[RISCV][Clang] Support policy functions for vmerge, vfmerge and vcompress.

[kito-cheng]

LGTM, and verified with internal testsuite :)

[nlopes]

While at it, could you switch those UndefValue with PoisonValue if possible? Thank you!

[khchen]

In D126745#3678776, @nlopes wrote:

While at it, could you switch those UndefValue with PoisonValue if possible? Thank you!

I'm not sure UndefValue will work because the backend would only check isUndef() to generate correct code...

[craig.topper]

In D126745#3691528, @khchen wrote:

In D126745#3678776, @nlopes wrote:

While at it, could you switch those UndefValue with PoisonValue if possible? Thank you!

I'm not sure UndefValue will work because the backend would only check isUndef() to generate correct code...

PoisonValue and UndefValue are both turned into undef in SelectionDAG today.

[khchen]

In D126745#3691544, @craig.topper wrote:

In D126745#3691528, @khchen wrote:

In D126745#3678776, @nlopes wrote:

While at it, could you switch those UndefValue with PoisonValue if possible? Thank you!

I'm not sure UndefValue will work because the backend would only check isUndef() to generate correct code...

PoisonValue and UndefValue are both turned into undef in SelectionDAG today.

thanks @craig.topper, I will switch UndefValue with PoisonValue.

[khchen]

oops.. all tests need to updated and include all intrinsic IR tests...
Is there any benefit to switch UndefValue to PoisonValue?
maybe those changed should be in other follow-up patches?

[khchen]

@nlopes we will update all undef to poison in follow up patches.

[nlopes]

In D126745#3693790, @khchen wrote:

@nlopes we will update all undef to poison in follow up patches.

Thank you!
The benefit is that we want to remove undef altogether, so I'm trying to avoid having more undefs introduced, and if possible have everyone help reducing its usage whenever possible.
Thank you for your help!

clang/include/clang/Basic/riscv_vector.td

 // shape of the builtins generated. Refer to the comments below for them.
 
 class PolicyScheme<int val>{
   int Value = val;
 }
 def NonePolicy : PolicyScheme<0>;
 def HasPassthruOperand : PolicyScheme<1>;
 def HasPolicyOperand : PolicyScheme<2>;
+// Specail case for passthru operand which is not a first opeand.
+def HasPassthruOperandAtIdx1 : PolicyScheme<3>;
 
 class RVVBuiltin<string suffix, string prototype, string type_range,
                  string overloaded_suffix = ""> {
   // Base name that will be prepended in __builtin_rvv_ and appended the
   // computed Suffix.
   string Name = NAME;
 
   // If not empty, each instantiated builtin will have this appended after an
defm vwmaccsu : RVVOutOp1Op2BuiltinSet<"vwmaccsu", "csi",
                                        [["vv", "w", "wwvUv"],
                                         ["vx", "w", "wweUv"]]>;
 defm vwmaccus : RVVOutOp1Op2BuiltinSet<"vwmaccus", "csi",
                                        [["vx", "w", "wwUev"]]>;
 }
 }
 
 // 12.15. Vector Integer Merge Instructions
-// C/C++ Operand: (mask, op1, op2, vl), Intrinsic: (op1, op2, mask, vl)
-let HasMasked = false, MaskedPolicyScheme = NonePolicy,
+// C/C++ Operand: (mask, op1, op2, vl), Intrinsic: (passthru, op1, op2, mask, vl)
+let HasMasked = false,
+    UnMaskedPolicyScheme = HasPassthruOperandAtIdx1,
+    MaskedPolicyScheme = NonePolicy,
     ManualCodegen = [{
-      std::rotate(Ops.begin(), Ops.begin() + 1, Ops.begin() + 3);
-      IntrinsicTypes = {ResultType, Ops[1]->getType(), Ops[3]->getType()};
+      std::rotate(Ops.begin(), Ops.begin() + 1, Ops.end() - 1);
       // insert undef passthru
+      if (DefaultPolicy == TAIL_AGNOSTIC)
         Ops.insert(Ops.begin(), llvm::UndefValue::get(ResultType));
+      IntrinsicTypes = {ResultType, Ops[2]->getType(), Ops.back()->getType()};
     }] in {
   defm vmerge : RVVOutOp1BuiltinSet<"vmerge", "csil",
                                     [["vvm", "v", "vmvv"],
                                      ["vxm", "v", "vmve"],
                                      ["vvm", "Uv", "UvmUvUv"],
                                      ["vxm", "Uv", "UvmUvUe"]]>;
 }
 

 }
 
 // 14.14. Vector Floating-Point Classify Instruction
 let Name = "vfclass_v", UnMaskedPolicyScheme = HasPassthruOperand in
   def vfclass : RVVOp0Builtin<"Uv", "Uvv", "xfd">;
 
