CHECK-LABEL: vfwnmacc_vv_nxv1f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vfwnmacc.vv v8, v9, v10
+; CHECK-NEXT:    ret
+  %vd = fpext <vscale x 1 x half> %vb to <vscale x 1 x float>
+  %ve = fpext <vscale x 1 x half> %vc to <vscale x 1 x float>
+  %vf = fneg <vscale x 1 x float> %va
+  %vg = fneg <vscale x 1 x float> %vd
+  %vh = call <vscale x 1 x float> @llvm.experimental.constrained.fma.v1f32(<vscale x 1 x float> %vg, <vscale x 1 x float> %ve, <vscale x 1 x float> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x float> %vh
+}
+
+define <vscale x 1 x float> @vfwnmacc_vf_nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x half> %vb, half %c) {
+; CHECK-LABEL: vfwnmacc_vf_nxv1f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 1 x half> %head, <vscale x 1 x half> poison, <vscale x 1 x i32> zeroinitializer
+  %vd = fpext <vscale x 1 x half> %vb to <vscale x 1 x float>
+  %ve = fpext <vscale x 1 x half> %splat to <vscale x 1 x float>
+  %vf = fneg <vscale x 1 x float> %va
+  %vg = fneg <vscale x 1 x float> %vd
+  %vh = call <vscale x 1 x float> @llvm.experimental.constrained.fma.v1f32(<vscale x 1 x float> %vg, <vscale x 1 x float> %ve, <vscale x 1 x float> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x float> %vh
+}
+
+define <vscale x 1 x float> @vfwnmacc_fv_nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x half> %vb, half %c) {
+; CHECK-LABEL: vfwnmacc_fv_nxv1f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 1 x half> %head, <vscale x 1 x half> poison, <vscale x 1 x i32> zeroinitializer
+  %vd = fpext <vscale x 1 x half> %vb to <vscale x 1 x float>
+  %ve = fpext <vscale x 1 x half> %splat to <vscale x 1 x float>
+  %vf = fneg <vscale x 1 x float> %va
+  %vg = fneg <vscale x 1 x float> %ve
+  %vh = call <vscale x 1 x float> @llvm.experimental.constrained.fma.v1f32(<vscale x 1 x float> %vd, <vscale x 1 x float> %vg, <vscale x 1 x float> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x float> %vh
+}
+
+define <vscale x 1 x float> @vfwmsac_vv_nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x half> %vb, <vscale x 1 x half> %vc) {
+; CHECK-LABEL: vfwmsac_vv_nxv1f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vfwmsac.vv v8, v9, v10
+; CHECK-NEXT:    ret
+  %vd = fpext <vscale x 1 x half> %vb to <vscale x 1 x float>
+  %ve = fpext <vscale x 1 x half> %vc to <vscale x 1 x float>
+  %vf = fneg <vscale x 1 x float> %va
+  %vg = call <vscale x 1 x float> @llvm.experimental.constrained.fma.v1f32(<vscale x 1 x float> %vd, <vscale x 1 x float> %ve, <vscale x 1 x float> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x float> %vg
+}
+
+define <vscale x 1 x float> @vfwmsac_vf_nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x half> %vb, half %c) {
+; CHECK-LABEL: vfwmsac_vf_nxv1f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vfwmsac.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 1 x half> %head, <vscale x 1 x half> poison, <vscale x 1 x i32> zeroinitializer
+  %vd = fpext <vscale x 1 x half> %vb to <vscale x 1 x float>
+  %ve = fpext <vscale x 1 x half> %splat to <vscale x 1 x float>
+  %vf = fneg <vscale x 1 x float> %va
+  %vg = call <vscale x 1 x float> @llvm.experimental.constrained.fma.v1f32(<vscale x 1 x float> %vd, <vscale x 1 x float> %ve, <vscale x 1 x float> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x float> %vg
+}
+
+define <vscale x 1 x float> @vfwnmsac_vv_nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x half> %vb, <vscale x 1 x half> %vc) {
+; CHECK-LABEL: vfwnmsac_vv_nxv1f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vfwnmsac.vv v8, v9, v10
+; CHECK-NEXT:    ret
+  %vd = fpext <vscale x 1 x half> %vb to <vscale x 1 x float>
+  %ve = fpext <vscale x 1 x half> %vc to <vscale x 1 x float>
+  %vf = fneg <vscale x 1 x float> %vd
+  %vg = call <vscale x 1 x float> @llvm.experimental.constrained.fma.v1f32(<vscale x 1 x float> %vf, <vscale x 1 x float> %ve, <vscale x 1 x float> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x float> %vg
+}
+
+define <vscale x 1 x float> @vfwnmsac_vf_nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x half> %vb, half %c) {
+; CHECK-LABEL: vfwnmsac_vf_nxv1f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 1 x half> %head, <vscale x 1 x half> poison, <vscale x 1 x i32> zeroinitializer
+  %vd = fpext <vscale x 1 x half> %vb to <vscale x 1 x float>
+  %ve = fpext <vscale x 1 x half> %splat to <vscale x 1 x float>
+  %vf = fneg <vscale x 1 x float> %vd
+  %vg = call <vscale x 1 x float> @llvm.experimental.constrained.fma.v1f32(<vscale x 1 x float> %vf, <vscale x 1 x float> %ve, <vscale x 1 x float> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x float> %vg
+}
+
+define <vscale x 1 x float> @vfwnmsac_fv_nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x half> %vb, half %c) {
+; CHECK-LABEL: vfwnmsac_fv_nxv1f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 1 x half> %head, <vscale x 1 x half> poison, <vscale x 1 x i32> zeroinitializer
+  %vd = fpext <vscale x 1 x half> %vb to <vscale x 1 x float>
+  %ve = fpext <vscale x 1 x half> %splat to <vscale x 1 x float>
+  %vf = fneg <vscale x 1 x float> %ve
+  %vg = call <vscale x 1 x float> @llvm.experimental.constrained.fma.v1f32(<vscale x 1 x float> %vd, <vscale x 1 x float> %vf, <vscale x 1 x float> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x float> %vg
+}
+
+declare <vscale x 2 x float> @llvm.experimental.constrained.fma.v2f32(<vscale x 2 x float>, <vscale x 2 x float>, <vscale x 2 x float>, metadata, metadata)
+
+define <vscale x 2 x float> @vfwmacc_vv_nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x half> %vb, <vscale x 2 x half> %vc) {
+; CHECK-LABEL: vfwmacc_vv_nxv2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vfwmacc.vv v8, v9, v10
+; CHECK-NEXT:    ret
+  %vd = fpext <vscale x 2 x half> %vb to <vscale x 2 x float>
+  %ve = fpext <vscale x 2 x half> %vc to <vscale x 2 x float>
+  %vf = call <vscale x 2 x float> @llvm.experimental.constrained.fma.v2f32(<vscale x 2 x float> %vd, <vscale x 2 x float> %ve, <vscale x 2 x float> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x float> %vf
+}
+
+define <vscale x 2 x float> @vfwmacc_vf_nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x half> %vb, half %c) {
+; CHECK-LABEL: vfwmacc_vf_nxv2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vfwmacc.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 2 x half> %head, <vscale x 2 x half> poison, <vscale x 2 x i32> zeroinitializer
+  %vd = fpext <vscale x 2 x half> %vb to <vscale x 2 x float>
+  %ve = fpext <vscale x 2 x half> %splat to <vscale x 2 x float>
+  %vf = call <vscale x 2 x float> @llvm.experimental.constrained.fma.v2f32(<vscale x 2 x float> %vd, <vscale x 2 x float> %ve, <vscale x 2 x float> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x float> %vf
+}
+
+define <vscale x 2 x float> @vfwnmacc_vv_nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x half> %vb, <vscale x 2 x half> %vc) {
+; CHECK-LABEL: vfwnmacc_vv_nxv2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vfwnmacc.vv v8, v9, v10
+; CHECK-NEXT:    ret
+  %vd = fpext <vscale x 2 x half> %vb to <vscale x 2 x float>
+  %ve = fpext <vscale x 2 x half> %vc to <vscale x 2 x float>
+  %vf = fneg <vscale x 2 x float> %va
+  %vg = fneg <vscale x 2 x float> %vd
+  %vh = call <vscale x 2 x float> @llvm.experimental.constrained.fma.v2f32(<vscale x 2 x float> %vg, <vscale x 2 x float> %ve, <vscale x 2 x float> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x float> %vh
+}
+
+define <vscale x 2 x float> @vfwnmacc_vf_nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x half> %vb, half %c) {
+; CHECK-LABEL: vfwnmacc_vf_nxv2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 2 x half> %head, <vscale x 2 x half> poison, <vscale x 2 x i32> zeroinitializer
+  %vd = fpext <vscale x 2 x half> %vb to <vscale x 2 x float>
+  %ve = fpext <vscale x 2 x half> %splat to <vscale x 2 x float>
+  %vf = fneg <vscale x 2 x float> %va
+  %vg = fneg <vscale x 2 x float> %vd
+  %vh = call <vscale x 2 x float> @llvm.experimental.constrained.fma.v2f32(<vscale x 2 x float> %vg, <vscale x 2 x float> %ve, <vscale x 2 x float> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x float> %vh
+}
+
+define <vscale x 2 x float> @vfwnmacc_fv_nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x half> %vb, half %c) {
+; CHECK-LABEL: vfwnmacc_fv_nxv2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 2 x half> %head, <vscale x 2 x half> poison, <vscale x 2 x i32> zeroinitializer
+  %vd = fpext <vscale x 2 x half> %vb to <vscale x 2 x float>
+  %ve = fpext <vscale x 2 x half> %splat to <vscale x 2 x float>
+  %vf = fneg <vscale x 2 x float> %va
+  %vg = fneg <vscale x 2 x float> %ve
+  %vh = call <vscale x 2 x float> @llvm.experimental.constrained.fma.v2f32(<vscale x 2 x float> %vd, <vscale x 2 x float> %vg, <vscale x 2 x float> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x float> %vh
+}
+
+define <vscale x 2 x float> @vfwmsac_vv_nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x half> %vb, <vscale x 2 x half> %vc) {
+; CHECK-LABEL: vfwmsac_vv_nxv2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vfwmsac.vv v8, v9, v10
+; CHECK-NEXT:    ret
+  %vd = fpext <vscale x 2 x half> %vb to <vscale x 2 x float>
+  %ve = fpext <vscale x 2 x half> %vc to <vscale x 2 x float>
+  %vf = fneg <vscale x 2 x float> %va
+  %vg = call <vscale x 2 x float> @llvm.experimental.constrained.fma.v2f32(<vscale x 2 x float> %vd, <vscale x 2 x float> %ve, <vscale x 2 x float> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x float> %vg
+}
+
+define <vscale x 2 x float> @vfwmsac_vf_nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x half> %vb, half %c) {
