[RISCV] Add intrinsics for vmv.x.s and vmv.s.x

[craig.topper]

Summary

This adds intrinsics for vmv.x.s and vmv.s.x.

I've used stricter type constraints on these intrinsics than what we've been doing on the arithmetic intrinsics so far. This will allow us to not need to pass the scalar type to the Intrinsic::getDeclaration call when creating these intrinsics.

A custom ISD is used for vmv.x.s in order to implement the change in computeNumSignBitsForTargetNode which can remove sign extends on the result.

I also modified the MC layer description of these instructions to show the tied source/dest operand. This is different than what we do for masked instructions where we drop the tied source operand when converting to MC. But it is a more accurate description of the instruction. We can't do this for masked instructions since we use the same MC instruction for masked and unmasked. Tools like llvm-mca operate in the MC layer and rely on ins/outs and Uses/Defs for analysis so I don't know if we'll be able to maintain the current behavior for masked instructions. So I went with the accurate description here since it was easy.

[craig.topper]

llvm/test/CodeGen/RISCV/rvv/vmv.x.s-rv32.ll

7

The signext attribute on the results here gives test coverage on the computeNumSignBitsForTargetNode change.

[craig.topper]

Rebase and use anyint_ty instead of anyvector_ty to be consistent with an upcoming vfmv.s.f/vfmv.f.s patch

[HsiangKai]

llvm/lib/Target/RISCV/RISCVInstrInfoVPseudos.td

1195

Do we need to add a constraint "$rd = $rs1"?

[craig.topper]

llvm/lib/Target/RISCV/RISCVInstrInfoVPseudos.td

1195

Yes we do. I thought I had done that. Thanks!

[craig.topper]

Add tied register constraint. Remove unnecessary type from intrinsic names in tests.

[frasercrmck]

LGTM. Was it a TableGen type contradiction error that you saw without the custom node?

[craig.topper]

In D93365#2459803, @frasercrmck wrote:

LGTM. Was it a TableGen type contradiction error that you saw without the custom node?

Well I thought it was, but after debugging a different issue with fmv it might not have been. The DAG isel patterns don't know about LLVMVectorElementType<0> so the change to anyint_ty might have fixed it.

[frasercrmck]

In D93365#2459816, @craig.topper wrote:

Well I thought it was, but after debugging a different issue with fmv it might not have been. The DAG isel patterns don't know about LLVMVectorElementType<0> so the change to anyint_ty might have fixed it.

Oh neat, is it worth trying without the custom nodes, then?

[craig.topper]

Remove custom node for vmv.s.x. Using llvm_anyint_ty for the vector instead of llvm_anyvector_ty made the isel table work after all.

[craig.topper]

Update comment I missed when removing one of the custom ISD opcodes

[craig.topper]

@frasercrmck does this look ok to you with the custom node removed?

[frasercrmck]

Yep, LGTM.

llvm/include/llvm/IR/IntrinsicsRISCV.td

let TargetPrefix = "riscv" in {
 
   defm vwmul : RISCVBinaryABX;
   defm vwmulu : RISCVBinaryABX;
   defm vwmulsu : RISCVBinaryABX;
 
   defm vfadd : RISCVBinaryAAX;
   defm vfsub : RISCVBinaryAAX;
   defm vfrsub : RISCVBinaryAAX;
+
+  def int_riscv_vmv_x_s : Intrinsic<[LLVMVectorElementType<0>],
+                                    [llvm_anyint_ty],
+                                    [IntrNoMem]>, RISCVVIntrinsic;
+  def int_riscv_vmv_s_x : Intrinsic<[llvm_anyint_ty],
+                                    [LLVMMatchType<0>, LLVMVectorElementType<0>,
+                                     llvm_anyint_ty],
+                                    [IntrNoMem]>, RISCVVIntrinsic {
+    let ExtendOperand = 2;
+  }
 } // TargetPrefix = "riscv"

llvm/lib/Target/RISCV/RISCVISelLowering.h

 //===-- RISCVISelLowering.h - RISCV DAG Lowering Interface ------*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 //
enum NodeType : unsigned {
   // TableGen chokes when faced with commutative permutations in deeply-nested
   // DAGs. Each node takes an input operand and a TargetConstant immediate
   // shift amount, and outputs a bit-manipulated version of input. All operands
   // are of type XLenVT.
   GREVI,
   GREVIW,
   GORCI,
   GORCIW,
+  // Vector Extension
+  // VMV_X_S matches the semantics of vmv.x.s. The result is always XLenVT
+  // sign extended from the vector element size. NOTE: The result size will
+  // never be less than the vector element size.
+  VMV_X_S,
 };
 } // namespace RISCVISD
 
 class RISCVTargetLowering : public TargetLowering {
   const RISCVSubtarget &Subtarget;
 
 public:
   explicit RISCVTargetLowering(const TargetMachine &TM,

llvm/lib/Target/RISCV/RISCVISelLowering.cpp

 //===-- RISCVISelLowering.cpp - RISCV DAG Lowering Implementation  --------===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 //
RISCVTargetLowering::RISCVTargetLowering(const TargetMachine &TM,
   }
 
   setBooleanContents(ZeroOrOneBooleanContent);
 
   if (Subtarget.hasStdExtV()) {
     setBooleanVectorContents(ZeroOrOneBooleanContent);
 
     // RVV intrinsics may have illegal operands.
+    // We also need to custom legalize vmv.x.s and vmv.s.x.
     setOperationAction(ISD::INTRINSIC_WO_CHAIN, MVT::i8, Custom);
     setOperationAction(ISD::INTRINSIC_WO_CHAIN, MVT::i16, Custom);
-    if (Subtarget.is64Bit())
-      setOperationAction(ISD::INTRINSIC_WO_CHAIN, MVT::i32, Custom);
+    if (Subtarget.is64Bit())
+      setOperationAction(ISD::INTRINSIC_WO_CHAIN, MVT::i64, Custom);
   }
 
   // Function alignments.
   const Align FunctionAlignment(Subtarget.hasStdExtC() ? 2 : 4);
   setMinFunctionAlignment(FunctionAlignment);
   setPrefFunctionAlignment(FunctionAlignment);
 