 // 14.15. Vector Floating-Point Merge Instructio
 // C/C++ Operand: (mask, op1, op2, vl), Builtin: (op1, op2, mask, vl)
-let HasMasked = false, MaskedPolicyScheme = NonePolicy,
+let HasMasked = false,
+    UnMaskedPolicyScheme = HasPassthruOperandAtIdx1,
+    MaskedPolicyScheme = NonePolicy,
     ManualCodegen = [{
-      std::rotate(Ops.begin(), Ops.begin() + 1, Ops.begin() + 3);
-      IntrinsicTypes = {ResultType, Ops[1]->getType(), Ops[3]->getType()};
+      std::rotate(Ops.begin(), Ops.begin() + 1, Ops.end() - 1);
       // insert undef passthru
+      if (DefaultPolicy == TAIL_AGNOSTIC)
         Ops.insert(Ops.begin(), llvm::UndefValue::get(ResultType));
+      IntrinsicTypes = {ResultType, Ops[2]->getType(), Ops.back()->getType()};
     }] in {
   defm vmerge : RVVOutOp1BuiltinSet<"vmerge", "xfd",
                                     [["vvm", "v", "vmvv"]]>;
   defm vfmerge : RVVOutOp1BuiltinSet<"vfmerge", "xfd",
                                     [["vfm", "v", "vmve"]]>;
 }
 
 // 14.16. Vector Floating-Point Move Instruction
defm vrgather : RVVOutBuiltinSet<"vrgather_vv", "csil",
                                  [["vv", "Uv", "UvUvUv"]]>;
 defm vrgather : RVVOutBuiltinSet<"vrgather_vx", "csil",
                                  [["vx", "Uv", "UvUvz"]]>;
 defm vrgatherei16 : RVVOutBuiltinSet<"vrgatherei16_vv", "csil",
                                      [["vv", "Uv", "UvUv(Log2EEW:4)Uv"]]>;
 }
 
 // 17.5. Vector Compress Instruction
-let HasMasked = false, MaskedPolicyScheme = NonePolicy,
+let IsPrototypeDefaultTU = true,
+    HasMasked = false,
+    UnMaskedPolicyScheme = HasPassthruOperandAtIdx1,
+    MaskedPolicyScheme = NonePolicy,
     ManualCodegen = [{
-      std::rotate(Ops.begin(), Ops.begin() + 1, Ops.begin() + 3);
-      IntrinsicTypes = {ResultType, Ops[3]->getType()};
+      std::rotate(Ops.begin(), Ops.begin() + 1, Ops.end() - 1);
+      // insert undef passthru
+      if (DefaultPolicy == TAIL_AGNOSTIC)
+        Ops.insert(Ops.begin(), llvm::UndefValue::get(ResultType));
+      IntrinsicTypes = {ResultType, Ops.back()->getType()};
     }] in {
   // signed and floating type
   defm vcompress : RVVOutBuiltinSet<"vcompress", "csilxfd",
                                     [["vm", "v", "vmvv"]]>;
   // unsigned type
   defm vcompress : RVVOutBuiltinSet<"vcompress", "csil",
                                     [["vm", "Uv", "UvmUvUv"]]>;
 }

clang/include/clang/Support/RISCVVIntrinsicUtils.h

public:
   computeTypes(BasicType BT, int Log2LMUL, unsigned NF,
                llvm::ArrayRef<PrototypeDescriptor> Prototype);
   static llvm::Optional<RVVTypePtr> computeType(BasicType BT, int Log2LMUL,
                                                 PrototypeDescriptor Proto);
 };
 
 enum PolicyScheme : uint8_t {
   SchemeNone,
+  // Passthru operand is at first parameter in C builtin.
   HasPassthruOperand,
   HasPolicyOperand,
+  // Special case for vmerge, the passthru operand is second
+  // parameter in C builtin.
+  HasPassthruOperandAtIdx1,
 };
 
 // TODO refactor RVVIntrinsic class design after support all intrinsic
 // combination. This represents an instantiation of an intrinsic with a
 // particular type and prototype
 class RVVIntrinsic {
 
 private:

clang/lib/Support/RISCVVIntrinsicUtils.cpp

if (IsMasked) {
     }
   } else if (NF == 1) {
     bool HasPassthruOp = DefaultScheme == PolicyScheme::HasPassthruOperand;
     if (DefaultPolicy == Policy::TU && HasPassthruOp && !IsPrototypeDefaultTU)
       NewPrototype.insert(NewPrototype.begin(), NewPrototype[0]);
     else if (DefaultPolicy == Policy::TA && HasPassthruOp &&
              IsPrototypeDefaultTU)
       NewPrototype.erase(NewPrototype.begin() + 1);
+    if (DefaultScheme == PolicyScheme::HasPassthruOperandAtIdx1) {
+      if (DefaultPolicy == Policy::TU && !IsPrototypeDefaultTU) {
+        // Insert undisturbed output to index 1
+        NewPrototype.insert(NewPrototype.begin() + 2, NewPrototype[0]);
+      } else if (DefaultPolicy == Policy::TA && IsPrototypeDefaultTU) {
+        // Erase passthru for TA policy
+        NewPrototype.erase(NewPrototype.begin() + 2);
+      }
+    }
   }
 
   // If HasVL, append PrototypeDescriptor:VL to last operand
   if (HasVL)
     NewPrototype.push_back(PrototypeDescriptor::VL);
   return NewPrototype;
 }
 

clang/test/CodeGen/RISCV/rvv-intrinsics-overloaded/vcompress.c

 // CHECK-RV64-NEXT:  entry:
 // CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vcompress.nxv8f64.i64(<vscale x 8 x double> [[DEST:%.*]], <vscale x 8 x double> [[SRC:%.*]], <vscale x 8 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vcompress_vm_f64m8 (vbool8_t mask, vfloat64m8_t dest, vfloat64m8_t src, size_t vl) {
   return vcompress(mask, dest, src, vl);
 }
 