+; CHECK-LABEL: vfwmsac_vf_nxv2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vfwmsac.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 2 x half> %head, <vscale x 2 x half> poison, <vscale x 2 x i32> zeroinitializer
+  %vd = fpext <vscale x 2 x half> %vb to <vscale x 2 x float>
+  %ve = fpext <vscale x 2 x half> %splat to <vscale x 2 x float>
+  %vf = fneg <vscale x 2 x float> %va
+  %vg = call <vscale x 2 x float> @llvm.experimental.constrained.fma.v2f32(<vscale x 2 x float> %vd, <vscale x 2 x float> %ve, <vscale x 2 x float> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x float> %vg
+}
+
+define <vscale x 2 x float> @vfwnmsac_vv_nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x half> %vb, <vscale x 2 x half> %vc) {
+; CHECK-LABEL: vfwnmsac_vv_nxv2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vfwnmsac.vv v8, v9, v10
+; CHECK-NEXT:    ret
+  %vd = fpext <vscale x 2 x half> %vb to <vscale x 2 x float>
+  %ve = fpext <vscale x 2 x half> %vc to <vscale x 2 x float>
+  %vf = fneg <vscale x 2 x float> %vd
+  %vg = call <vscale x 2 x float> @llvm.experimental.constrained.fma.v2f32(<vscale x 2 x float> %vf, <vscale x 2 x float> %ve, <vscale x 2 x float> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x float> %vg
+}
+
+define <vscale x 2 x float> @vfwnmsac_vf_nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x half> %vb, half %c) {
+; CHECK-LABEL: vfwnmsac_vf_nxv2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 2 x half> %head, <vscale x 2 x half> poison, <vscale x 2 x i32> zeroinitializer
+  %vd = fpext <vscale x 2 x half> %vb to <vscale x 2 x float>
+  %ve = fpext <vscale x 2 x half> %splat to <vscale x 2 x float>
+  %vf = fneg <vscale x 2 x float> %vd
+  %vg = call <vscale x 2 x float> @llvm.experimental.constrained.fma.v2f32(<vscale x 2 x float> %vf, <vscale x 2 x float> %ve, <vscale x 2 x float> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x float> %vg
+}
+
+define <vscale x 2 x float> @vfwnmsac_fv_nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x half> %vb, half %c) {
+; CHECK-LABEL: vfwnmsac_fv_nxv2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 2 x half> %head, <vscale x 2 x half> poison, <vscale x 2 x i32> zeroinitializer
+  %vd = fpext <vscale x 2 x half> %vb to <vscale x 2 x float>
+  %ve = fpext <vscale x 2 x half> %splat to <vscale x 2 x float>
+  %vf = fneg <vscale x 2 x float> %ve
+  %vg = call <vscale x 2 x float> @llvm.experimental.constrained.fma.v2f32(<vscale x 2 x float> %vd, <vscale x 2 x float> %vf, <vscale x 2 x float> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x float> %vg
+}
+
+
+declare <vscale x 4 x float> @llvm.experimental.constrained.fma.v4f32(<vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, metadata, metadata)
+
+define <vscale x 4 x float> @vfwmacc_vv_nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x half> %vb, <vscale x 4 x half> %vc) {
+; CHECK-LABEL: vfwmacc_vv_nxv4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vfwmacc.vv v8, v10, v11
+; CHECK-NEXT:    ret
+  %vd = fpext <vscale x 4 x half> %vb to <vscale x 4 x float>
+  %ve = fpext <vscale x 4 x half> %vc to <vscale x 4 x float>
+  %vf = call <vscale x 4 x float> @llvm.experimental.constrained.fma.v4f32(<vscale x 4 x float> %vd, <vscale x 4 x float> %ve, <vscale x 4 x float> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x float> %vf
+}
+
+define <vscale x 4 x float> @vfwmacc_vf_nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x half> %vb, half %c) {
+; CHECK-LABEL: vfwmacc_vf_nxv4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vfwmacc.vf v8, fa0, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 4 x half> %head, <vscale x 4 x half> poison, <vscale x 4 x i32> zeroinitializer
+  %vd = fpext <vscale x 4 x half> %vb to <vscale x 4 x float>
+  %ve = fpext <vscale x 4 x half> %splat to <vscale x 4 x float>
+  %vf = call <vscale x 4 x float> @llvm.experimental.constrained.fma.v4f32(<vscale x 4 x float> %vd, <vscale x 4 x float> %ve, <vscale x 4 x float> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x float> %vf
+}
+
+define <vscale x 4 x float> @vfwnmacc_vv_nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x half> %vb, <vscale x 4 x half> %vc) {
+; CHECK-LABEL: vfwnmacc_vv_nxv4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vfwnmacc.vv v8, v10, v11
+; CHECK-NEXT:    ret
+  %vd = fpext <vscale x 4 x half> %vb to <vscale x 4 x float>
+  %ve = fpext <vscale x 4 x half> %vc to <vscale x 4 x float>
+  %vf = fneg <vscale x 4 x float> %va
+  %vg = fneg <vscale x 4 x float> %vd
+  %vh = call <vscale x 4 x float> @llvm.experimental.constrained.fma.v4f32(<vscale x 4 x float> %vg, <vscale x 4 x float> %ve, <vscale x 4 x float> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x float> %vh
+}
+
+define <vscale x 4 x float> @vfwnmacc_vf_nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x half> %vb, half %c) {
+; CHECK-LABEL: vfwnmacc_vf_nxv4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 4 x half> %head, <vscale x 4 x half> poison, <vscale x 4 x i32> zeroinitializer
+  %vd = fpext <vscale x 4 x half> %vb to <vscale x 4 x float>
+  %ve = fpext <vscale x 4 x half> %splat to <vscale x 4 x float>
+  %vf = fneg <vscale x 4 x float> %va
+  %vg = fneg <vscale x 4 x float> %vd
+  %vh = call <vscale x 4 x float> @llvm.experimental.constrained.fma.v4f32(<vscale x 4 x float> %vg, <vscale x 4 x float> %ve, <vscale x 4 x float> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x float> %vh
+}
+
+define <vscale x 4 x float> @vfwnmacc_fv_nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x half> %vb, half %c) {
+; CHECK-LABEL: vfwnmacc_fv_nxv4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 4 x half> %head, <vscale x 4 x half> poison, <vscale x 4 x i32> zeroinitializer
+  %vd = fpext <vscale x 4 x half> %vb to <vscale x 4 x float>
+  %ve = fpext <vscale x 4 x half> %splat to <vscale x 4 x float>
+  %vf = fneg <vscale x 4 x float> %va
+  %vg = fneg <vscale x 4 x float> %ve
+  %vh = call <vscale x 4 x float> @llvm.experimental.constrained.fma.v4f32(<vscale x 4 x float> %vd, <vscale x 4 x float> %vg, <vscale x 4 x float> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x float> %vh
+}
+
+define <vscale x 4 x float> @vfwmsac_vv_nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x half> %vb, <vscale x 4 x half> %vc) {
+; CHECK-LABEL: vfwmsac_vv_nxv4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vfwmsac.vv v8, v10, v11
+; CHECK-NEXT:    ret
+  %vd = fpext <vscale x 4 x half> %vb to <vscale x 4 x float>
+  %ve = fpext <vscale x 4 x half> %vc to <vscale x 4 x float>
+  %vf = fneg <vscale x 4 x float> %va
+  %vg = call <vscale x 4 x float> @llvm.experimental.constrained.fma.v4f32(<vscale x 4 x float> %vd, <vscale x 4 x float> %ve, <vscale x 4 x float> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x float> %vg
+}
+
+define <vscale x 4 x float> @vfwmsac_vf_nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x half> %vb, half %c) {
+; CHECK-LABEL: vfwmsac_vf_nxv4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vfwmsac.vf v8, fa0, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 4 x half> %head, <vscale x 4 x half> poison, <vscale x 4 x i32> zeroinitializer
+  %vd = fpext <vscale x 4 x half> %vb to <vscale x 4 x float>
+  %ve = fpext <vscale x 4 x half> %splat to <vscale x 4 x float>
+  %vf = fneg <vscale x 4 x float> %va
+  %vg = call <vscale x 4 x float> @llvm.experimental.constrained.fma.v4f32(<vscale x 4 x float> %vd, <vscale x 4 x float> %ve, <vscale x 4 x float> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x float> %vg
+}
+
+define <vscale x 4 x float> @vfwnmsac_vv_nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x half> %vb, <vscale x 4 x half> %vc) {
+; CHECK-LABEL: vfwnmsac_vv_nxv4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vfwnmsac.vv v8, v10, v11
+; CHECK-NEXT:    ret
+  %vd = fpext <vscale x 4 x half> %vb to <vscale x 4 x float>
+  %ve = fpext <vscale x 4 x half> %vc to <vscale x 4 x float>
+  %vf = fneg <vscale x 4 x float> %vd
+  %vg = call <vscale x 4 x float> @llvm.experimental.constrained.fma.v4f32(<vscale x 4 x float> %vf, <vscale x 4 x float> %ve, <vscale x 4 x float> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x float> %vg
+}
+
+define <vscale x 4 x float> @vfwnmsac_vf_nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x half> %vb, half %c) {
+; CHECK-LABEL: vfwnmsac_vf_nxv4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 4 x half> %head, <vscale x 4 x half> poison, <vscale x 4 x i32> zeroinitializer
+  %vd = fpext <vscale x 4 x half> %vb to <vscale x 4 x float>
+  %ve = fpext <vscale x 4 x half> %splat to <vscale x 4 x float>
+  %vf = fneg <vscale x 4 x float> %vd
+  %vg = call <vscale x 4 x float> @llvm.experimental.constrained.fma.v4f32(<vscale x 4 x float> %vf, <vscale x 4 x float> %ve, <vscale x 4 x float> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x float> %vg
+}
+
+define <vscale x 4 x float> @vfwnmsac_fv_nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x half> %vb, half %c) {
+; CHECK-LABEL: vfwnmsac_fv_nxv4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 4 x half> %head, <vscale x 4 x half> poison, <vscale x 4 x i32> zeroinitializer
+  %vd = fpext <vscale x 4 x half> %vb to <vscale x 4 x float>
+  %ve = fpext <vscale x 4 x half> %splat to <vscale x 4 x float>
+  %vf = fneg <vscale x 4 x float> %ve
+  %vg = call <vscale x 4 x float> @llvm.experimental.constrained.fma.v4f32(<vscale x 4 x float> %vd, <vscale x 4 x float> %vf, <vscale x 4 x float> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x float> %vg
+}
+
+declare <vscale x 8 x float> @llvm.experimental.constrained.fma.v8f32(<vscale x 8 x float>, <vscale x 8 x float>, <vscale x 8 x float>, metadata, metadata)
+