   // Effectively disable jump table generation.
if (Subtarget.hasStdExtV()) {
     // Some RVV intrinsics may claim that they want an integer operand to be
     // extended.
     if (const RISCVVIntrinsicsTable::RISCVVIntrinsicInfo *II =
             RISCVVIntrinsicsTable::getRISCVVIntrinsicInfo(IntNo)) {
       if (II->ExtendedOperand) {
         assert(II->ExtendedOperand < Op.getNumOperands());
         SmallVector<SDValue, 8> Operands(Op->op_begin(), Op->op_end());
         SDValue &ScalarOp = Operands[II->ExtendedOperand];
-        if (ScalarOp.getValueType() == MVT::i8 ||
-            ScalarOp.getValueType() == MVT::i16 ||
-            ScalarOp.getValueType() == MVT::i32) {
+        EVT OpVT = ScalarOp.getValueType();
+        if (OpVT == MVT::i8 || OpVT == MVT::i16 ||
+            (OpVT == MVT::i32 && Subtarget.is64Bit())) {
           ScalarOp =
               DAG.getNode(ISD::ANY_EXTEND, DL, Subtarget.getXLenVT(), ScalarOp);
           return DAG.getNode(ISD::INTRINSIC_WO_CHAIN, DL, Op.getValueType(),
                              Operands);
         }
       }
     }
   }
 
   switch (IntNo) {
   default:
     return SDValue();    // Don't custom lower most intrinsics.
   case Intrinsic::thread_pointer: {
     EVT PtrVT = getPointerTy(DAG.getDataLayout());
     return DAG.getRegister(RISCV::X4, PtrVT);
   }
+  case Intrinsic::riscv_vmv_x_s:
+    assert(Op.getValueType() == Subtarget.getXLenVT() && "Unexpected VT!");
+    return DAG.getNode(RISCVISD::VMV_X_S, DL, Op.getValueType(),
+                       Op.getOperand(1));
   }
 }
 
 // Returns the opcode of the target-specific SDNode that implements the 32-bit
 // form of the given Opcode.
 static RISCVISD::NodeType getRISCVWOpcode(unsigned Opcode) {
   switch (Opcode) {
   default:
case ISD::FSHR: {
     NewOp2 = DAG.getNode(ISD::AND, DL, MVT::i64, NewOp2,
                          DAG.getConstant(0x1f, DL, MVT::i64));
     unsigned Opc =
         N->getOpcode() == ISD::FSHL ? RISCVISD::FSLW : RISCVISD::FSRW;
     SDValue NewOp = DAG.getNode(Opc, DL, MVT::i64, NewOp0, NewOp1, NewOp2);
     Results.push_back(DAG.getNode(ISD::TRUNCATE, DL, MVT::i32, NewOp));
     break;
   }
+  case ISD::INTRINSIC_WO_CHAIN: {
+    unsigned IntNo = cast<ConstantSDNode>(N->getOperand(0))->getZExtValue();
+    switch (IntNo) {
+    default:
+      llvm_unreachable(
+          "Don't know how to custom type legalize this intrinsic!");
+    case Intrinsic::riscv_vmv_x_s: {
+      EVT VT = N->getValueType(0);
+      assert((VT == MVT::i8 || VT == MVT::i16 ||
+              (Subtarget.is64Bit() && VT == MVT::i32)) &&
+             "Unexpected custom legalisation!");
+      SDValue Extract = DAG.getNode(RISCVISD::VMV_X_S, DL,
+                                    Subtarget.getXLenVT(), N->getOperand(1));
+      Results.push_back(DAG.getNode(ISD::TRUNCATE, DL, VT, Extract));
+      break;
+    }
+    }
+    break;
+  }
   }
 }
 
 // A structure to hold one of the bit-manipulation patterns below. Together, a
 // SHL and non-SHL pattern may form a bit-manipulation pair on a single source:
 //   (or (and (shl x, 1), 0xAAAAAAAA),
 //       (and (srl x, 1), 0x55555555))
 struct RISCVBitmanipPat {
unsigned RISCVTargetLowering::ComputeNumSignBitsForTargetNode(
   case RISCVISD::GREVIW:
   case RISCVISD::GORCIW:
   case RISCVISD::FSLW:
   case RISCVISD::FSRW:
     // TODO: As the result is sign-extended, this is conservatively correct. A
     // more precise answer could be calculated for SRAW depending on known
     // bits in the shift amount.
     return 33;
+  case RISCVISD::VMV_X_S:
+    // The number of sign bits of the scalar result is computed by obtaining the
+    // element type of the input vector operand, substracting its width from the
+    // XLEN, and then adding one (sign bit within the element type).
+    return Subtarget.getXLen() - Op.getOperand(0).getScalarValueSizeInBits() + 1;
   }
 
   return 1;
 }
 
 static MachineBasicBlock *emitReadCycleWidePseudo(MachineInstr &MI,
                                                   MachineBasicBlock *BB) {
   assert(MI.getOpcode() == RISCV::ReadCycleWide && "Unexpected instruction");
#define NODE_NAME_CASE(NODE)                                                   \
   NODE_NAME_CASE(FMV_X_ANYEXTH)
   NODE_NAME_CASE(FMV_W_X_RV64)
   NODE_NAME_CASE(FMV_X_ANYEXTW_RV64)
   NODE_NAME_CASE(READ_CYCLE_WIDE)
   NODE_NAME_CASE(GREVI)
   NODE_NAME_CASE(GREVIW)
   NODE_NAME_CASE(GORCI)
   NODE_NAME_CASE(GORCIW)
+  NODE_NAME_CASE(VMV_X_S)
   }
   // clang-format on
   return nullptr;
 #undef NODE_NAME_CASE
 }
 
 /// getConstraintType - Given a constraint letter, return the type of
 /// constraint it is for this target.

llvm/lib/Target/RISCV/RISCVInstrInfoV.td

def VID_V : RVInstV<0b010100, 0b10001, OPMVV, (outs VR:$vd),
                       (ins VMaskOp:$vm), "vid.v", "$vd$vm"> {
   let vs2 = 0;
 }
 
 // Integer Scalar Move Instructions
 let vm = 1 in {
 def VMV_X_S : RVInstV<0b010000, 0b00000, OPMVV, (outs GPR:$vd),
                       (ins VR:$vs2), "vmv.x.s", "$vd, $vs2">;
-def VMV_S_X : RVInstV2<0b010000, 0b00000, OPMVX, (outs VR:$vd),
-                      (ins GPR:$rs1), "vmv.s.x", "$vd, $rs1">;
+let Constraints = "$vd = $vd_wb" in
+def VMV_S_X : RVInstV2<0b010000, 0b00000, OPMVX, (outs VR:$vd_wb),
+                      (ins VR:$vd, GPR:$rs1), "vmv.s.x", "$vd, $rs1">;
 