+// CHECK-RV64-LABEL: @test_vcompress_vm_i32mf2_tu(
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i32> @llvm.riscv.vcompress.nxv1i32.i64(<vscale x 1 x i32> [[MERGE:%.*]], <vscale x 1 x i32> [[SRC:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i32> [[TMP0]]
+//
+vint32mf2_t test_vcompress_vm_i32mf2_tu(vbool64_t mask, vint32mf2_t merge, vint32mf2_t src, size_t vl) {
+  return vcompress_tu(mask, merge, src, vl);
+}
+
+// CHECK-RV64-LABEL: @test_vcompress_vm_u32mf2_tu(
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i32> @llvm.riscv.vcompress.nxv1i32.i64(<vscale x 1 x i32> [[MERGE:%.*]], <vscale x 1 x i32> [[SRC:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i32> [[TMP0]]
+//
+vuint32mf2_t test_vcompress_vm_u32mf2_tu(vbool64_t mask, vuint32mf2_t merge, vuint32mf2_t src, size_t vl) {
+  return vcompress_tu(mask, merge, src, vl);
+}
+
+// CHECK-RV64-LABEL: @test_vcompress_vm_f32mf2_tu(
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vcompress.nxv1f32.i64(<vscale x 1 x float> [[MERGE:%.*]], <vscale x 1 x float> [[SRC:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vcompress_vm_f32mf2_tu(vbool64_t mask, vfloat32mf2_t merge, vfloat32mf2_t src, size_t vl) {
+  return vcompress_tu(mask, merge, src, vl);
+}
 
+// CHECK-RV64-LABEL: @test_vcompress_vm_i32mf2_ta(
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i32> @llvm.riscv.vcompress.nxv1i32.i64(<vscale x 1 x i32> undef, <vscale x 1 x i32> [[SRC:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i32> [[TMP0]]
+//
+vint32mf2_t test_vcompress_vm_i32mf2_ta(vbool64_t mask, vint32mf2_t src, size_t vl) {
+  return vcompress_ta(mask, src, vl);
+}
+
+// CHECK-RV64-LABEL: @test_vcompress_vm_u32mf2_ta(
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i32> @llvm.riscv.vcompress.nxv1i32.i64(<vscale x 1 x i32> undef, <vscale x 1 x i32> [[SRC:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i32> [[TMP0]]
+//
+vuint32mf2_t test_vcompress_vm_u32mf2_ta(vbool64_t mask, vuint32mf2_t src, size_t vl) {
+  return vcompress_ta(mask, src, vl);
+}
+
+// CHECK-RV64-LABEL: @test_vcompress_vm_f32mf2_ta(
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vcompress.nxv1f32.i64(<vscale x 1 x float> undef, <vscale x 1 x float> [[SRC:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vcompress_vm_f32mf2_ta(vbool64_t mask, vfloat32mf2_t src, size_t vl) {
+  return vcompress_ta(mask, src, vl);
+}

clang/test/CodeGen/RISCV/rvv-intrinsics-overloaded/vfmerge.c

 // CHECK-RV64-NEXT:  entry:
 // CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vfmerge.nxv8f64.f64.i64(<vscale x 8 x double> undef, <vscale x 8 x double> [[OP1:%.*]], double [[OP2:%.*]], <vscale x 8 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vfmerge_vfm_f64m8(vbool8_t mask, vfloat64m8_t op1, double op2,
                                     size_t vl) {
   return vfmerge(mask, op1, op2, vl);
 }
+
+// CHECK-RV64-LABEL: @test_vfmerge_vfm_f32mf2_tu(
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfmerge.nxv1f32.f32.i64(<vscale x 1 x float> [[MERGE:%.*]], <vscale x 1 x float> [[OP1:%.*]], float [[OP2:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfmerge_vfm_f32mf2_tu(vbool64_t mask, vfloat32mf2_t merge, vfloat32mf2_t op1, float op2, size_t vl) {
+  return vfmerge_tu(mask, merge, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: @test_vfmerge_vfm_f32mf2_ta(
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfmerge.nxv1f32.f32.i64(<vscale x 1 x float> undef, <vscale x 1 x float> [[OP1:%.*]], float [[OP2:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfmerge_vfm_f32mf2_ta(vbool64_t mask, vfloat32mf2_t op1, float op2, size_t vl) {
+  return vfmerge_ta(mask, op1, op2, vl);
+}