+define <vscale x 8 x float> @vfwmacc_vv_nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x half> %vb, <vscale x 8 x half> %vc) {
+; CHECK-LABEL: vfwmacc_vv_nxv8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m2, ta, ma
+; CHECK-NEXT:    vfwmacc.vv v8, v12, v14
+; CHECK-NEXT:    ret
+  %vd = fpext <vscale x 8 x half> %vb to <vscale x 8 x float>
+  %ve = fpext <vscale x 8 x half> %vc to <vscale x 8 x float>
+  %vf = call <vscale x 8 x float> @llvm.experimental.constrained.fma.v8f32(<vscale x 8 x float> %vd, <vscale x 8 x float> %ve, <vscale x 8 x float> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x float> %vf
+}
+
+define <vscale x 8 x float> @vfwmacc_vf_nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x half> %vb, half %c) {
+; CHECK-LABEL: vfwmacc_vf_nxv8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m2, ta, ma
+; CHECK-NEXT:    vfwmacc.vf v8, fa0, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 8 x half> %head, <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
+  %vd = fpext <vscale x 8 x half> %vb to <vscale x 8 x float>
+  %ve = fpext <vscale x 8 x half> %splat to <vscale x 8 x float>
+  %vf = call <vscale x 8 x float> @llvm.experimental.constrained.fma.v8f32(<vscale x 8 x float> %vd, <vscale x 8 x float> %ve, <vscale x 8 x float> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x float> %vf
+}
+
+define <vscale x 8 x float> @vfwnmacc_vv_nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x half> %vb, <vscale x 8 x half> %vc) {
+; CHECK-LABEL: vfwnmacc_vv_nxv8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m2, ta, ma
+; CHECK-NEXT:    vfwnmacc.vv v8, v12, v14
+; CHECK-NEXT:    ret
+  %vd = fpext <vscale x 8 x half> %vb to <vscale x 8 x float>
+  %ve = fpext <vscale x 8 x half> %vc to <vscale x 8 x float>
+  %vf = fneg <vscale x 8 x float> %va
+  %vg = fneg <vscale x 8 x float> %vd
+  %vh = call <vscale x 8 x float> @llvm.experimental.constrained.fma.v8f32(<vscale x 8 x float> %vg, <vscale x 8 x float> %ve, <vscale x 8 x float> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x float> %vh
+}
+
+define <vscale x 8 x float> @vfwnmacc_vf_nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x half> %vb, half %c) {
+; CHECK-LABEL: vfwnmacc_vf_nxv8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m2, ta, ma
+; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 8 x half> %head, <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
+  %vd = fpext <vscale x 8 x half> %vb to <vscale x 8 x float>
+  %ve = fpext <vscale x 8 x half> %splat to <vscale x 8 x float>
+  %vf = fneg <vscale x 8 x float> %va
+  %vg = fneg <vscale x 8 x float> %vd
+  %vh = call <vscale x 8 x float> @llvm.experimental.constrained.fma.v8f32(<vscale x 8 x float> %vg, <vscale x 8 x float> %ve, <vscale x 8 x float> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x float> %vh
+}
+
+define <vscale x 8 x float> @vfwnmacc_fv_nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x half> %vb, half %c) {
+; CHECK-LABEL: vfwnmacc_fv_nxv8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m2, ta, ma
+; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 8 x half> %head, <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
+  %vd = fpext <vscale x 8 x half> %vb to <vscale x 8 x float>
+  %ve = fpext <vscale x 8 x half> %splat to <vscale x 8 x float>
+  %vf = fneg <vscale x 8 x float> %va
+  %vg = fneg <vscale x 8 x float> %ve
+  %vh = call <vscale x 8 x float> @llvm.experimental.constrained.fma.v8f32(<vscale x 8 x float> %vd, <vscale x 8 x float> %vg, <vscale x 8 x float> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x float> %vh
+}
+
+define <vscale x 8 x float> @vfwmsac_vv_nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x half> %vb, <vscale x 8 x half> %vc) {
+; CHECK-LABEL: vfwmsac_vv_nxv8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m2, ta, ma
+; CHECK-NEXT:    vfwmsac.vv v8, v12, v14
+; CHECK-NEXT:    ret
+  %vd = fpext <vscale x 8 x half> %vb to <vscale x 8 x float>
+  %ve = fpext <vscale x 8 x half> %vc to <vscale x 8 x float>
+  %vf = fneg <vscale x 8 x float> %va
+  %vg = call <vscale x 8 x float> @llvm.experimental.constrained.fma.v8f32(<vscale x 8 x float> %vd, <vscale x 8 x float> %ve, <vscale x 8 x float> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x float> %vg
+}
+
+define <vscale x 8 x float> @vfwmsac_vf_nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x half> %vb, half %c) {
+; CHECK-LABEL: vfwmsac_vf_nxv8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m2, ta, ma
+; CHECK-NEXT:    vfwmsac.vf v8, fa0, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 8 x half> %head, <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
+  %vd = fpext <vscale x 8 x half> %vb to <vscale x 8 x float>
+  %ve = fpext <vscale x 8 x half> %splat to <vscale x 8 x float>
+  %vf = fneg <vscale x 8 x float> %va
+  %vg = call <vscale x 8 x float> @llvm.experimental.constrained.fma.v8f32(<vscale x 8 x float> %vd, <vscale x 8 x float> %ve, <vscale x 8 x float> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x float> %vg
+}
+
+define <vscale x 8 x float> @vfwnmsac_vv_nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x half> %vb, <vscale x 8 x half> %vc) {
+; CHECK-LABEL: vfwnmsac_vv_nxv8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m2, ta, ma
+; CHECK-NEXT:    vfwnmsac.vv v8, v12, v14
+; CHECK-NEXT:    ret
+  %vd = fpext <vscale x 8 x half> %vb to <vscale x 8 x float>
+  %ve = fpext <vscale x 8 x half> %vc to <vscale x 8 x float>
+  %vf = fneg <vscale x 8 x float> %vd
+  %vg = call <vscale x 8 x float> @llvm.experimental.constrained.fma.v8f32(<vscale x 8 x float> %vf, <vscale x 8 x float> %ve, <vscale x 8 x float> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x float> %vg
+}
+
+define <vscale x 8 x float> @vfwnmsac_vf_nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x half> %vb, half %c) {
+; CHECK-LABEL: vfwnmsac_vf_nxv8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m2, ta, ma
+; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 8 x half> %head, <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
+  %vd = fpext <vscale x 8 x half> %vb to <vscale x 8 x float>
+  %ve = fpext <vscale x 8 x half> %splat to <vscale x 8 x float>
+  %vf = fneg <vscale x 8 x float> %vd
+  %vg = call <vscale x 8 x float> @llvm.experimental.constrained.fma.v8f32(<vscale x 8 x float> %vf, <vscale x 8 x float> %ve, <vscale x 8 x float> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x float> %vg
+}
+
+define <vscale x 8 x float> @vfwnmsac_fv_nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x half> %vb, half %c) {
+; CHECK-LABEL: vfwnmsac_fv_nxv8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m2, ta, ma
+; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 8 x half> %head, <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
+  %vd = fpext <vscale x 8 x half> %vb to <vscale x 8 x float>
+  %ve = fpext <vscale x 8 x half> %splat to <vscale x 8 x float>
+  %vf = fneg <vscale x 8 x float> %ve
+  %vg = call <vscale x 8 x float> @llvm.experimental.constrained.fma.v8f32(<vscale x 8 x float> %vd, <vscale x 8 x float> %vf, <vscale x 8 x float> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x float> %vg
+}
+
+declare <vscale x 16 x float> @llvm.experimental.constrained.fma.v16f32(<vscale x 16 x float>, <vscale x 16 x float>, <vscale x 16 x float>, metadata, metadata)
+
+define <vscale x 16 x float> @vfwmacc_vv_nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x half> %vb, <vscale x 16 x half> %vc) {
+; CHECK-LABEL: vfwmacc_vv_nxv16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m4, ta, ma
+; CHECK-NEXT:    vfwmacc.vv v8, v16, v20
+; CHECK-NEXT:    ret
+  %vd = fpext <vscale x 16 x half> %vb to <vscale x 16 x float>
+  %ve = fpext <vscale x 16 x half> %vc to <vscale x 16 x float>
+  %vf = call <vscale x 16 x float> @llvm.experimental.constrained.fma.v16f32(<vscale x 16 x float> %vd, <vscale x 16 x float> %ve, <vscale x 16 x float> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x float> %vf
+}
+
+define <vscale x 16 x float> @vfwmacc_vf_nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x half> %vb, half %c) {
+; CHECK-LABEL: vfwmacc_vf_nxv16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m4, ta, ma
+; CHECK-NEXT:    vfwmacc.vf v8, fa0, v16
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 16 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 16 x half> %head, <vscale x 16 x half> poison, <vscale x 16 x i32> zeroinitializer
+  %vd = fpext <vscale x 16 x half> %vb to <vscale x 16 x float>
+  %ve = fpext <vscale x 16 x half> %splat to <vscale x 16 x float>
+  %vf = call <vscale x 16 x float> @llvm.experimental.constrained.fma.v16f32(<vscale x 16 x float> %vd, <vscale x 16 x float> %ve, <vscale x 16 x float> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x float> %vf
+}
+
+define <vscale x 16 x float> @vfwnmacc_vv_nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x half> %vb, <vscale x 16 x half> %vc) {
+; CHECK-LABEL: vfwnmacc_vv_nxv16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m4, ta, ma
+; CHECK-NEXT:    vfwnmacc.vv v8, v16, v20
+; CHECK-NEXT:    ret
+  %vd = fpext <vscale x 16 x half> %vb to <vscale x 16 x float>
+  %ve = fpext <vscale x 16 x half> %vc to <vscale x 16 x float>
+  %vf = fneg <vscale x 16 x float> %va
+  %vg = fneg <vscale x 16 x float> %vd
+  %vh = call <vscale x 16 x float> @llvm.experimental.constrained.fma.v16f32(<vscale x 16 x float> %vg, <vscale x 16 x float> %ve, <vscale x 16 x float> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x float> %vh
+}
+
+define <vscale x 16 x float> @vfwnmacc_vf_nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x half> %vb, half %c) {
+; CHECK-LABEL: vfwnmacc_vf_nxv16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m4, ta, ma
+; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v16
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 16 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 16 x half> %head, <vscale x 16 x half> poison, <vscale x 16 x i32> zeroinitializer
+  %vd = fpext <vscale x 16 x half> %vb to <vscale x 16 x float>
+  %ve = fpext <vscale x 16 x half> %splat to <vscale x 16 x float>
+  %vf = fneg <vscale x 16 x float> %va
+  %vg = fneg <vscale x 16 x float> %vd
+  %vh = call <vscale x 16 x float> @llvm.experimental.constrained.fma.v16f32(<vscale x 16 x float> %vg, <vscale x 16 x float> %ve, <vscale x 16 x float> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x float> %vh
+}
+
+define <vscale x 16 x float> @vfwnmacc_fv_nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x half> %vb, half %c) {
+; CHECK-LABEL: vfwnmacc_fv_nxv16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m4, ta, ma
+; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v16
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 16 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 16 x half> %head, <vscale x 16 x half> poison, <vscale x 16 x i32> zeroinitializer
+  %vd = fpext <vscale x 16 x half> %vb to <vscale x 16 x float>
+  %ve = fpext <vscale x 16 x half> %splat to <vscale x 16 x float>
+  %vf = fneg <vscale x 16 x float> %va
+  %vg = fneg <vscale x 16 x float> %ve
+  %vh = call <vscale x 16 x float> @llvm.experimental.constrained.fma.v16f32(<vscale x 16 x float> %vd, <vscale x 16 x float> %vg, <vscale x 16 x float> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x float> %vh
+}
+
+define <vscale x 16 x float> @vfwmsac_vv_nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x half> %vb, <vscale x 16 x half> %vc) {
+; CHECK-LABEL: vfwmsac_vv_nxv16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m4, ta, ma
+; CHECK-NEXT:    vfwmsac.vv v8, v16, v20
+; CHECK-NEXT:    ret
+  %vd = fpext <vscale x 16 x half> %vb to <vscale x 16 x float>
+  %ve = fpext <vscale x 16 x half> %vc to <vscale x 16 x float>
+  %vf = fneg <vscale x 16 x float> %va
+  %vg = call <vscale x 16 x float> @llvm.experimental.constrained.fma.v16f32(<vscale x 16 x float> %vd, <vscale x 16 x float> %ve, <vscale x 16 x float> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x float> %vg
+}
+
+define <vscale x 16 x float> @vfwmsac_vf_nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x half> %vb, half %c) {
+; CHECK-LABEL: vfwmsac_vf_nxv16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m4, ta, ma
+; CHECK-NEXT:    vfwmsac.vf v8, fa0, v16
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 16 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 16 x half> %head, <vscale x 16 x half> poison, <vscale x 16 x i32> zeroinitializer
+  %vd = fpext <vscale x 16 x half> %vb to <vscale x 16 x float>
+  %ve = fpext <vscale x 16 x half> %splat to <vscale x 16 x float>
+  %vf = fneg <vscale x 16 x float> %va
+  %vg = call <vscale x 16 x float> @llvm.experimental.constrained.fma.v16f32(<vscale x 16 x float> %vd, <vscale x 16 x float> %ve, <vscale x 16 x float> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x float> %vg
+}
+
+define <vscale x 16 x float> @vfwnmsac_vv_nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x half> %vb, <vscale x 16 x half> %vc) {
+; CHECK-LABEL: vfwnmsac_vv_nxv16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m4, ta, ma
+; CHECK-NEXT:    vfwnmsac.vv v8, v16, v20
+; CHECK-NEXT:    ret
+  %vd = fpext <vscale x 16 x half> %vb to <vscale x 16 x float>
+  %ve = fpext <vscale x 16 x half> %vc to <vscale x 16 x float>
+  %vf = fneg <vscale x 16 x float> %vd
+  %vg = call <vscale x 16 x float> @llvm.experimental.constrained.fma.v16f32(<vscale x 16 x float> %vf, <vscale x 16 x float> %ve, <vscale x 16 x float> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x float> %vg
+}
+
+define <vscale x 16 x float> @vfwnmsac_vf_nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x half> %vb, half %c) {
+; CHECK-LABEL: vfwnmsac_vf_nxv16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m4, ta, ma
+; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v16
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 16 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 16 x half> %head, <vscale x 16 x half> poison, <vscale x 16 x i32> zeroinitializer
+  %vd = fpext <vscale x 16 x half> %vb to <vscale x 16 x float>
+  %ve = fpext <vscale x 16 x half> %splat to <vscale x 16 x float>
+  %vf = fneg <vscale x 16 x float> %vd
+  %vg = call <vscale x 16 x float> @llvm.experimental.constrained.fma.v16f32(<vscale x 16 x float> %vf, <vscale x 16 x float> %ve, <vscale x 16 x float> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x float> %vg
+}
+
+define <vscale x 16 x float> @vfwnmsac_fv_nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x half> %vb, half %c) {
+; CHECK-LABEL: vfwnmsac_fv_nxv16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e16, m4, ta, ma
+; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v16
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 16 x half> poison, half %c, i32 0
+  %splat = shufflevector <vscale x 16 x half> %head, <vscale x 16 x half> poison, <vscale x 16 x i32> zeroinitializer
+  %vd = fpext <vscale x 16 x half> %vb to <vscale x 16 x float>
+  %ve = fpext <vscale x 16 x half> %splat to <vscale x 16 x float>
+  %vf = fneg <vscale x 16 x float> %ve
+  %vg = call <vscale x 16 x float> @llvm.experimental.constrained.fma.v16f32(<vscale x 16 x float> %vd, <vscale x 16 x float> %vf, <vscale x 16 x float> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 16 x float> %vg
+}
+
+declare <vscale x 1 x double> @llvm.experimental.constrained.fma.v1f64(<vscale x 1 x double>, <vscale x 1 x double>, <vscale x 1 x double>, metadata, metadata)
+
+define <vscale x 1 x double> @vfwmacc_vv_nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x float> %vb, <vscale x 1 x float> %vc) {
+; CHECK-LABEL: vfwmacc_vv_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfwmacc.vv v8, v9, v10
+; CHECK-NEXT:    ret
+  %vd = fpext <vscale x 1 x float> %vb to <vscale x 1 x double>
+  %ve = fpext <vscale x 1 x float> %vc to <vscale x 1 x double>
+  %vf = call <vscale x 1 x double> @llvm.experimental.constrained.fma.v1f64(<vscale x 1 x double> %vd, <vscale x 1 x double> %ve, <vscale x 1 x double> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x double> %vf
+}
+
+define <vscale x 1 x double> @vfwmacc_vf_nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x float> %vb, float %c) {
+; CHECK-LABEL: vfwmacc_vf_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfwmacc.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 1 x float> %head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
+  %vd = fpext <vscale x 1 x float> %vb to <vscale x 1 x double>
+  %ve = fpext <vscale x 1 x float> %splat to <vscale x 1 x double>
+  %vf = call <vscale x 1 x double> @llvm.experimental.constrained.fma.v1f64(<vscale x 1 x double> %vd, <vscale x 1 x double> %ve, <vscale x 1 x double> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x double> %vf
+}
+
+define <vscale x 1 x double> @vfwnmacc_vv_nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x float> %vb, <vscale x 1 x float> %vc) {
+; CHECK-LABEL: vfwnmacc_vv_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfwnmacc.vv v8, v9, v10
+; CHECK-NEXT:    ret
+  %vd = fpext <vscale x 1 x float> %vb to <vscale x 1 x double>
+  %ve = fpext <vscale x 1 x float> %vc to <vscale x 1 x double>
+  %vf = fneg <vscale x 1 x double> %va
+  %vg = fneg <vscale x 1 x double> %vd
+  %vh = call <vscale x 1 x double> @llvm.experimental.constrained.fma.v1f64(<vscale x 1 x double> %vg, <vscale x 1 x double> %ve, <vscale x 1 x double> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x double> %vh
+}
+
+define <vscale x 1 x double> @vfwnmacc_vf_nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x float> %vb, float %c) {
+; CHECK-LABEL: vfwnmacc_vf_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 1 x float> %head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
+  %vd = fpext <vscale x 1 x float> %vb to <vscale x 1 x double>
+  %ve = fpext <vscale x 1 x float> %splat to <vscale x 1 x double>
+  %vf = fneg <vscale x 1 x double> %va
+  %vg = fneg <vscale x 1 x double> %vd
+  %vh = call <vscale x 1 x double> @llvm.experimental.constrained.fma.v1f64(<vscale x 1 x double> %vg, <vscale x 1 x double> %ve, <vscale x 1 x double> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x double> %vh
+}
+
+define <vscale x 1 x double> @vfwnmacc_fv_nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x float> %vb, float %c) {
+; CHECK-LABEL: vfwnmacc_fv_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 1 x float> %head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
+  %vd = fpext <vscale x 1 x float> %vb to <vscale x 1 x double>
+  %ve = fpext <vscale x 1 x float> %splat to <vscale x 1 x double>
+  %vf = fneg <vscale x 1 x double> %va
+  %vg = fneg <vscale x 1 x double> %ve
+  %vh = call <vscale x 1 x double> @llvm.experimental.constrained.fma.v1f64(<vscale x 1 x double> %vd, <vscale x 1 x double> %vg, <vscale x 1 x double> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x double> %vh
+}
+
+define <vscale x 1 x double> @vfwmsac_vv_nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x float> %vb, <vscale x 1 x float> %vc) {
+; CHECK-LABEL: vfwmsac_vv_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfwmsac.vv v8, v9, v10
+; CHECK-NEXT:    ret
+  %vd = fpext <vscale x 1 x float> %vb to <vscale x 1 x double>
+  %ve = fpext <vscale x 1 x float> %vc to <vscale x 1 x double>
+  %vf = fneg <vscale x 1 x double> %va
+  %vg = call <vscale x 1 x double> @llvm.experimental.constrained.fma.v1f64(<vscale x 1 x double> %vd, <vscale x 1 x double> %ve, <vscale x 1 x double> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x double> %vg