 }
 } // hasSideEffects = 0, mayLoad = 0, mayStore = 0
 } // Predicates = [HasStdExtV]
 
 let Predicates = [HasStdExtV, HasStdExtF] in {
 let hasSideEffects = 0, mayLoad = 0, mayStore = 0, vm = 1 in {
 // Floating-Point Scalar Move Instructions

llvm/lib/Target/RISCV/RISCVInstrInfoVPseudos.td

 //===-- RISCVInstrInfoVPseudos.td - RISC-V 'V' Pseudos -----*- tablegen -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 ///
 /// This file contains the required infrastructure to support code generation
 /// for the standard 'V' (Vector) extension, version 0.9.  This version is still
 /// experimental as the 'V' extension hasn't been ratified yet.
 ///
 /// This file is included from RISCVInstrInfoV.td
 ///
 //===----------------------------------------------------------------------===//
 
+def riscv_vmv_x_s : SDNode<"RISCVISD::VMV_X_S",
+                           SDTypeProfile<1, 1, [SDTCisInt<0>, SDTCisVec<1>,
+                                                SDTCisInt<1>]>>;
+
 // X0 has special meaning for vsetvl/vsetvli.
 //  rd | rs1 |   AVL value | Effect on vl
 //--------------------------------------------------------------
 // !X0 |  X0 |       VLMAX | Set vl to VLMAX
 //  X0 |  X0 | Value in vl | Keep current vl, just change vtype.
 def NoX0 : SDNodeXForm<undef,
 [{
   auto *C = dyn_cast<ConstantSDNode>(N);

 //===----------------------------------------------------------------------===//
 defm PseudoVFADD       : VPseudoBinaryV_VV_VX</*IsFloat=*/1>;
 defm PseudoVFSUB       : VPseudoBinaryV_VV_VX</*IsFloat=*/1>;
 defm PseudoVFRSUB      : VPseudoBinaryV_VX</*IsFloat=*/1>;
 
 } // Predicates = [HasStdExtV, HasStdExtF]
 
 //===----------------------------------------------------------------------===//
+// 17.1. Integer Scalar Move Instructions
+//===----------------------------------------------------------------------===//
+
+let Predicates = [HasStdExtV] in {
+let mayLoad = 0, mayStore = 0, hasSideEffects = 0, usesCustomInserter = 1,
+    Uses = [VL, VTYPE] in {
+  foreach m = MxList.m in {
+    let VLMul = m.value in {
+      let SEWIndex = 2, BaseInstr = VMV_X_S in
+      def PseudoVMV_X_S # "_" # m.MX: Pseudo<(outs GPR:$rd),
+                                             (ins m.vrclass:$rs2, ixlenimm:$sew),
+                                             []>, RISCVVPseudo;
+      let VLIndex = 3, SEWIndex = 4, BaseInstr = VMV_S_X,
+          Constraints = "$rd = $rs1" in
+      def PseudoVMV_S_X # "_" # m.MX: Pseudo<(outs m.vrclass:$rd),
+                                             (ins m.vrclass:$rs1, GPR:$rs2,
+                                             GPR:$vl, ixlenimm:$sew),
+                                             []>, RISCVVPseudo;
+    }
+  }
+}
+}
+
+//===----------------------------------------------------------------------===//
 // Patterns.
 //===----------------------------------------------------------------------===//
 let Predicates = [HasStdExtV] in {
 
 // Whole-register vector patterns.
 
 // 7.4. Vector Unit-Stride Instructions
 defm "" : VPatUSLoadStoreSDNodes<GPR>;

 //===----------------------------------------------------------------------===//
 // 14.2. Vector Single-Width Floating-Point Add/Subtract Instructions
 //===----------------------------------------------------------------------===//
 defm "" : VPatBinaryV_VV_VX<"int_riscv_vfadd", "PseudoVFADD", AllFloatVectors>;
 defm "" : VPatBinaryV_VV_VX<"int_riscv_vfsub", "PseudoVFSUB", AllFloatVectors>;
 defm "" : VPatBinaryV_VX<"int_riscv_vfrsub", "PseudoVFRSUB", AllFloatVectors>;
 
 } // Predicates = [HasStdExtV, HasStdExtF]
+
+//===----------------------------------------------------------------------===//
+// 17.1. Integer Scalar Move Instructions
+//===----------------------------------------------------------------------===//
+
+let Predicates = [HasStdExtV] in {
+foreach vti = AllIntegerVectors in {
+  def : Pat<(riscv_vmv_x_s (vti.Vector vti.RegClass:$rs2)),
+            (!cast<Instruction>("PseudoVMV_X_S_" # vti.LMul.MX) $rs2, vti.SEW)>;
+  def : Pat<(vti.Vector (int_riscv_vmv_s_x (vti.Vector vti.RegClass:$rs1),
+                                           GPR:$rs2, GPR:$vl)),
+            (!cast<Instruction>("PseudoVMV_S_X_" # vti.LMul.MX)
+             (vti.Vector $rs1), $rs2, (NoX0 GPR:$vl), vti.SEW)>;
+}
+} // Predicates = [HasStdExtV]