clang/test/CodeGen/RISCV/rvv-intrinsics-overloaded/vmerge.c

 // CHECK-RV64-NEXT:  entry:
 // CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x double> @llvm.riscv.vmerge.nxv8f64.nxv8f64.i64(<vscale x 8 x double> undef, <vscale x 8 x double> [[OP1:%.*]], <vscale x 8 x double> [[OP2:%.*]], <vscale x 8 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
 // CHECK-RV64-NEXT:    ret <vscale x 8 x double> [[TMP0]]
 //
 vfloat64m8_t test_vmerge_vvm_f64m8(vbool8_t mask, vfloat64m8_t op1,
                                    vfloat64m8_t op2, size_t vl) {
   return vmerge(mask, op1, op2, vl);
 }
+
+// CHECK-RV64-LABEL: @test_vmerge_vvm_i32mf2_tu(
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i32> @llvm.riscv.vmerge.nxv1i32.nxv1i32.i64(<vscale x 1 x i32> [[MERGE:%.*]], <vscale x 1 x i32> [[OP1:%.*]], <vscale x 1 x i32> [[OP2:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i32> [[TMP0]]
+//
+vint32mf2_t test_vmerge_vvm_i32mf2_tu(vbool64_t mask, vint32mf2_t merge, vint32mf2_t op1, vint32mf2_t op2, size_t vl) {
+  return vmerge_tu(mask, merge, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: @test_vmerge_vxm_i32mf2_tu(
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i32> @llvm.riscv.vmerge.nxv1i32.i32.i64(<vscale x 1 x i32> [[MERGE:%.*]], <vscale x 1 x i32> [[OP1:%.*]], i32 [[OP2:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i32> [[TMP0]]
+//
+vint32mf2_t test_vmerge_vxm_i32mf2_tu(vbool64_t mask, vint32mf2_t merge, vint32mf2_t op1, int32_t op2, size_t vl) {
+  return vmerge_tu(mask, merge, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: @test_vmerge_vvm_u32mf2_tu(
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i32> @llvm.riscv.vmerge.nxv1i32.nxv1i32.i64(<vscale x 1 x i32> [[MERGE:%.*]], <vscale x 1 x i32> [[OP1:%.*]], <vscale x 1 x i32> [[OP2:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i32> [[TMP0]]
+//
+vuint32mf2_t test_vmerge_vvm_u32mf2_tu(vbool64_t mask, vuint32mf2_t merge, vuint32mf2_t op1, vuint32mf2_t op2, size_t vl) {
+  return vmerge_tu(mask, merge, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: @test_vmerge_vxm_u32mf2_tu(
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i32> @llvm.riscv.vmerge.nxv1i32.i32.i64(<vscale x 1 x i32> [[MERGE:%.*]], <vscale x 1 x i32> [[OP1:%.*]], i32 [[OP2:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i32> [[TMP0]]
+//
+vuint32mf2_t test_vmerge_vxm_u32mf2_tu(vbool64_t mask, vuint32mf2_t merge, vuint32mf2_t op1, uint32_t op2, size_t vl) {
+  return vmerge_tu(mask, merge, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: @test_vmerge_vvm_i32mf2_ta(
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i32> @llvm.riscv.vmerge.nxv1i32.nxv1i32.i64(<vscale x 1 x i32> undef, <vscale x 1 x i32> [[OP1:%.*]], <vscale x 1 x i32> [[OP2:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i32> [[TMP0]]
+//
+vint32mf2_t test_vmerge_vvm_i32mf2_ta(vbool64_t mask, vint32mf2_t op1, vint32mf2_t op2, size_t vl) {
+  return vmerge_ta(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: @test_vmerge_vxm_i32mf2_ta(
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i32> @llvm.riscv.vmerge.nxv1i32.i32.i64(<vscale x 1 x i32> undef, <vscale x 1 x i32> [[OP1:%.*]], i32 [[OP2:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i32> [[TMP0]]
+//
+vint32mf2_t test_vmerge_vxm_i32mf2_ta(vbool64_t mask, vint32mf2_t op1, int32_t op2, size_t vl) {
+  return vmerge_ta(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: @test_vmerge_vvm_u32mf2_ta(
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i32> @llvm.riscv.vmerge.nxv1i32.nxv1i32.i64(<vscale x 1 x i32> undef, <vscale x 1 x i32> [[OP1:%.*]], <vscale x 1 x i32> [[OP2:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i32> [[TMP0]]
+//
+vuint32mf2_t test_vmerge_vvm_u32mf2_ta(vbool64_t mask, vuint32mf2_t op1, vuint32mf2_t op2, size_t vl) {
+  return vmerge_ta(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: @test_vmerge_vxm_u32mf2_ta(
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i32> @llvm.riscv.vmerge.nxv1i32.i32.i64(<vscale x 1 x i32> undef, <vscale x 1 x i32> [[OP1:%.*]], i32 [[OP2:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i32> [[TMP0]]
+//
+vuint32mf2_t test_vmerge_vxm_u32mf2_ta(vbool64_t mask, vuint32mf2_t op1, uint32_t op2, size_t vl) {
+  return vmerge_ta(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: @test_vmerge_vvm_f32mf2_tu(
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vmerge.nxv1f32.nxv1f32.i64(<vscale x 1 x float> [[MERGE:%.*]], <vscale x 1 x float> [[OP1:%.*]], <vscale x 1 x float> [[OP2:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vmerge_vvm_f32mf2_tu(vbool64_t mask, vfloat32mf2_t merge, vfloat32mf2_t op1, vfloat32mf2_t op2, size_t vl) {
+  return vmerge_tu(mask, merge, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: @test_vmerge_vvm_f32mf2_ta(
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vmerge.nxv1f32.nxv1f32.i64(<vscale x 1 x float> undef, <vscale x 1 x float> [[OP1:%.*]], <vscale x 1 x float> [[OP2:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vmerge_vvm_f32mf2_ta(vbool64_t mask, vfloat32mf2_t op1, vfloat32mf2_t op2, size_t vl) {
+  return vmerge_ta(mask, op1, op2, vl);
+}