+}
+
+define <vscale x 1 x double> @vfwmsac_vf_nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x float> %vb, float %c) {
+; CHECK-LABEL: vfwmsac_vf_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfwmsac.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 1 x float> %head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
+  %vd = fpext <vscale x 1 x float> %vb to <vscale x 1 x double>
+  %ve = fpext <vscale x 1 x float> %splat to <vscale x 1 x double>
+  %vf = fneg <vscale x 1 x double> %va
+  %vg = call <vscale x 1 x double> @llvm.experimental.constrained.fma.v1f64(<vscale x 1 x double> %vd, <vscale x 1 x double> %ve, <vscale x 1 x double> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x double> %vg
+}
+
+define <vscale x 1 x double> @vfwnmsac_vv_nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x float> %vb, <vscale x 1 x float> %vc) {
+; CHECK-LABEL: vfwnmsac_vv_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfwnmsac.vv v8, v9, v10
+; CHECK-NEXT:    ret
+  %vd = fpext <vscale x 1 x float> %vb to <vscale x 1 x double>
+  %ve = fpext <vscale x 1 x float> %vc to <vscale x 1 x double>
+  %vf = fneg <vscale x 1 x double> %vd
+  %vg = call <vscale x 1 x double> @llvm.experimental.constrained.fma.v1f64(<vscale x 1 x double> %vf, <vscale x 1 x double> %ve, <vscale x 1 x double> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x double> %vg
+}
+
+define <vscale x 1 x double> @vfwnmsac_vf_nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x float> %vb, float %c) {
+; CHECK-LABEL: vfwnmsac_vf_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 1 x float> %head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
+  %vd = fpext <vscale x 1 x float> %vb to <vscale x 1 x double>
+  %ve = fpext <vscale x 1 x float> %splat to <vscale x 1 x double>
+  %vf = fneg <vscale x 1 x double> %vd
+  %vg = call <vscale x 1 x double> @llvm.experimental.constrained.fma.v1f64(<vscale x 1 x double> %vf, <vscale x 1 x double> %ve, <vscale x 1 x double> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x double> %vg
+}
+
+define <vscale x 1 x double> @vfwnmsac_fv_nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x float> %vb, float %c) {
+; CHECK-LABEL: vfwnmsac_fv_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 1 x float> %head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
+  %vd = fpext <vscale x 1 x float> %vb to <vscale x 1 x double>
+  %ve = fpext <vscale x 1 x float> %splat to <vscale x 1 x double>
+  %vf = fneg <vscale x 1 x double> %ve
+  %vg = call <vscale x 1 x double> @llvm.experimental.constrained.fma.v1f64(<vscale x 1 x double> %vd, <vscale x 1 x double> %vf, <vscale x 1 x double> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x double> %vg
+}
+
+declare <vscale x 2 x double> @llvm.experimental.constrained.fma.v2f64(<vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, metadata, metadata)
+
+define <vscale x 2 x double> @vfwmacc_vv_nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x float> %vb, <vscale x 2 x float> %vc) {
+; CHECK-LABEL: vfwmacc_vv_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vfwmacc.vv v8, v10, v11
+; CHECK-NEXT:    ret
+  %vd = fpext <vscale x 2 x float> %vb to <vscale x 2 x double>
+  %ve = fpext <vscale x 2 x float> %vc to <vscale x 2 x double>
+  %vf = call <vscale x 2 x double> @llvm.experimental.constrained.fma.v2f64(<vscale x 2 x double> %vd, <vscale x 2 x double> %ve, <vscale x 2 x double> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x double> %vf
+}
+
+define <vscale x 2 x double> @vfwmacc_vf_nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x float> %vb, float %c) {
+; CHECK-LABEL: vfwmacc_vf_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vfwmacc.vf v8, fa0, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 2 x float> %head, <vscale x 2 x float> poison, <vscale x 2 x i32> zeroinitializer
+  %vd = fpext <vscale x 2 x float> %vb to <vscale x 2 x double>
+  %ve = fpext <vscale x 2 x float> %splat to <vscale x 2 x double>
+  %vf = call <vscale x 2 x double> @llvm.experimental.constrained.fma.v2f64(<vscale x 2 x double> %vd, <vscale x 2 x double> %ve, <vscale x 2 x double> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x double> %vf
+}
+
+define <vscale x 2 x double> @vfwnmacc_vv_nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x float> %vb, <vscale x 2 x float> %vc) {
+; CHECK-LABEL: vfwnmacc_vv_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vfwnmacc.vv v8, v10, v11
+; CHECK-NEXT:    ret
+  %vd = fpext <vscale x 2 x float> %vb to <vscale x 2 x double>
+  %ve = fpext <vscale x 2 x float> %vc to <vscale x 2 x double>
+  %vf = fneg <vscale x 2 x double> %va
+  %vg = fneg <vscale x 2 x double> %vd
+  %vh = call <vscale x 2 x double> @llvm.experimental.constrained.fma.v2f64(<vscale x 2 x double> %vg, <vscale x 2 x double> %ve, <vscale x 2 x double> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x double> %vh
+}
+
+define <vscale x 2 x double> @vfwnmacc_vf_nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x float> %vb, float %c) {
+; CHECK-LABEL: vfwnmacc_vf_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 2 x float> %head, <vscale x 2 x float> poison, <vscale x 2 x i32> zeroinitializer
+  %vd = fpext <vscale x 2 x float> %vb to <vscale x 2 x double>
+  %ve = fpext <vscale x 2 x float> %splat to <vscale x 2 x double>
+  %vf = fneg <vscale x 2 x double> %va
+  %vg = fneg <vscale x 2 x double> %vd
+  %vh = call <vscale x 2 x double> @llvm.experimental.constrained.fma.v2f64(<vscale x 2 x double> %vg, <vscale x 2 x double> %ve, <vscale x 2 x double> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x double> %vh
+}
+
+define <vscale x 2 x double> @vfwnmacc_fv_nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x float> %vb, float %c) {
+; CHECK-LABEL: vfwnmacc_fv_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 2 x float> %head, <vscale x 2 x float> poison, <vscale x 2 x i32> zeroinitializer
+  %vd = fpext <vscale x 2 x float> %vb to <vscale x 2 x double>
+  %ve = fpext <vscale x 2 x float> %splat to <vscale x 2 x double>
+  %vf = fneg <vscale x 2 x double> %va
+  %vg = fneg <vscale x 2 x double> %ve
+  %vh = call <vscale x 2 x double> @llvm.experimental.constrained.fma.v2f64(<vscale x 2 x double> %vd, <vscale x 2 x double> %vg, <vscale x 2 x double> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x double> %vh
+}
+
+define <vscale x 2 x double> @vfwmsac_vv_nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x float> %vb, <vscale x 2 x float> %vc) {
+; CHECK-LABEL: vfwmsac_vv_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vfwmsac.vv v8, v10, v11
+; CHECK-NEXT:    ret
+  %vd = fpext <vscale x 2 x float> %vb to <vscale x 2 x double>
+  %ve = fpext <vscale x 2 x float> %vc to <vscale x 2 x double>
+  %vf = fneg <vscale x 2 x double> %va
+  %vg = call <vscale x 2 x double> @llvm.experimental.constrained.fma.v2f64(<vscale x 2 x double> %vd, <vscale x 2 x double> %ve, <vscale x 2 x double> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x double> %vg
+}
+
+define <vscale x 2 x double> @vfwmsac_vf_nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x float> %vb, float %c) {
+; CHECK-LABEL: vfwmsac_vf_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vfwmsac.vf v8, fa0, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 2 x float> %head, <vscale x 2 x float> poison, <vscale x 2 x i32> zeroinitializer
+  %vd = fpext <vscale x 2 x float> %vb to <vscale x 2 x double>
+  %ve = fpext <vscale x 2 x float> %splat to <vscale x 2 x double>
+  %vf = fneg <vscale x 2 x double> %va
+  %vg = call <vscale x 2 x double> @llvm.experimental.constrained.fma.v2f64(<vscale x 2 x double> %vd, <vscale x 2 x double> %ve, <vscale x 2 x double> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x double> %vg
+}
+
+define <vscale x 2 x double> @vfwnmsac_vv_nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x float> %vb, <vscale x 2 x float> %vc) {
+; CHECK-LABEL: vfwnmsac_vv_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vfwnmsac.vv v8, v10, v11
+; CHECK-NEXT:    ret
+  %vd = fpext <vscale x 2 x float> %vb to <vscale x 2 x double>
+  %ve = fpext <vscale x 2 x float> %vc to <vscale x 2 x double>
+  %vf = fneg <vscale x 2 x double> %vd
+  %vg = call <vscale x 2 x double> @llvm.experimental.constrained.fma.v2f64(<vscale x 2 x double> %vf, <vscale x 2 x double> %ve, <vscale x 2 x double> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x double> %vg
+}
+
+define <vscale x 2 x double> @vfwnmsac_vf_nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x float> %vb, float %c) {
+; CHECK-LABEL: vfwnmsac_vf_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 2 x float> %head, <vscale x 2 x float> poison, <vscale x 2 x i32> zeroinitializer
+  %vd = fpext <vscale x 2 x float> %vb to <vscale x 2 x double>
+  %ve = fpext <vscale x 2 x float> %splat to <vscale x 2 x double>
+  %vf = fneg <vscale x 2 x double> %vd
+  %vg = call <vscale x 2 x double> @llvm.experimental.constrained.fma.v2f64(<vscale x 2 x double> %vf, <vscale x 2 x double> %ve, <vscale x 2 x double> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x double> %vg
+}
+
+define <vscale x 2 x double> @vfwnmsac_fv_nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x float> %vb, float %c) {
+; CHECK-LABEL: vfwnmsac_fv_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 2 x float> %head, <vscale x 2 x float> poison, <vscale x 2 x i32> zeroinitializer
+  %vd = fpext <vscale x 2 x float> %vb to <vscale x 2 x double>
+  %ve = fpext <vscale x 2 x float> %splat to <vscale x 2 x double>
+  %vf = fneg <vscale x 2 x double> %ve