llvm/test/CodeGen/RISCV/rvv/vmv.s.x-rv32.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=riscv32 -mattr=+experimental-v -verify-machineinstrs < %s | FileCheck %s
+
+declare <vscale x 1 x i8> @llvm.riscv.vmv.s.x.nxv1i8(<vscale x 1 x i8>, i8, i32)
+
+define <vscale x 1 x i8> @intrinsic_vmv.s.x_x_nxv1i8(<vscale x 1 x i8> %0, i8 %1, i32 %2) nounwind {
+; CHECK-LABEL: intrinsic_vmv.s.x_x_nxv1i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a1, a1, e8,mf8,ta,mu
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x i8> @llvm.riscv.vmv.s.x.nxv1i8(<vscale x 1 x i8> %0, i8 %1, i32 %2)
+  ret <vscale x 1 x i8> %a
+}
+
+declare <vscale x 2 x i8> @llvm.riscv.vmv.s.x.nxv2i8(<vscale x 2 x i8>, i8, i32)
+
+define <vscale x 2 x i8> @intrinsic_vmv.s.x_x_nxv2i8(<vscale x 2 x i8> %0, i8 %1, i32 %2) nounwind {
+; CHECK-LABEL: intrinsic_vmv.s.x_x_nxv2i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a1, a1, e8,mf4,ta,mu
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x i8> @llvm.riscv.vmv.s.x.nxv2i8(<vscale x 2 x i8> %0, i8 %1, i32 %2)
+  ret <vscale x 2 x i8> %a
+}
+
+declare <vscale x 4 x i8> @llvm.riscv.vmv.s.x.nxv4i8(<vscale x 4 x i8>, i8, i32)
+
+define <vscale x 4 x i8> @intrinsic_vmv.s.x_x_nxv4i8(<vscale x 4 x i8> %0, i8 %1, i32 %2) nounwind {
+; CHECK-LABEL: intrinsic_vmv.s.x_x_nxv4i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a1, a1, e8,mf2,ta,mu
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x i8> @llvm.riscv.vmv.s.x.nxv4i8(<vscale x 4 x i8> %0, i8 %1, i32 %2)
+  ret <vscale x 4 x i8> %a
+}
+
+declare <vscale x 8 x i8> @llvm.riscv.vmv.s.x.nxv8i8(<vscale x 8 x i8>, i8, i32)
+
+define <vscale x 8 x i8> @intrinsic_vmv.s.x_x_nxv8i8(<vscale x 8 x i8> %0, i8 %1, i32 %2) nounwind {
+; CHECK-LABEL: intrinsic_vmv.s.x_x_nxv8i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a1, a1, e8,m1,ta,mu
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x i8> @llvm.riscv.vmv.s.x.nxv8i8(<vscale x 8 x i8> %0, i8 %1, i32 %2)
+  ret <vscale x 8 x i8> %a
+}
+
+declare <vscale x 16 x i8> @llvm.riscv.vmv.s.x.nxv16i8(<vscale x 16 x i8>, i8, i32)
+
+define <vscale x 16 x i8> @intrinsic_vmv.s.x_x_nxv16i8(<vscale x 16 x i8> %0, i8 %1, i32 %2) nounwind {
+; CHECK-LABEL: intrinsic_vmv.s.x_x_nxv16i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a1, a1, e8,m2,ta,mu
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x i8> @llvm.riscv.vmv.s.x.nxv16i8(<vscale x 16 x i8> %0, i8 %1, i32 %2)
+  ret <vscale x 16 x i8> %a
+}
+
+declare <vscale x 32 x i8> @llvm.riscv.vmv.s.x.nxv32i8(<vscale x 32 x i8>, i8, i32)
+
+define <vscale x 32 x i8> @intrinsic_vmv.s.x_x_nxv32i8(<vscale x 32 x i8> %0, i8 %1, i32 %2) nounwind {
+; CHECK-LABEL: intrinsic_vmv.s.x_x_nxv32i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a1, a1, e8,m4,ta,mu
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x i8> @llvm.riscv.vmv.s.x.nxv32i8(<vscale x 32 x i8> %0, i8 %1, i32 %2)
+  ret <vscale x 32 x i8> %a
+}
+
+declare <vscale x 64 x i8> @llvm.riscv.vmv.s.x.nxv64i8(<vscale x 64 x i8>, i8, i32)
+
+define <vscale x 64 x i8> @intrinsic_vmv.s.x_x_nxv64i8(<vscale x 64 x i8> %0, i8 %1, i32 %2) nounwind {
+; CHECK-LABEL: intrinsic_vmv.s.x_x_nxv64i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a1, a1, e8,m8,ta,mu
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 64 x i8> @llvm.riscv.vmv.s.x.nxv64i8(<vscale x 64 x i8> %0, i8 %1, i32 %2)
+  ret <vscale x 64 x i8> %a
+}
+
+declare <vscale x 1 x i16> @llvm.riscv.vmv.s.x.nxv1i16(<vscale x 1 x i16>, i16, i32)
+
+define <vscale x 1 x i16> @intrinsic_vmv.s.x_x_nxv1i16(<vscale x 1 x i16> %0, i16 %1, i32 %2) nounwind {
+; CHECK-LABEL: intrinsic_vmv.s.x_x_nxv1i16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a1, a1, e16,mf4,ta,mu
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x i16> @llvm.riscv.vmv.s.x.nxv1i16(<vscale x 1 x i16> %0, i16 %1, i32 %2)
+  ret <vscale x 1 x i16> %a
+}
+
+declare <vscale x 2 x i16> @llvm.riscv.vmv.s.x.nxv2i16(<vscale x 2 x i16>, i16, i32)
+
+define <vscale x 2 x i16> @intrinsic_vmv.s.x_x_nxv2i16(<vscale x 2 x i16> %0, i16 %1, i32 %2) nounwind {
+; CHECK-LABEL: intrinsic_vmv.s.x_x_nxv2i16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a1, a1, e16,mf2,ta,mu
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x i16> @llvm.riscv.vmv.s.x.nxv2i16(<vscale x 2 x i16> %0, i16 %1, i32 %2)