clang/test/CodeGen/RISCV/rvv-intrinsics/vcompress.c

 // CHECK-RV64-LABEL: @test_vcompress_vm_f16m8(
 // CHECK-RV64-NEXT:  entry:
 // CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x half> @llvm.riscv.vcompress.nxv32f16.i64(<vscale x 32 x half> [[DEST:%.*]], <vscale x 32 x half> [[SRC:%.*]], <vscale x 32 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
 // CHECK-RV64-NEXT:    ret <vscale x 32 x half> [[TMP0]]
 //
 vfloat16m8_t test_vcompress_vm_f16m8 (vbool2_t mask, vfloat16m8_t dest, vfloat16m8_t src, size_t vl) {
   return vcompress_vm_f16m8(mask, dest, src, vl);
 }
+
+// CHECK-RV64-LABEL: @test_vcompress_vm_i32mf2_tu(
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i32> @llvm.riscv.vcompress.nxv1i32.i64(<vscale x 1 x i32> [[MERGE:%.*]], <vscale x 1 x i32> [[SRC:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i32> [[TMP0]]
+//
+vint32mf2_t test_vcompress_vm_i32mf2_tu(vbool64_t mask, vint32mf2_t merge, vint32mf2_t src, size_t vl) {
+  return vcompress_vm_i32mf2_tu(mask, merge, src, vl);
+}
+
+// CHECK-RV64-LABEL: @test_vcompress_vm_u32mf2_tu(
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i32> @llvm.riscv.vcompress.nxv1i32.i64(<vscale x 1 x i32> [[MERGE:%.*]], <vscale x 1 x i32> [[SRC:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i32> [[TMP0]]
+//
+vuint32mf2_t test_vcompress_vm_u32mf2_tu(vbool64_t mask, vuint32mf2_t merge, vuint32mf2_t src, size_t vl) {
+  return vcompress_vm_u32mf2_tu(mask, merge, src, vl);
+}
+
+// CHECK-RV64-LABEL: @test_vcompress_vm_f32mf2_tu(
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vcompress.nxv1f32.i64(<vscale x 1 x float> [[MERGE:%.*]], <vscale x 1 x float> [[SRC:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vcompress_vm_f32mf2_tu(vbool64_t mask, vfloat32mf2_t merge, vfloat32mf2_t src, size_t vl) {
+  return vcompress_vm_f32mf2_tu(mask, merge, src, vl);
+}
+
+// CHECK-RV64-LABEL: @test_vcompress_vm_i32mf2_ta(
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i32> @llvm.riscv.vcompress.nxv1i32.i64(<vscale x 1 x i32> undef, <vscale x 1 x i32> [[SRC:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i32> [[TMP0]]
+//
+vint32mf2_t test_vcompress_vm_i32mf2_ta(vbool64_t mask, vint32mf2_t src, size_t vl) {
+  return vcompress_vm_i32mf2_ta(mask, src, vl);
+}
+
+// CHECK-RV64-LABEL: @test_vcompress_vm_u32mf2_ta(
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i32> @llvm.riscv.vcompress.nxv1i32.i64(<vscale x 1 x i32> undef, <vscale x 1 x i32> [[SRC:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i32> [[TMP0]]
+//
+vuint32mf2_t test_vcompress_vm_u32mf2_ta(vbool64_t mask, vuint32mf2_t src, size_t vl) {
+  return vcompress_vm_u32mf2_ta(mask, src, vl);
+}
+
+// CHECK-RV64-LABEL: @test_vcompress_vm_f32mf2_ta(
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vcompress.nxv1f32.i64(<vscale x 1 x float> undef, <vscale x 1 x float> [[SRC:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vcompress_vm_f32mf2_ta(vbool64_t mask, vfloat32mf2_t src, size_t vl) {
+  return vcompress_vm_f32mf2_ta(mask, src, vl);
+}

clang/test/CodeGen/RISCV/rvv-intrinsics/vfmerge.c

vfloat64m8_t test_vfmerge_vfm_f64m8(vbool8_t mask, vfloat64m8_t op1, double op2,
   return vfmerge_vfm_f64m8(mask, op1, op2, vl);
 }
 
 // CHECK-RV64-LABEL: @test_vfmerge_vfm_f16mf4(
 // CHECK-RV64-NEXT:  entry:
 // CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x half> @llvm.riscv.vfmerge.nxv1f16.f16.i64(<vscale x 1 x half> undef, <vscale x 1 x half> [[OP1:%.*]], half [[OP2:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x half> [[TMP0]]
 //
-vfloat16mf4_t test_vfmerge_vfm_f16mf4 (vbool64_t mask, vfloat16mf4_t op1, _Float16 op2, size_t vl) {
+vfloat16mf4_t test_vfmerge_vfm_f16mf4(vbool64_t mask, vfloat16mf4_t op1, _Float16 op2, size_t vl) {
   return vfmerge_vfm_f16mf4(mask, op1, op2, vl);
 }
 
 // CHECK-RV64-LABEL: @test_vfmerge_vfm_f16mf2(
 // CHECK-RV64-NEXT:  entry:
 // CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x half> @llvm.riscv.vfmerge.nxv2f16.f16.i64(<vscale x 2 x half> undef, <vscale x 2 x half> [[OP1:%.*]], half [[OP2:%.*]], <vscale x 2 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x half> [[TMP0]]
 //
-vfloat16mf2_t test_vfmerge_vfm_f16mf2 (vbool32_t mask, vfloat16mf2_t op1, _Float16 op2, size_t vl) {
+vfloat16mf2_t test_vfmerge_vfm_f16mf2(vbool32_t mask, vfloat16mf2_t op1, _Float16 op2, size_t vl) {
   return vfmerge_vfm_f16mf2(mask, op1, op2, vl);
 }
 