+  %vg = call <vscale x 2 x double> @llvm.experimental.constrained.fma.v2f64(<vscale x 2 x double> %vd, <vscale x 2 x double> %vf, <vscale x 2 x double> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x double> %vg
+}
+
+
+declare <vscale x 4 x double> @llvm.experimental.constrained.fma.v4f64(<vscale x 4 x double>, <vscale x 4 x double>, <vscale x 4 x double>, metadata, metadata)
+
+define <vscale x 4 x double> @vfwmacc_vv_nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x float> %vb, <vscale x 4 x float> %vc) {
+; CHECK-LABEL: vfwmacc_vv_nxv4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
+; CHECK-NEXT:    vfwmacc.vv v8, v12, v14
+; CHECK-NEXT:    ret
+  %vd = fpext <vscale x 4 x float> %vb to <vscale x 4 x double>
+  %ve = fpext <vscale x 4 x float> %vc to <vscale x 4 x double>
+  %vf = call <vscale x 4 x double> @llvm.experimental.constrained.fma.v4f64(<vscale x 4 x double> %vd, <vscale x 4 x double> %ve, <vscale x 4 x double> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x double> %vf
+}
+
+define <vscale x 4 x double> @vfwmacc_vf_nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x float> %vb, float %c) {
+; CHECK-LABEL: vfwmacc_vf_nxv4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
+; CHECK-NEXT:    vfwmacc.vf v8, fa0, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 4 x float> %head, <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
+  %vd = fpext <vscale x 4 x float> %vb to <vscale x 4 x double>
+  %ve = fpext <vscale x 4 x float> %splat to <vscale x 4 x double>
+  %vf = call <vscale x 4 x double> @llvm.experimental.constrained.fma.v4f64(<vscale x 4 x double> %vd, <vscale x 4 x double> %ve, <vscale x 4 x double> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x double> %vf
+}
+
+define <vscale x 4 x double> @vfwnmacc_vv_nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x float> %vb, <vscale x 4 x float> %vc) {
+; CHECK-LABEL: vfwnmacc_vv_nxv4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
+; CHECK-NEXT:    vfwnmacc.vv v8, v12, v14
+; CHECK-NEXT:    ret
+  %vd = fpext <vscale x 4 x float> %vb to <vscale x 4 x double>
+  %ve = fpext <vscale x 4 x float> %vc to <vscale x 4 x double>
+  %vf = fneg <vscale x 4 x double> %va
+  %vg = fneg <vscale x 4 x double> %vd
+  %vh = call <vscale x 4 x double> @llvm.experimental.constrained.fma.v4f64(<vscale x 4 x double> %vg, <vscale x 4 x double> %ve, <vscale x 4 x double> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x double> %vh
+}
+
+define <vscale x 4 x double> @vfwnmacc_vf_nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x float> %vb, float %c) {
+; CHECK-LABEL: vfwnmacc_vf_nxv4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
+; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 4 x float> %head, <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
+  %vd = fpext <vscale x 4 x float> %vb to <vscale x 4 x double>
+  %ve = fpext <vscale x 4 x float> %splat to <vscale x 4 x double>
+  %vf = fneg <vscale x 4 x double> %va
+  %vg = fneg <vscale x 4 x double> %vd
+  %vh = call <vscale x 4 x double> @llvm.experimental.constrained.fma.v4f64(<vscale x 4 x double> %vg, <vscale x 4 x double> %ve, <vscale x 4 x double> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x double> %vh
+}
+
+define <vscale x 4 x double> @vfwnmacc_fv_nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x float> %vb, float %c) {
+; CHECK-LABEL: vfwnmacc_fv_nxv4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
+; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 4 x float> %head, <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
+  %vd = fpext <vscale x 4 x float> %vb to <vscale x 4 x double>
+  %ve = fpext <vscale x 4 x float> %splat to <vscale x 4 x double>
+  %vf = fneg <vscale x 4 x double> %va
+  %vg = fneg <vscale x 4 x double> %ve
+  %vh = call <vscale x 4 x double> @llvm.experimental.constrained.fma.v4f64(<vscale x 4 x double> %vd, <vscale x 4 x double> %vg, <vscale x 4 x double> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x double> %vh
+}
+
+define <vscale x 4 x double> @vfwmsac_vv_nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x float> %vb, <vscale x 4 x float> %vc) {
+; CHECK-LABEL: vfwmsac_vv_nxv4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
+; CHECK-NEXT:    vfwmsac.vv v8, v12, v14
+; CHECK-NEXT:    ret
+  %vd = fpext <vscale x 4 x float> %vb to <vscale x 4 x double>
+  %ve = fpext <vscale x 4 x float> %vc to <vscale x 4 x double>
+  %vf = fneg <vscale x 4 x double> %va
+  %vg = call <vscale x 4 x double> @llvm.experimental.constrained.fma.v4f64(<vscale x 4 x double> %vd, <vscale x 4 x double> %ve, <vscale x 4 x double> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x double> %vg
+}
+
+define <vscale x 4 x double> @vfwmsac_vf_nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x float> %vb, float %c) {
+; CHECK-LABEL: vfwmsac_vf_nxv4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
+; CHECK-NEXT:    vfwmsac.vf v8, fa0, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 4 x float> %head, <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
+  %vd = fpext <vscale x 4 x float> %vb to <vscale x 4 x double>
+  %ve = fpext <vscale x 4 x float> %splat to <vscale x 4 x double>
+  %vf = fneg <vscale x 4 x double> %va
+  %vg = call <vscale x 4 x double> @llvm.experimental.constrained.fma.v4f64(<vscale x 4 x double> %vd, <vscale x 4 x double> %ve, <vscale x 4 x double> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x double> %vg
+}
+
+define <vscale x 4 x double> @vfwnmsac_vv_nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x float> %vb, <vscale x 4 x float> %vc) {
+; CHECK-LABEL: vfwnmsac_vv_nxv4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
+; CHECK-NEXT:    vfwnmsac.vv v8, v12, v14
+; CHECK-NEXT:    ret
+  %vd = fpext <vscale x 4 x float> %vb to <vscale x 4 x double>
+  %ve = fpext <vscale x 4 x float> %vc to <vscale x 4 x double>
+  %vf = fneg <vscale x 4 x double> %vd
+  %vg = call <vscale x 4 x double> @llvm.experimental.constrained.fma.v4f64(<vscale x 4 x double> %vf, <vscale x 4 x double> %ve, <vscale x 4 x double> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x double> %vg
+}
+
+define <vscale x 4 x double> @vfwnmsac_vf_nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x float> %vb, float %c) {
+; CHECK-LABEL: vfwnmsac_vf_nxv4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
+; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 4 x float> %head, <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
+  %vd = fpext <vscale x 4 x float> %vb to <vscale x 4 x double>
+  %ve = fpext <vscale x 4 x float> %splat to <vscale x 4 x double>
+  %vf = fneg <vscale x 4 x double> %vd
+  %vg = call <vscale x 4 x double> @llvm.experimental.constrained.fma.v4f64(<vscale x 4 x double> %vf, <vscale x 4 x double> %ve, <vscale x 4 x double> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x double> %vg
+}
+
+define <vscale x 4 x double> @vfwnmsac_fv_nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x float> %vb, float %c) {
+; CHECK-LABEL: vfwnmsac_fv_nxv4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
+; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 4 x float> %head, <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
+  %vd = fpext <vscale x 4 x float> %vb to <vscale x 4 x double>
+  %ve = fpext <vscale x 4 x float> %splat to <vscale x 4 x double>
+  %vf = fneg <vscale x 4 x double> %ve
+  %vg = call <vscale x 4 x double> @llvm.experimental.constrained.fma.v4f64(<vscale x 4 x double> %vd, <vscale x 4 x double> %vf, <vscale x 4 x double> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x double> %vg
+}
+
+declare <vscale x 8 x double> @llvm.experimental.constrained.fma.v8f64(<vscale x 8 x double>, <vscale x 8 x double>, <vscale x 8 x double>, metadata, metadata)
+
+define <vscale x 8 x double> @vfwmacc_vv_nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x float> %vb, <vscale x 8 x float> %vc) {
+; CHECK-LABEL: vfwmacc_vv_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m4, ta, ma
+; CHECK-NEXT:    vfwmacc.vv v8, v16, v20
+; CHECK-NEXT:    ret
+  %vd = fpext <vscale x 8 x float> %vb to <vscale x 8 x double>
+  %ve = fpext <vscale x 8 x float> %vc to <vscale x 8 x double>
+  %vf = call <vscale x 8 x double> @llvm.experimental.constrained.fma.v8f64(<vscale x 8 x double> %vd, <vscale x 8 x double> %ve, <vscale x 8 x double> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x double> %vf
+}
+
+define <vscale x 8 x double> @vfwmacc_vf_nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x float> %vb, float %c) {
+; CHECK-LABEL: vfwmacc_vf_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m4, ta, ma
+; CHECK-NEXT:    vfwmacc.vf v8, fa0, v16
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 8 x float> %head, <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer
+  %vd = fpext <vscale x 8 x float> %vb to <vscale x 8 x double>
+  %ve = fpext <vscale x 8 x float> %splat to <vscale x 8 x double>
+  %vf = call <vscale x 8 x double> @llvm.experimental.constrained.fma.v8f64(<vscale x 8 x double> %vd, <vscale x 8 x double> %ve, <vscale x 8 x double> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x double> %vf
+}
+
+define <vscale x 8 x double> @vfwnmacc_vv_nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x float> %vb, <vscale x 8 x float> %vc) {
+; CHECK-LABEL: vfwnmacc_vv_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m4, ta, ma
+; CHECK-NEXT:    vfwnmacc.vv v8, v16, v20
+; CHECK-NEXT:    ret
+  %vd = fpext <vscale x 8 x float> %vb to <vscale x 8 x double>
+  %ve = fpext <vscale x 8 x float> %vc to <vscale x 8 x double>
+  %vf = fneg <vscale x 8 x double> %va
+  %vg = fneg <vscale x 8 x double> %vd