+  ret <vscale x 2 x i16> %a
+}
+
+declare <vscale x 4 x i16> @llvm.riscv.vmv.s.x.nxv4i16(<vscale x 4 x i16>, i16, i32)
+
+define <vscale x 4 x i16> @intrinsic_vmv.s.x_x_nxv4i16(<vscale x 4 x i16> %0, i16 %1, i32 %2) nounwind {
+; CHECK-LABEL: intrinsic_vmv.s.x_x_nxv4i16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a1, a1, e16,m1,ta,mu
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x i16> @llvm.riscv.vmv.s.x.nxv4i16(<vscale x 4 x i16> %0, i16 %1, i32 %2)
+  ret <vscale x 4 x i16> %a
+}
+
+declare <vscale x 8 x i16> @llvm.riscv.vmv.s.x.nxv8i16(<vscale x 8 x i16>, i16, i32)
+
+define <vscale x 8 x i16> @intrinsic_vmv.s.x_x_nxv8i16(<vscale x 8 x i16> %0, i16 %1, i32 %2) nounwind {
+; CHECK-LABEL: intrinsic_vmv.s.x_x_nxv8i16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a1, a1, e16,m2,ta,mu
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x i16> @llvm.riscv.vmv.s.x.nxv8i16(<vscale x 8 x i16> %0, i16 %1, i32 %2)
+  ret <vscale x 8 x i16> %a
+}
+
+declare <vscale x 16 x i16> @llvm.riscv.vmv.s.x.nxv16i16(<vscale x 16 x i16>, i16, i32)
+
+define <vscale x 16 x i16> @intrinsic_vmv.s.x_x_nxv16i16(<vscale x 16 x i16> %0, i16 %1, i32 %2) nounwind {
+; CHECK-LABEL: intrinsic_vmv.s.x_x_nxv16i16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a1, a1, e16,m4,ta,mu
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x i16> @llvm.riscv.vmv.s.x.nxv16i16(<vscale x 16 x i16> %0, i16 %1, i32 %2)
+  ret <vscale x 16 x i16> %a
+}
+
+declare <vscale x 32 x i16> @llvm.riscv.vmv.s.x.nxv32i16(<vscale x 32 x i16>, i16, i32)
+
+define <vscale x 32 x i16> @intrinsic_vmv.s.x_x_nxv32i16(<vscale x 32 x i16> %0, i16 %1, i32 %2) nounwind {
+; CHECK-LABEL: intrinsic_vmv.s.x_x_nxv32i16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a1, a1, e16,m8,ta,mu
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x i16> @llvm.riscv.vmv.s.x.nxv32i16(<vscale x 32 x i16> %0, i16 %1, i32 %2)
+  ret <vscale x 32 x i16> %a
+}
+
+declare <vscale x 1 x i32> @llvm.riscv.vmv.s.x.nxv1i32(<vscale x 1 x i32>, i32, i32)
+
+define <vscale x 1 x i32> @intrinsic_vmv.s.x_x_nxv1i32(<vscale x 1 x i32> %0, i32 %1, i32 %2) nounwind {
+; CHECK-LABEL: intrinsic_vmv.s.x_x_nxv1i32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a1, a1, e32,mf2,ta,mu
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x i32> @llvm.riscv.vmv.s.x.nxv1i32(<vscale x 1 x i32> %0, i32 %1, i32 %2)
+  ret <vscale x 1 x i32> %a
+}
+
+declare <vscale x 2 x i32> @llvm.riscv.vmv.s.x.nxv2i32(<vscale x 2 x i32>, i32, i32)
+
+define <vscale x 2 x i32> @intrinsic_vmv.s.x_x_nxv2i32(<vscale x 2 x i32> %0, i32 %1, i32 %2) nounwind {
+; CHECK-LABEL: intrinsic_vmv.s.x_x_nxv2i32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a1, a1, e32,m1,ta,mu
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x i32> @llvm.riscv.vmv.s.x.nxv2i32(<vscale x 2 x i32> %0, i32 %1, i32 %2)
+  ret <vscale x 2 x i32> %a
+}
+
+declare <vscale x 4 x i32> @llvm.riscv.vmv.s.x.nxv4i32(<vscale x 4 x i32>, i32, i32)
+
+define <vscale x 4 x i32> @intrinsic_vmv.s.x_x_nxv4i32(<vscale x 4 x i32> %0, i32 %1, i32 %2) nounwind {
+; CHECK-LABEL: intrinsic_vmv.s.x_x_nxv4i32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a1, a1, e32,m2,ta,mu
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x i32> @llvm.riscv.vmv.s.x.nxv4i32(<vscale x 4 x i32> %0, i32 %1, i32 %2)
+  ret <vscale x 4 x i32> %a
+}
+
+declare <vscale x 8 x i32> @llvm.riscv.vmv.s.x.nxv8i32(<vscale x 8 x i32>, i32, i32)
+
+define <vscale x 8 x i32> @intrinsic_vmv.s.x_x_nxv8i32(<vscale x 8 x i32> %0, i32 %1, i32 %2) nounwind {
+; CHECK-LABEL: intrinsic_vmv.s.x_x_nxv8i32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a1, a1, e32,m4,ta,mu
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x i32> @llvm.riscv.vmv.s.x.nxv8i32(<vscale x 8 x i32> %0, i32 %1, i32 %2)
+  ret <vscale x 8 x i32> %a
+}
+
+declare <vscale x 16 x i32> @llvm.riscv.vmv.s.x.nxv16i32(<vscale x 16 x i32>, i32, i32)
+
+define <vscale x 16 x i32> @intrinsic_vmv.s.x_x_nxv16i32(<vscale x 16 x i32> %0, i32 %1, i32 %2) nounwind {
+; CHECK-LABEL: intrinsic_vmv.s.x_x_nxv16i32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a1, a1, e32,m8,ta,mu
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x i32> @llvm.riscv.vmv.s.x.nxv16i32(<vscale x 16 x i32> %0, i32 %1, i32 %2)
+  ret <vscale x 16 x i32> %a
+}