 // CHECK-RV64-LABEL: @test_vfmerge_vfm_f16m1(
 // CHECK-RV64-NEXT:  entry:
 // CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vfmerge.nxv4f16.f16.i64(<vscale x 4 x half> undef, <vscale x 4 x half> [[OP1:%.*]], half [[OP2:%.*]], <vscale x 4 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
-vfloat16m1_t test_vfmerge_vfm_f16m1 (vbool16_t mask, vfloat16m1_t op1, _Float16 op2, size_t vl) {
+vfloat16m1_t test_vfmerge_vfm_f16m1(vbool16_t mask, vfloat16m1_t op1, _Float16 op2, size_t vl) {
   return vfmerge_vfm_f16m1(mask, op1, op2, vl);
 }
 
 // CHECK-RV64-LABEL: @test_vfmerge_vfm_f16m2(
 // CHECK-RV64-NEXT:  entry:
 // CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x half> @llvm.riscv.vfmerge.nxv8f16.f16.i64(<vscale x 8 x half> undef, <vscale x 8 x half> [[OP1:%.*]], half [[OP2:%.*]], <vscale x 8 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
 // CHECK-RV64-NEXT:    ret <vscale x 8 x half> [[TMP0]]
 //
-vfloat16m2_t test_vfmerge_vfm_f16m2 (vbool8_t mask, vfloat16m2_t op1, _Float16 op2, size_t vl) {
+vfloat16m2_t test_vfmerge_vfm_f16m2(vbool8_t mask, vfloat16m2_t op1, _Float16 op2, size_t vl) {
   return vfmerge_vfm_f16m2(mask, op1, op2, vl);
 }
 
 // CHECK-RV64-LABEL: @test_vfmerge_vfm_f16m4(
 // CHECK-RV64-NEXT:  entry:
 // CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x half> @llvm.riscv.vfmerge.nxv16f16.f16.i64(<vscale x 16 x half> undef, <vscale x 16 x half> [[OP1:%.*]], half [[OP2:%.*]], <vscale x 16 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
 // CHECK-RV64-NEXT:    ret <vscale x 16 x half> [[TMP0]]
 //
-vfloat16m4_t test_vfmerge_vfm_f16m4 (vbool4_t mask, vfloat16m4_t op1, _Float16 op2, size_t vl) {
+vfloat16m4_t test_vfmerge_vfm_f16m4(vbool4_t mask, vfloat16m4_t op1, _Float16 op2, size_t vl) {
   return vfmerge_vfm_f16m4(mask, op1, op2, vl);
 }
 
 // CHECK-RV64-LABEL: @test_vfmerge_vfm_f16m8(
 // CHECK-RV64-NEXT:  entry:
 // CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x half> @llvm.riscv.vfmerge.nxv32f16.f16.i64(<vscale x 32 x half> undef, <vscale x 32 x half> [[OP1:%.*]], half [[OP2:%.*]], <vscale x 32 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
 // CHECK-RV64-NEXT:    ret <vscale x 32 x half> [[TMP0]]
 //
-vfloat16m8_t test_vfmerge_vfm_f16m8 (vbool2_t mask, vfloat16m8_t op1, _Float16 op2, size_t vl) {
+vfloat16m8_t test_vfmerge_vfm_f16m8(vbool2_t mask, vfloat16m8_t op1, _Float16 op2, size_t vl) {
   return vfmerge_vfm_f16m8(mask, op1, op2, vl);
 }
+
+// CHECK-RV64-LABEL: @test_vfmerge_vfm_f32mf2_tu(
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfmerge.nxv1f32.f32.i64(<vscale x 1 x float> [[MERGE:%.*]], <vscale x 1 x float> [[OP1:%.*]], float [[OP2:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfmerge_vfm_f32mf2_tu(vbool64_t mask, vfloat32mf2_t merge, vfloat32mf2_t op1, float op2, size_t vl) {
+  return vfmerge_vfm_f32mf2_tu(mask, merge, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: @test_vfmerge_vfm_f32mf2_ta(
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vfmerge.nxv1f32.f32.i64(<vscale x 1 x float> undef, <vscale x 1 x float> [[OP1:%.*]], float [[OP2:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vfmerge_vfm_f32mf2_ta(vbool64_t mask, vfloat32mf2_t op1, float op2, size_t vl) {
+  return vfmerge_vfm_f32mf2_ta(mask, op1, op2, vl);
+}

clang/test/CodeGen/RISCV/rvv-intrinsics/vmerge.c

vfloat64m8_t test_vmerge_vvm_f64m8(vbool8_t mask, vfloat64m8_t op1,
   return vmerge_vvm_f64m8(mask, op1, op2, vl);
 }
 
 // CHECK-RV64-LABEL: @test_vmerge_vvm_f16mf4(
 // CHECK-RV64-NEXT:  entry:
 // CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x half> @llvm.riscv.vmerge.nxv1f16.nxv1f16.i64(<vscale x 1 x half> undef, <vscale x 1 x half> [[OP1:%.*]], <vscale x 1 x half> [[OP2:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
 // CHECK-RV64-NEXT:    ret <vscale x 1 x half> [[TMP0]]
 //
-vfloat16mf4_t test_vmerge_vvm_f16mf4 (vbool64_t mask, vfloat16mf4_t op1, vfloat16mf4_t op2, size_t vl) {
+vfloat16mf4_t test_vmerge_vvm_f16mf4(vbool64_t mask, vfloat16mf4_t op1, vfloat16mf4_t op2, size_t vl) {
   return vmerge_vvm_f16mf4(mask, op1, op2, vl);
 }
 