+  %vh = call <vscale x 8 x double> @llvm.experimental.constrained.fma.v8f64(<vscale x 8 x double> %vg, <vscale x 8 x double> %ve, <vscale x 8 x double> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x double> %vh
+}
+
+define <vscale x 8 x double> @vfwnmacc_vf_nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x float> %vb, float %c) {
+; CHECK-LABEL: vfwnmacc_vf_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m4, ta, ma
+; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v16
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 8 x float> %head, <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer
+  %vd = fpext <vscale x 8 x float> %vb to <vscale x 8 x double>
+  %ve = fpext <vscale x 8 x float> %splat to <vscale x 8 x double>
+  %vf = fneg <vscale x 8 x double> %va
+  %vg = fneg <vscale x 8 x double> %vd
+  %vh = call <vscale x 8 x double> @llvm.experimental.constrained.fma.v8f64(<vscale x 8 x double> %vg, <vscale x 8 x double> %ve, <vscale x 8 x double> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x double> %vh
+}
+
+define <vscale x 8 x double> @vfwnmacc_fv_nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x float> %vb, float %c) {
+; CHECK-LABEL: vfwnmacc_fv_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m4, ta, ma
+; CHECK-NEXT:    vfwnmacc.vf v8, fa0, v16
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 8 x float> %head, <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer
+  %vd = fpext <vscale x 8 x float> %vb to <vscale x 8 x double>
+  %ve = fpext <vscale x 8 x float> %splat to <vscale x 8 x double>
+  %vf = fneg <vscale x 8 x double> %va
+  %vg = fneg <vscale x 8 x double> %ve
+  %vh = call <vscale x 8 x double> @llvm.experimental.constrained.fma.v8f64(<vscale x 8 x double> %vd, <vscale x 8 x double> %vg, <vscale x 8 x double> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x double> %vh
+}
+
+define <vscale x 8 x double> @vfwmsac_vv_nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x float> %vb, <vscale x 8 x float> %vc) {
+; CHECK-LABEL: vfwmsac_vv_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m4, ta, ma
+; CHECK-NEXT:    vfwmsac.vv v8, v16, v20
+; CHECK-NEXT:    ret
+  %vd = fpext <vscale x 8 x float> %vb to <vscale x 8 x double>
+  %ve = fpext <vscale x 8 x float> %vc to <vscale x 8 x double>
+  %vf = fneg <vscale x 8 x double> %va
+  %vg = call <vscale x 8 x double> @llvm.experimental.constrained.fma.v8f64(<vscale x 8 x double> %vd, <vscale x 8 x double> %ve, <vscale x 8 x double> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x double> %vg
+}
+
+define <vscale x 8 x double> @vfwmsac_vf_nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x float> %vb, float %c) {
+; CHECK-LABEL: vfwmsac_vf_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m4, ta, ma
+; CHECK-NEXT:    vfwmsac.vf v8, fa0, v16
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 8 x float> %head, <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer
+  %vd = fpext <vscale x 8 x float> %vb to <vscale x 8 x double>
+  %ve = fpext <vscale x 8 x float> %splat to <vscale x 8 x double>
+  %vf = fneg <vscale x 8 x double> %va
+  %vg = call <vscale x 8 x double> @llvm.experimental.constrained.fma.v8f64(<vscale x 8 x double> %vd, <vscale x 8 x double> %ve, <vscale x 8 x double> %vf,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x double> %vg
+}
+
+define <vscale x 8 x double> @vfwnmsac_vv_nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x float> %vb, <vscale x 8 x float> %vc) {
+; CHECK-LABEL: vfwnmsac_vv_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m4, ta, ma
+; CHECK-NEXT:    vfwnmsac.vv v8, v16, v20
+; CHECK-NEXT:    ret
+  %vd = fpext <vscale x 8 x float> %vb to <vscale x 8 x double>
+  %ve = fpext <vscale x 8 x float> %vc to <vscale x 8 x double>
+  %vf = fneg <vscale x 8 x double> %vd
+  %vg = call <vscale x 8 x double> @llvm.experimental.constrained.fma.v8f64(<vscale x 8 x double> %vf, <vscale x 8 x double> %ve, <vscale x 8 x double> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x double> %vg
+}
+
+define <vscale x 8 x double> @vfwnmsac_vf_nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x float> %vb, float %c) {
+; CHECK-LABEL: vfwnmsac_vf_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m4, ta, ma
+; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v16
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 8 x float> %head, <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer
+  %vd = fpext <vscale x 8 x float> %vb to <vscale x 8 x double>
+  %ve = fpext <vscale x 8 x float> %splat to <vscale x 8 x double>
+  %vf = fneg <vscale x 8 x double> %vd
+  %vg = call <vscale x 8 x double> @llvm.experimental.constrained.fma.v8f64(<vscale x 8 x double> %vf, <vscale x 8 x double> %ve, <vscale x 8 x double> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x double> %vg
+}
+
+define <vscale x 8 x double> @vfwnmsac_fv_nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x float> %vb, float %c) {
+; CHECK-LABEL: vfwnmsac_fv_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m4, ta, ma
+; CHECK-NEXT:    vfwnmsac.vf v8, fa0, v16
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x float> poison, float %c, i32 0
+  %splat = shufflevector <vscale x 8 x float> %head, <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer
+  %vd = fpext <vscale x 8 x float> %vb to <vscale x 8 x double>
+  %ve = fpext <vscale x 8 x float> %splat to <vscale x 8 x double>
+  %vf = fneg <vscale x 8 x double> %ve
+  %vg = call <vscale x 8 x double> @llvm.experimental.constrained.fma.v8f64(<vscale x 8 x double> %vd, <vscale x 8 x double> %vf, <vscale x 8 x double> %va,metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x double> %vg
+}

llvm/test/CodeGen/RISCV/rvv/vfwmul-constrained-sdnode.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+experimental-zvfh,+v -target-abi=ilp32d \
+; RUN:     -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+experimental-zvfh,+v -target-abi=lp64d \
+; RUN:     -verify-machineinstrs < %s | FileCheck %s
+
+declare <vscale x 1 x double> @llvm.experimental.constrained.fmul.nxv1f64(<vscale x 1 x double>, <vscale x 1 x double>, metadata, metadata)
+define <vscale x 1 x double> @vfwmul_vv_nxv1f64(<vscale x 1 x float> %va, <vscale x 1 x float> %vb) {
+; CHECK-LABEL: vfwmul_vv_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfwmul.vv v10, v8, v9
+; CHECK-NEXT:    vmv1r.v v8, v10
+; CHECK-NEXT:    ret
+  %vc = fpext <vscale x 1 x float> %va to <vscale x 1 x double>
+  %vd = fpext <vscale x 1 x float> %vb to <vscale x 1 x double>
+  %ve =call <vscale x 1 x double> @llvm.experimental.constrained.fmul.nxv1f64(<vscale x 1 x double> %vc, <vscale x 1 x double> %vd,  metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x double> %ve
+}
+
+define <vscale x 1 x double> @vfwmul_vf_nxv1f64(<vscale x 1 x float> %va, float %b) {
+; CHECK-LABEL: vfwmul_vf_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfwmul.vf v9, v8, fa0
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x float> poison, float %b, i32 0
+  %splat = shufflevector <vscale x 1 x float> %head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
+  %vc = fpext <vscale x 1 x float> %va to <vscale x 1 x double>
+  %vd = fpext <vscale x 1 x float> %splat to <vscale x 1 x double>
+  %ve =call <vscale x 1 x double> @llvm.experimental.constrained.fmul.nxv1f64(<vscale x 1 x double> %vc, <vscale x 1 x double> %vd,  metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x double> %ve
+}
+
+define <vscale x 1 x double> @vfwmul_vf_nxv1f64_2(<vscale x 1 x float> %va, float %b) {
+; CHECK-LABEL: vfwmul_vf_nxv1f64_2:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfwmul.vf v9, v8, fa0
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+  %fpext = fpext float %b to double
+  %head = insertelement <vscale x 1 x double> poison, double %fpext, i32 0
+  %splat = shufflevector <vscale x 1 x double> %head, <vscale x 1 x double> poison, <vscale x 1 x i32> zeroinitializer
+  %vc = fpext <vscale x 1 x float> %va to <vscale x 1 x double>
+  %ve =call <vscale x 1 x double> @llvm.experimental.constrained.fmul.nxv1f64(<vscale x 1 x double> %vc, <vscale x 1 x double> %splat,  metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x double> %ve
+}
+
+declare <vscale x 2 x double> @llvm.experimental.constrained.fmul.nxv2f64(<vscale x 2 x double>, <vscale x 2 x double>, metadata, metadata)
+define <vscale x 2 x double> @vfwmul_vv_nxv2f64(<vscale x 2 x float> %va, <vscale x 2 x float> %vb) {
+; CHECK-LABEL: vfwmul_vv_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vfwmul.vv v10, v8, v9
+; CHECK-NEXT:    vmv2r.v v8, v10
+; CHECK-NEXT:    ret
+  %vc = fpext <vscale x 2 x float> %va to <vscale x 2 x double>
+  %vd = fpext <vscale x 2 x float> %vb to <vscale x 2 x double>
+  %ve =call <vscale x 2 x double> @llvm.experimental.constrained.fmul.nxv2f64(<vscale x 2 x double> %vc, <vscale x 2 x double> %vd,  metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x double> %ve
+}
+
+define <vscale x 2 x double> @vfwmul_vf_nxv2f64(<vscale x 2 x float> %va, float %b) {
+; CHECK-LABEL: vfwmul_vf_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vfwmul.vf v10, v8, fa0
+; CHECK-NEXT:    vmv2r.v v8, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x float> poison, float %b, i32 0
+  %splat = shufflevector <vscale x 2 x float> %head, <vscale x 2 x float> poison, <vscale x 2 x i32> zeroinitializer
+  %vc = fpext <vscale x 2 x float> %va to <vscale x 2 x double>
+  %vd = fpext <vscale x 2 x float> %splat to <vscale x 2 x double>
+  %ve =call <vscale x 2 x double> @llvm.experimental.constrained.fmul.nxv2f64(<vscale x 2 x double> %vc, <vscale x 2 x double> %vd,  metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x double> %ve
+}
+
+define <vscale x 2 x double> @vfwmul_vf_nxv2f64_2(<vscale x 2 x float> %va, float %b) {