llvm/test/CodeGen/RISCV/rvv/vmv.s.x-rv64.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=riscv64 -mattr=+experimental-v -verify-machineinstrs < %s | FileCheck %s
+
+declare <vscale x 1 x i8> @llvm.riscv.vmv.s.x.nxv1i8(<vscale x 1 x i8>, i8, i64);
+
+define <vscale x 1 x i8> @intrinsic_vmv.s.x_x_nxv1i8(<vscale x 1 x i8> %0, i8 %1, i64 %2) nounwind {
+; CHECK-LABEL: intrinsic_vmv.s.x_x_nxv1i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a1, a1, e8,mf8,ta,mu
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x i8> @llvm.riscv.vmv.s.x.nxv1i8(<vscale x 1 x i8> %0, i8 %1, i64 %2)
+  ret <vscale x 1 x i8> %a
+}
+
+declare <vscale x 2 x i8> @llvm.riscv.vmv.s.x.nxv2i8(<vscale x 2 x i8>, i8, i64);
+
+define <vscale x 2 x i8> @intrinsic_vmv.s.x_x_nxv2i8(<vscale x 2 x i8> %0, i8 %1, i64 %2) nounwind {
+; CHECK-LABEL: intrinsic_vmv.s.x_x_nxv2i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a1, a1, e8,mf4,ta,mu
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x i8> @llvm.riscv.vmv.s.x.nxv2i8(<vscale x 2 x i8> %0, i8 %1, i64 %2)
+  ret <vscale x 2 x i8> %a
+}
+
+declare <vscale x 4 x i8> @llvm.riscv.vmv.s.x.nxv4i8(<vscale x 4 x i8>, i8, i64);
+
+define <vscale x 4 x i8> @intrinsic_vmv.s.x_x_nxv4i8(<vscale x 4 x i8> %0, i8 %1, i64 %2) nounwind {
+; CHECK-LABEL: intrinsic_vmv.s.x_x_nxv4i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a1, a1, e8,mf2,ta,mu
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x i8> @llvm.riscv.vmv.s.x.nxv4i8(<vscale x 4 x i8> %0, i8 %1, i64 %2)
+  ret <vscale x 4 x i8> %a
+}
+
+declare <vscale x 8 x i8> @llvm.riscv.vmv.s.x.nxv8i8(<vscale x 8 x i8>, i8, i64);
+
+define <vscale x 8 x i8> @intrinsic_vmv.s.x_x_nxv8i8(<vscale x 8 x i8> %0, i8 %1, i64 %2) nounwind {
+; CHECK-LABEL: intrinsic_vmv.s.x_x_nxv8i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a1, a1, e8,m1,ta,mu
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x i8> @llvm.riscv.vmv.s.x.nxv8i8(<vscale x 8 x i8> %0, i8 %1, i64 %2)
+  ret <vscale x 8 x i8> %a
+}
+
+declare <vscale x 16 x i8> @llvm.riscv.vmv.s.x.nxv16i8(<vscale x 16 x i8>, i8, i64);
+
+define <vscale x 16 x i8> @intrinsic_vmv.s.x_x_nxv16i8(<vscale x 16 x i8> %0, i8 %1, i64 %2) nounwind {
+; CHECK-LABEL: intrinsic_vmv.s.x_x_nxv16i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a1, a1, e8,m2,ta,mu
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x i8> @llvm.riscv.vmv.s.x.nxv16i8(<vscale x 16 x i8> %0, i8 %1, i64 %2)
+  ret <vscale x 16 x i8> %a
+}
+
+declare <vscale x 32 x i8> @llvm.riscv.vmv.s.x.nxv32i8(<vscale x 32 x i8>, i8, i64);
+
+define <vscale x 32 x i8> @intrinsic_vmv.s.x_x_nxv32i8(<vscale x 32 x i8> %0, i8 %1, i64 %2) nounwind {
+; CHECK-LABEL: intrinsic_vmv.s.x_x_nxv32i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a1, a1, e8,m4,ta,mu
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x i8> @llvm.riscv.vmv.s.x.nxv32i8(<vscale x 32 x i8> %0, i8 %1, i64 %2)
+  ret <vscale x 32 x i8> %a
+}
+
+declare <vscale x 64 x i8> @llvm.riscv.vmv.s.x.nxv64i8(<vscale x 64 x i8>, i8, i64);
+
+define <vscale x 64 x i8> @intrinsic_vmv.s.x_x_nxv64i8(<vscale x 64 x i8> %0, i8 %1, i64 %2) nounwind {
+; CHECK-LABEL: intrinsic_vmv.s.x_x_nxv64i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a1, a1, e8,m8,ta,mu
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 64 x i8> @llvm.riscv.vmv.s.x.nxv64i8(<vscale x 64 x i8> %0, i8 %1, i64 %2)
+  ret <vscale x 64 x i8> %a
+}
+
+declare <vscale x 1 x i16> @llvm.riscv.vmv.s.x.nxv1i16(<vscale x 1 x i16>, i16, i64);
+
+define <vscale x 1 x i16> @intrinsic_vmv.s.x_x_nxv1i16(<vscale x 1 x i16> %0, i16 %1, i64 %2) nounwind {
+; CHECK-LABEL: intrinsic_vmv.s.x_x_nxv1i16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a1, a1, e16,mf4,ta,mu
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x i16> @llvm.riscv.vmv.s.x.nxv1i16(<vscale x 1 x i16> %0, i16 %1, i64 %2)
+  ret <vscale x 1 x i16> %a
+}
+
+declare <vscale x 2 x i16> @llvm.riscv.vmv.s.x.nxv2i16(<vscale x 2 x i16>, i16, i64);
+
+define <vscale x 2 x i16> @intrinsic_vmv.s.x_x_nxv2i16(<vscale x 2 x i16> %0, i16 %1, i64 %2) nounwind {
+; CHECK-LABEL: intrinsic_vmv.s.x_x_nxv2i16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a1, a1, e16,mf2,ta,mu
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x i16> @llvm.riscv.vmv.s.x.nxv2i16(<vscale x 2 x i16> %0, i16 %1, i64 %2)
+  ret <vscale x 2 x i16> %a
+}
+
+declare <vscale x 4 x i16> @llvm.riscv.vmv.s.x.nxv4i16(<vscale x 4 x i16>, i16, i64);
+
+define <vscale x 4 x i16> @intrinsic_vmv.s.x_x_nxv4i16(<vscale x 4 x i16> %0, i16 %1, i64 %2) nounwind {
+; CHECK-LABEL: intrinsic_vmv.s.x_x_nxv4i16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a1, a1, e16,m1,ta,mu
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x i16> @llvm.riscv.vmv.s.x.nxv4i16(<vscale x 4 x i16> %0, i16 %1, i64 %2)
+  ret <vscale x 4 x i16> %a
+}
+
+declare <vscale x 8 x i16> @llvm.riscv.vmv.s.x.nxv8i16(<vscale x 8 x i16>, i16, i64);
+
+define <vscale x 8 x i16> @intrinsic_vmv.s.x_x_nxv8i16(<vscale x 8 x i16> %0, i16 %1, i64 %2) nounwind {
+; CHECK-LABEL: intrinsic_vmv.s.x_x_nxv8i16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a1, a1, e16,m2,ta,mu
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x i16> @llvm.riscv.vmv.s.x.nxv8i16(<vscale x 8 x i16> %0, i16 %1, i64 %2)
+  ret <vscale x 8 x i16> %a