 // CHECK-RV64-LABEL: @test_vmerge_vvm_f16mf2(
 // CHECK-RV64-NEXT:  entry:
 // CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x half> @llvm.riscv.vmerge.nxv2f16.nxv2f16.i64(<vscale x 2 x half> undef, <vscale x 2 x half> [[OP1:%.*]], <vscale x 2 x half> [[OP2:%.*]], <vscale x 2 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
 // CHECK-RV64-NEXT:    ret <vscale x 2 x half> [[TMP0]]
 //
-vfloat16mf2_t test_vmerge_vvm_f16mf2 (vbool32_t mask, vfloat16mf2_t op1, vfloat16mf2_t op2, size_t vl) {
+vfloat16mf2_t test_vmerge_vvm_f16mf2(vbool32_t mask, vfloat16mf2_t op1, vfloat16mf2_t op2, size_t vl) {
   return vmerge_vvm_f16mf2(mask, op1, op2, vl);
 }
 
 // CHECK-RV64-LABEL: @test_vmerge_vvm_f16m1(
 // CHECK-RV64-NEXT:  entry:
 // CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.riscv.vmerge.nxv4f16.nxv4f16.i64(<vscale x 4 x half> undef, <vscale x 4 x half> [[OP1:%.*]], <vscale x 4 x half> [[OP2:%.*]], <vscale x 4 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
 // CHECK-RV64-NEXT:    ret <vscale x 4 x half> [[TMP0]]
 //
-vfloat16m1_t test_vmerge_vvm_f16m1 (vbool16_t mask, vfloat16m1_t op1, vfloat16m1_t op2, size_t vl) {
+vfloat16m1_t test_vmerge_vvm_f16m1(vbool16_t mask, vfloat16m1_t op1, vfloat16m1_t op2, size_t vl) {
   return vmerge_vvm_f16m1(mask, op1, op2, vl);
 }
 
 // CHECK-RV64-LABEL: @test_vmerge_vvm_f16m2(
 // CHECK-RV64-NEXT:  entry:
 // CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x half> @llvm.riscv.vmerge.nxv8f16.nxv8f16.i64(<vscale x 8 x half> undef, <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
 // CHECK-RV64-NEXT:    ret <vscale x 8 x half> [[TMP0]]
 //
-vfloat16m2_t test_vmerge_vvm_f16m2 (vbool8_t mask, vfloat16m2_t op1, vfloat16m2_t op2, size_t vl) {
+vfloat16m2_t test_vmerge_vvm_f16m2(vbool8_t mask, vfloat16m2_t op1, vfloat16m2_t op2, size_t vl) {
   return vmerge_vvm_f16m2(mask, op1, op2, vl);
 }
 
 // CHECK-RV64-LABEL: @test_vmerge_vvm_f16m4(
 // CHECK-RV64-NEXT:  entry:
 // CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x half> @llvm.riscv.vmerge.nxv16f16.nxv16f16.i64(<vscale x 16 x half> undef, <vscale x 16 x half> [[OP1:%.*]], <vscale x 16 x half> [[OP2:%.*]], <vscale x 16 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
 // CHECK-RV64-NEXT:    ret <vscale x 16 x half> [[TMP0]]
 //
-vfloat16m4_t test_vmerge_vvm_f16m4 (vbool4_t mask, vfloat16m4_t op1, vfloat16m4_t op2, size_t vl) {
+vfloat16m4_t test_vmerge_vvm_f16m4(vbool4_t mask, vfloat16m4_t op1, vfloat16m4_t op2, size_t vl) {
   return vmerge_vvm_f16m4(mask, op1, op2, vl);
 }
 