+; CHECK-LABEL: vfwmul_vf_nxv2f64_2:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vfwmul.vf v10, v8, fa0
+; CHECK-NEXT:    vmv2r.v v8, v10
+; CHECK-NEXT:    ret
+  %fpext = fpext float %b to double
+  %head = insertelement <vscale x 2 x double> poison, double %fpext, i32 0
+  %splat = shufflevector <vscale x 2 x double> %head, <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
+  %vc = fpext <vscale x 2 x float> %va to <vscale x 2 x double>
+  %ve =call <vscale x 2 x double> @llvm.experimental.constrained.fmul.nxv2f64(<vscale x 2 x double> %vc, <vscale x 2 x double> %splat,  metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 2 x double> %ve
+}
+
+declare <vscale x 4 x double> @llvm.experimental.constrained.fmul.nxv4f64(<vscale x 4 x double>, <vscale x 4 x double>, metadata, metadata)
+define <vscale x 4 x double> @vfwmul_vv_nxv4f64(<vscale x 4 x float> %va, <vscale x 4 x float> %vb) {
+; CHECK-LABEL: vfwmul_vv_nxv4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
+; CHECK-NEXT:    vfwmul.vv v12, v8, v10
+; CHECK-NEXT:    vmv4r.v v8, v12
+; CHECK-NEXT:    ret
+  %vc = fpext <vscale x 4 x float> %va to <vscale x 4 x double>
+  %vd = fpext <vscale x 4 x float> %vb to <vscale x 4 x double>
+  %ve =call <vscale x 4 x double> @llvm.experimental.constrained.fmul.nxv4f64(<vscale x 4 x double> %vc, <vscale x 4 x double> %vd,  metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x double> %ve
+}
+
+define <vscale x 4 x double> @vfwmul_vf_nxv4f64(<vscale x 4 x float> %va, float %b) {
+; CHECK-LABEL: vfwmul_vf_nxv4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
+; CHECK-NEXT:    vfwmul.vf v12, v8, fa0
+; CHECK-NEXT:    vmv4r.v v8, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x float> poison, float %b, i32 0
+  %splat = shufflevector <vscale x 4 x float> %head, <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
+  %vc = fpext <vscale x 4 x float> %va to <vscale x 4 x double>
+  %vd = fpext <vscale x 4 x float> %splat to <vscale x 4 x double>
+  %ve =call <vscale x 4 x double> @llvm.experimental.constrained.fmul.nxv4f64(<vscale x 4 x double> %vc, <vscale x 4 x double> %vd,  metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x double> %ve
+}
+
+define <vscale x 4 x double> @vfwmul_vf_nxv4f64_2(<vscale x 4 x float> %va, float %b) {
+; CHECK-LABEL: vfwmul_vf_nxv4f64_2:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, ma
+; CHECK-NEXT:    vfwmul.vf v12, v8, fa0
+; CHECK-NEXT:    vmv4r.v v8, v12
+; CHECK-NEXT:    ret
+  %fpext = fpext float %b to double
+  %head = insertelement <vscale x 4 x double> poison, double %fpext, i32 0
+  %splat = shufflevector <vscale x 4 x double> %head, <vscale x 4 x double> poison, <vscale x 4 x i32> zeroinitializer
+  %vc = fpext <vscale x 4 x float> %va to <vscale x 4 x double>
+  %ve =call <vscale x 4 x double> @llvm.experimental.constrained.fmul.nxv4f64(<vscale x 4 x double> %vc, <vscale x 4 x double> %splat,  metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 4 x double> %ve
+}
+
+declare <vscale x 8 x double> @llvm.experimental.constrained.fmul.nxv8f64(<vscale x 8 x double>, <vscale x 8 x double>, metadata, metadata)
+define <vscale x 8 x double> @vfwmul_vv_nxv8f64(<vscale x 8 x float> %va, <vscale x 8 x float> %vb) {
+; CHECK-LABEL: vfwmul_vv_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m4, ta, ma
+; CHECK-NEXT:    vfwmul.vv v16, v8, v12
+; CHECK-NEXT:    vmv8r.v v8, v16
+; CHECK-NEXT:    ret
+  %vc = fpext <vscale x 8 x float> %va to <vscale x 8 x double>
+  %vd = fpext <vscale x 8 x float> %vb to <vscale x 8 x double>
+  %ve =call <vscale x 8 x double> @llvm.experimental.constrained.fmul.nxv8f64(<vscale x 8 x double> %vc, <vscale x 8 x double> %vd,  metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x double> %ve
+}
+
+define <vscale x 8 x double> @vfwmul_vf_nxv8f64(<vscale x 8 x float> %va, float %b) {
+; CHECK-LABEL: vfwmul_vf_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m4, ta, ma
+; CHECK-NEXT:    vfwmul.vf v16, v8, fa0
+; CHECK-NEXT:    vmv8r.v v8, v16
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x float> poison, float %b, i32 0
+  %splat = shufflevector <vscale x 8 x float> %head, <vscale x 8 x float> poison, <vscale x 8 x i32> zeroinitializer
+  %vc = fpext <vscale x 8 x float> %va to <vscale x 8 x double>
+  %vd = fpext <vscale x 8 x float> %splat to <vscale x 8 x double>
+  %ve =call <vscale x 8 x double> @llvm.experimental.constrained.fmul.nxv8f64(<vscale x 8 x double> %vc, <vscale x 8 x double> %vd,  metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x double> %ve
+}
+
+define <vscale x 8 x double> @vfwmul_vf_nxv8f64_2(<vscale x 8 x float> %va, float %b) {
+; CHECK-LABEL: vfwmul_vf_nxv8f64_2:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, m4, ta, ma
+; CHECK-NEXT:    vfwmul.vf v16, v8, fa0
+; CHECK-NEXT:    vmv8r.v v8, v16
+; CHECK-NEXT:    ret
+  %fpext = fpext float %b to double
+  %head = insertelement <vscale x 8 x double> poison, double %fpext, i32 0
+  %splat = shufflevector <vscale x 8 x double> %head, <vscale x 8 x double> poison, <vscale x 8 x i32> zeroinitializer
+  %vc = fpext <vscale x 8 x float> %va to <vscale x 8 x double>
+  %ve =call <vscale x 8 x double> @llvm.experimental.constrained.fmul.nxv8f64(<vscale x 8 x double> %vc, <vscale x 8 x double> %splat,  metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 8 x double> %ve
+}

llvm/test/CodeGen/RISCV/rvv/vfwsub-constrained-sdnode.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+experimental-zvfh,+v -target-abi=ilp32d \
+; RUN:     -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+experimental-zvfh,+v -target-abi=lp64d \
+; RUN:     -verify-machineinstrs < %s | FileCheck %s
+
+declare <vscale x 1 x double> @llvm.experimental.constrained.fsub.nxv1f64(<vscale x 1 x double>, <vscale x 1 x double>, metadata, metadata)
+define <vscale x 1 x double> @vfwsub_vv_nxv1f64(<vscale x 1 x float> %va, <vscale x 1 x float> %vb) {
+; CHECK-LABEL: vfwsub_vv_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfwsub.vv v10, v8, v9
+; CHECK-NEXT:    vmv1r.v v8, v10
+; CHECK-NEXT:    ret
+  %vc = fpext <vscale x 1 x float> %va to <vscale x 1 x double>
+  %vd = fpext <vscale x 1 x float> %vb to <vscale x 1 x double>
+  %ve = call <vscale x 1 x double> @llvm.experimental.constrained.fsub.nxv1f64(<vscale x 1 x double> %vc, <vscale x 1 x double> %vd, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x double> %ve
+}
+
+define <vscale x 1 x double> @vfwsub_vf_nxv1f64(<vscale x 1 x float> %va, float %b) {
+; CHECK-LABEL: vfwsub_vf_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfwsub.vf v9, v8, fa0
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x float> poison, float %b, i32 0
+  %splat = shufflevector <vscale x 1 x float> %head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
+  %vc = fpext <vscale x 1 x float> %va to <vscale x 1 x double>
+  %vd = fpext <vscale x 1 x float> %splat to <vscale x 1 x double>
+  %ve = call <vscale x 1 x double> @llvm.experimental.constrained.fsub.nxv1f64(<vscale x 1 x double> %vc, <vscale x 1 x double> %vd, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x double> %ve
+}
+
+define <vscale x 1 x double> @vfwsub_vf_nxv1f64_2(<vscale x 1 x float> %va, float %b) {
+; CHECK-LABEL: vfwsub_vf_nxv1f64_2:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfwsub.vf v9, v8, fa0
+; CHECK-NEXT:    vmv1r.v v8, v9
+; CHECK-NEXT:    ret
+  %fpext = fpext float %b to double
+  %head = insertelement <vscale x 1 x double> poison, double %fpext, i32 0
+  %splat = shufflevector <vscale x 1 x double> %head, <vscale x 1 x double> poison, <vscale x 1 x i32> zeroinitializer
+  %vc = fpext <vscale x 1 x float> %va to <vscale x 1 x double>
+  %ve = call <vscale x 1 x double> @llvm.experimental.constrained.fsub.nxv1f64(<vscale x 1 x double> %vc, <vscale x 1 x double> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x double> %ve
+}
+
+define <vscale x 1 x double> @vfwsub_wv_nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x float> %vb) {
+; CHECK-LABEL: vfwsub_wv_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfwsub.wv v8, v8, v9
+; CHECK-NEXT:    ret
+  %vc = fpext <vscale x 1 x float> %vb to <vscale x 1 x double>
+  %vd = call <vscale x 1 x double> @llvm.experimental.constrained.fsub.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %vc, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x double> %vd
+}
+
+define <vscale x 1 x double> @vfwsub_wf_nxv1f64(<vscale x 1 x double> %va, float %b) {
+; CHECK-LABEL: vfwsub_wf_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfwsub.wf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x float> poison, float %b, i32 0
+  %splat = shufflevector <vscale x 1 x float> %head, <vscale x 1 x float> poison, <vscale x 1 x i32> zeroinitializer
+  %vc = fpext <vscale x 1 x float> %splat to <vscale x 1 x double>
+  %vd = call <vscale x 1 x double> @llvm.experimental.constrained.fsub.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %vc, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x double> %vd
+}
+
+define <vscale x 1 x double> @vfwsub_wf_nxv1f64_2(<vscale x 1 x double> %va, float %b) {
+; CHECK-LABEL: vfwsub_wf_nxv1f64_2:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vfwsub.wf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %fpext = fpext float %b to double
+  %head = insertelement <vscale x 1 x double> poison, double %fpext, i32 0
+  %splat = shufflevector <vscale x 1 x double> %head, <vscale x 1 x double> poison, <vscale x 1 x i32> zeroinitializer
+  %vd = call <vscale x 1 x double> @llvm.experimental.constrained.fsub.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %splat, metadata !"round.dynamic", metadata !"fpexcept.ignore")
+  ret <vscale x 1 x double> %vd
+}