+}
+
+declare <vscale x 16 x i16> @llvm.riscv.vmv.s.x.nxv16i16(<vscale x 16 x i16>, i16, i64);
+
+define <vscale x 16 x i16> @intrinsic_vmv.s.x_x_nxv16i16(<vscale x 16 x i16> %0, i16 %1, i64 %2) nounwind {
+; CHECK-LABEL: intrinsic_vmv.s.x_x_nxv16i16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a1, a1, e16,m4,ta,mu
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x i16> @llvm.riscv.vmv.s.x.nxv16i16(<vscale x 16 x i16> %0, i16 %1, i64 %2)
+  ret <vscale x 16 x i16> %a
+}
+
+declare <vscale x 32 x i16> @llvm.riscv.vmv.s.x.nxv32i16(<vscale x 32 x i16>, i16, i64);
+
+define <vscale x 32 x i16> @intrinsic_vmv.s.x_x_nxv32i16(<vscale x 32 x i16> %0, i16 %1, i64 %2) nounwind {
+; CHECK-LABEL: intrinsic_vmv.s.x_x_nxv32i16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a1, a1, e16,m8,ta,mu
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 32 x i16> @llvm.riscv.vmv.s.x.nxv32i16(<vscale x 32 x i16> %0, i16 %1, i64 %2)
+  ret <vscale x 32 x i16> %a
+}
+
+declare <vscale x 1 x i32> @llvm.riscv.vmv.s.x.nxv1i32(<vscale x 1 x i32>, i32, i64);
+
+define <vscale x 1 x i32> @intrinsic_vmv.s.x_x_nxv1i32(<vscale x 1 x i32> %0, i32 %1, i64 %2) nounwind {
+; CHECK-LABEL: intrinsic_vmv.s.x_x_nxv1i32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a1, a1, e32,mf2,ta,mu
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x i32> @llvm.riscv.vmv.s.x.nxv1i32(<vscale x 1 x i32> %0, i32 %1, i64 %2)
+  ret <vscale x 1 x i32> %a
+}
+
+declare <vscale x 2 x i32> @llvm.riscv.vmv.s.x.nxv2i32(<vscale x 2 x i32>, i32, i64);
+
+define <vscale x 2 x i32> @intrinsic_vmv.s.x_x_nxv2i32(<vscale x 2 x i32> %0, i32 %1, i64 %2) nounwind {
+; CHECK-LABEL: intrinsic_vmv.s.x_x_nxv2i32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a1, a1, e32,m1,ta,mu
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x i32> @llvm.riscv.vmv.s.x.nxv2i32(<vscale x 2 x i32> %0, i32 %1, i64 %2)
+  ret <vscale x 2 x i32> %a
+}
+
+declare <vscale x 4 x i32> @llvm.riscv.vmv.s.x.nxv4i32(<vscale x 4 x i32>, i32, i64);
+
+define <vscale x 4 x i32> @intrinsic_vmv.s.x_x_nxv4i32(<vscale x 4 x i32> %0, i32 %1, i64 %2) nounwind {
+; CHECK-LABEL: intrinsic_vmv.s.x_x_nxv4i32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a1, a1, e32,m2,ta,mu
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x i32> @llvm.riscv.vmv.s.x.nxv4i32(<vscale x 4 x i32> %0, i32 %1, i64 %2)
+  ret <vscale x 4 x i32> %a
+}
+
+declare <vscale x 8 x i32> @llvm.riscv.vmv.s.x.nxv8i32(<vscale x 8 x i32>, i32, i64);
+
+define <vscale x 8 x i32> @intrinsic_vmv.s.x_x_nxv8i32(<vscale x 8 x i32> %0, i32 %1, i64 %2) nounwind {
+; CHECK-LABEL: intrinsic_vmv.s.x_x_nxv8i32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a1, a1, e32,m4,ta,mu
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x i32> @llvm.riscv.vmv.s.x.nxv8i32(<vscale x 8 x i32> %0, i32 %1, i64 %2)
+  ret <vscale x 8 x i32> %a
+}
+
+declare <vscale x 16 x i32> @llvm.riscv.vmv.s.x.nxv16i32(<vscale x 16 x i32>, i32, i64);
+
+define <vscale x 16 x i32> @intrinsic_vmv.s.x_x_nxv16i32(<vscale x 16 x i32> %0, i32 %1, i64 %2) nounwind {
+; CHECK-LABEL: intrinsic_vmv.s.x_x_nxv16i32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a1, a1, e32,m8,ta,mu
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 16 x i32> @llvm.riscv.vmv.s.x.nxv16i32(<vscale x 16 x i32> %0, i32 %1, i64 %2)
+  ret <vscale x 16 x i32> %a
+}
+
+declare <vscale x 1 x i64> @llvm.riscv.vmv.s.x.nxv1i64(<vscale x 1 x i64>, i64, i64);
+
+define <vscale x 1 x i64> @intrinsic_vmv.s.x_x_nxv1i64(<vscale x 1 x i64> %0, i64 %1, i64 %2) nounwind {
+; CHECK-LABEL: intrinsic_vmv.s.x_x_nxv1i64:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a1, a1, e64,m1,ta,mu
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 1 x i64> @llvm.riscv.vmv.s.x.nxv1i64(<vscale x 1 x i64> %0, i64 %1, i64 %2)
+  ret <vscale x 1 x i64> %a
+}
+
+declare <vscale x 2 x i64> @llvm.riscv.vmv.s.x.nxv2i64(<vscale x 2 x i64>, i64, i64);
+
+define <vscale x 2 x i64> @intrinsic_vmv.s.x_x_nxv2i64(<vscale x 2 x i64> %0, i64 %1, i64 %2) nounwind {
+; CHECK-LABEL: intrinsic_vmv.s.x_x_nxv2i64:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a1, a1, e64,m2,ta,mu
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 2 x i64> @llvm.riscv.vmv.s.x.nxv2i64(<vscale x 2 x i64> %0, i64 %1, i64 %2)
+  ret <vscale x 2 x i64> %a
+}
+
+declare <vscale x 4 x i64> @llvm.riscv.vmv.s.x.nxv4i64(<vscale x 4 x i64>, i64, i64);
+
+define <vscale x 4 x i64> @intrinsic_vmv.s.x_x_nxv4i64(<vscale x 4 x i64> %0, i64 %1, i64 %2) nounwind {
+; CHECK-LABEL: intrinsic_vmv.s.x_x_nxv4i64:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a1, a1, e64,m4,ta,mu
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 4 x i64> @llvm.riscv.vmv.s.x.nxv4i64(<vscale x 4 x i64> %0, i64 %1, i64 %2)
+  ret <vscale x 4 x i64> %a
+}
+
+declare <vscale x 8 x i64> @llvm.riscv.vmv.s.x.nxv8i64(<vscale x 8 x i64>, i64, i64);
+
+define <vscale x 8 x i64> @intrinsic_vmv.s.x_x_nxv8i64(<vscale x 8 x i64> %0, i64 %1, i64 %2) nounwind {
+; CHECK-LABEL: intrinsic_vmv.s.x_x_nxv8i64:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli a1, a1, e64,m8,ta,mu
+; CHECK-NEXT:    vmv.s.x v16, a0
+; CHECK-NEXT:    ret
+entry:
+  %a = call <vscale x 8 x i64> @llvm.riscv.vmv.s.x.nxv8i64(<vscale x 8 x i64> %0, i64 %1, i64 %2)
+  ret <vscale x 8 x i64> %a
+}