 // CHECK-RV64-LABEL: @test_vmerge_vvm_f16m8(
 // CHECK-RV64-NEXT:  entry:
 // CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 32 x half> @llvm.riscv.vmerge.nxv32f16.nxv32f16.i64(<vscale x 32 x half> undef, <vscale x 32 x half> [[OP1:%.*]], <vscale x 32 x half> [[OP2:%.*]], <vscale x 32 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
 // CHECK-RV64-NEXT:    ret <vscale x 32 x half> [[TMP0]]
 //
-vfloat16m8_t test_vmerge_vvm_f16m8 (vbool2_t mask, vfloat16m8_t op1, vfloat16m8_t op2, size_t vl) {
+vfloat16m8_t test_vmerge_vvm_f16m8(vbool2_t mask, vfloat16m8_t op1, vfloat16m8_t op2, size_t vl) {
   return vmerge_vvm_f16m8(mask, op1, op2, vl);
 }
+
+// CHECK-RV64-LABEL: @test_vmerge_vvm_i32mf2_tu(
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i32> @llvm.riscv.vmerge.nxv1i32.nxv1i32.i64(<vscale x 1 x i32> [[MERGE:%.*]], <vscale x 1 x i32> [[OP1:%.*]], <vscale x 1 x i32> [[OP2:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i32> [[TMP0]]
+//
+vint32mf2_t test_vmerge_vvm_i32mf2_tu(vbool64_t mask, vint32mf2_t merge, vint32mf2_t op1, vint32mf2_t op2, size_t vl) {
+  return vmerge_vvm_i32mf2_tu(mask, merge, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: @test_vmerge_vxm_i32mf2_tu(
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i32> @llvm.riscv.vmerge.nxv1i32.i32.i64(<vscale x 1 x i32> [[MERGE:%.*]], <vscale x 1 x i32> [[OP1:%.*]], i32 [[OP2:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i32> [[TMP0]]
+//
+vint32mf2_t test_vmerge_vxm_i32mf2_tu(vbool64_t mask, vint32mf2_t merge, vint32mf2_t op1, int32_t op2, size_t vl) {
+  return vmerge_vxm_i32mf2_tu(mask, merge, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: @test_vmerge_vvm_u32mf2_tu(
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i32> @llvm.riscv.vmerge.nxv1i32.nxv1i32.i64(<vscale x 1 x i32> [[MERGE:%.*]], <vscale x 1 x i32> [[OP1:%.*]], <vscale x 1 x i32> [[OP2:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i32> [[TMP0]]
+//
+vuint32mf2_t test_vmerge_vvm_u32mf2_tu(vbool64_t mask, vuint32mf2_t merge, vuint32mf2_t op1, vuint32mf2_t op2, size_t vl) {
+  return vmerge_vvm_u32mf2_tu(mask, merge, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: @test_vmerge_vxm_u32mf2_tu(
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i32> @llvm.riscv.vmerge.nxv1i32.i32.i64(<vscale x 1 x i32> [[MERGE:%.*]], <vscale x 1 x i32> [[OP1:%.*]], i32 [[OP2:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i32> [[TMP0]]
+//
+vuint32mf2_t test_vmerge_vxm_u32mf2_tu(vbool64_t mask, vuint32mf2_t merge, vuint32mf2_t op1, uint32_t op2, size_t vl) {
+  return vmerge_vxm_u32mf2_tu(mask, merge, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: @test_vmerge_vvm_i32mf2_ta(
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i32> @llvm.riscv.vmerge.nxv1i32.nxv1i32.i64(<vscale x 1 x i32> undef, <vscale x 1 x i32> [[OP1:%.*]], <vscale x 1 x i32> [[OP2:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i32> [[TMP0]]
+//
+vint32mf2_t test_vmerge_vvm_i32mf2_ta(vbool64_t mask, vint32mf2_t op1, vint32mf2_t op2, size_t vl) {
+  return vmerge_vvm_i32mf2_ta(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: @test_vmerge_vxm_i32mf2_ta(
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i32> @llvm.riscv.vmerge.nxv1i32.i32.i64(<vscale x 1 x i32> undef, <vscale x 1 x i32> [[OP1:%.*]], i32 [[OP2:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i32> [[TMP0]]
+//
+vint32mf2_t test_vmerge_vxm_i32mf2_ta(vbool64_t mask, vint32mf2_t op1, int32_t op2, size_t vl) {
+  return vmerge_vxm_i32mf2_ta(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: @test_vmerge_vvm_u32mf2_ta(
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i32> @llvm.riscv.vmerge.nxv1i32.nxv1i32.i64(<vscale x 1 x i32> undef, <vscale x 1 x i32> [[OP1:%.*]], <vscale x 1 x i32> [[OP2:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i32> [[TMP0]]
+//
+vuint32mf2_t test_vmerge_vvm_u32mf2_ta(vbool64_t mask, vuint32mf2_t op1, vuint32mf2_t op2, size_t vl) {
+  return vmerge_vvm_u32mf2_ta(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: @test_vmerge_vxm_u32mf2_ta(
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x i32> @llvm.riscv.vmerge.nxv1i32.i32.i64(<vscale x 1 x i32> undef, <vscale x 1 x i32> [[OP1:%.*]], i32 [[OP2:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x i32> [[TMP0]]
+//
+vuint32mf2_t test_vmerge_vxm_u32mf2_ta(vbool64_t mask, vuint32mf2_t op1, uint32_t op2, size_t vl) {
+  return vmerge_vxm_u32mf2_ta(mask, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: @test_vmerge_vvm_f32mf2_tu(
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vmerge.nxv1f32.nxv1f32.i64(<vscale x 1 x float> [[MERGE:%.*]], <vscale x 1 x float> [[OP1:%.*]], <vscale x 1 x float> [[OP2:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vmerge_vvm_f32mf2_tu(vbool64_t mask, vfloat32mf2_t merge, vfloat32mf2_t op1, vfloat32mf2_t op2, size_t vl) {
+  return vmerge_vvm_f32mf2_tu(mask, merge, op1, op2, vl);
+}
+
+// CHECK-RV64-LABEL: @test_vmerge_vvm_f32mf2_ta(
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = call <vscale x 1 x float> @llvm.riscv.vmerge.nxv1f32.nxv1f32.i64(<vscale x 1 x float> undef, <vscale x 1 x float> [[OP1:%.*]], <vscale x 1 x float> [[OP2:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]])
+// CHECK-RV64-NEXT:    ret <vscale x 1 x float> [[TMP0]]
+//
+vfloat32mf2_t test_vmerge_vvm_f32mf2_ta(vbool64_t mask, vfloat32mf2_t op1, vfloat32mf2_t op2, size_t vl) {
+  return vmerge_vvm_f32mf2_ta(mask, op1, op2, vl);
+}