llvm/test/CodeGen/RISCV/rvv/vmv.x.s-rv32.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=riscv32 -mattr=+experimental-v -verify-machineinstrs < %s | FileCheck %s
+
+declare i8 @llvm.riscv.vmv.x.s.nxv1i8(<vscale x 1 x i8>)
+
+define signext i8 @intrinsic_vmv.x.s_s_nxv1i8(<vscale x 1 x i8> %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.x.s_s_nxv1i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, zero, e8,mf8,ta,mu
+; CHECK-NEXT:    vmv.x.s a0, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call i8 @llvm.riscv.vmv.x.s.nxv1i8(<vscale x 1 x i8> %0)
+  ret i8 %a
+}
+
+declare i8 @llvm.riscv.vmv.x.s.nxv2i8(<vscale x 2 x i8>)
+
+define signext i8 @intrinsic_vmv.x.s_s_nxv2i8(<vscale x 2 x i8> %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.x.s_s_nxv2i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, zero, e8,mf4,ta,mu
+; CHECK-NEXT:    vmv.x.s a0, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call i8 @llvm.riscv.vmv.x.s.nxv2i8(<vscale x 2 x i8> %0)
+  ret i8 %a
+}
+
+declare i8 @llvm.riscv.vmv.x.s.nxv4i8(<vscale x 4 x i8>)
+
+define signext i8 @intrinsic_vmv.x.s_s_nxv4i8(<vscale x 4 x i8> %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.x.s_s_nxv4i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, zero, e8,mf2,ta,mu
+; CHECK-NEXT:    vmv.x.s a0, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call i8 @llvm.riscv.vmv.x.s.nxv4i8(<vscale x 4 x i8> %0)
+  ret i8 %a
+}
+
+declare i8 @llvm.riscv.vmv.x.s.nxv8i8(<vscale x 8 x i8>)
+
+define signext i8 @intrinsic_vmv.x.s_s_nxv8i8(<vscale x 8 x i8> %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.x.s_s_nxv8i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, zero, e8,m1,ta,mu
+; CHECK-NEXT:    vmv.x.s a0, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call i8 @llvm.riscv.vmv.x.s.nxv8i8(<vscale x 8 x i8> %0)
+  ret i8 %a
+}
+
+declare i8 @llvm.riscv.vmv.x.s.nxv16i8(<vscale x 16 x i8>)
+
+define signext i8 @intrinsic_vmv.x.s_s_nxv16i8(<vscale x 16 x i8> %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.x.s_s_nxv16i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, zero, e8,m2,ta,mu
+; CHECK-NEXT:    vmv.x.s a0, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call i8 @llvm.riscv.vmv.x.s.nxv16i8(<vscale x 16 x i8> %0)
+  ret i8 %a
+}
+
+declare i8 @llvm.riscv.vmv.x.s.nxv32i8(<vscale x 32 x i8>)
+
+define signext i8 @intrinsic_vmv.x.s_s_nxv32i8(<vscale x 32 x i8> %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.x.s_s_nxv32i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, zero, e8,m4,ta,mu
+; CHECK-NEXT:    vmv.x.s a0, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call i8 @llvm.riscv.vmv.x.s.nxv32i8(<vscale x 32 x i8> %0)
+  ret i8 %a
+}
+
+declare i8 @llvm.riscv.vmv.x.s.nxv64i8(<vscale x 64 x i8>)
+
+define signext i8 @intrinsic_vmv.x.s_s_nxv64i8(<vscale x 64 x i8> %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.x.s_s_nxv64i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, zero, e8,m8,ta,mu
+; CHECK-NEXT:    vmv.x.s a0, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call i8 @llvm.riscv.vmv.x.s.nxv64i8(<vscale x 64 x i8> %0)
+  ret i8 %a
+}
+
+declare i16 @llvm.riscv.vmv.x.s.nxv1i16(<vscale x 1 x i16>)
+
+define signext i16 @intrinsic_vmv.x.s_s_nxv1i16(<vscale x 1 x i16> %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.x.s_s_nxv1i16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, zero, e16,mf4,ta,mu
+; CHECK-NEXT:    vmv.x.s a0, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call i16 @llvm.riscv.vmv.x.s.nxv1i16(<vscale x 1 x i16> %0)
+  ret i16 %a
+}
+
+declare i16 @llvm.riscv.vmv.x.s.nxv2i16(<vscale x 2 x i16>)
+
+define signext i16 @intrinsic_vmv.x.s_s_nxv2i16(<vscale x 2 x i16> %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.x.s_s_nxv2i16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, zero, e16,mf2,ta,mu
+; CHECK-NEXT:    vmv.x.s a0, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call i16 @llvm.riscv.vmv.x.s.nxv2i16(<vscale x 2 x i16> %0)
+  ret i16 %a
+}
+
+declare i16 @llvm.riscv.vmv.x.s.nxv4i16(<vscale x 4 x i16>)
+
+define signext i16 @intrinsic_vmv.x.s_s_nxv4i16(<vscale x 4 x i16> %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.x.s_s_nxv4i16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, zero, e16,m1,ta,mu
+; CHECK-NEXT:    vmv.x.s a0, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call i16 @llvm.riscv.vmv.x.s.nxv4i16(<vscale x 4 x i16> %0)
+  ret i16 %a
+}
+
+declare i16 @llvm.riscv.vmv.x.s.nxv8i16(<vscale x 8 x i16>)
+
+define signext i16 @intrinsic_vmv.x.s_s_nxv8i16(<vscale x 8 x i16> %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.x.s_s_nxv8i16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, zero, e16,m2,ta,mu
+; CHECK-NEXT:    vmv.x.s a0, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call i16 @llvm.riscv.vmv.x.s.nxv8i16(<vscale x 8 x i16> %0)
+  ret i16 %a
+}
+
+declare i16 @llvm.riscv.vmv.x.s.nxv16i16(<vscale x 16 x i16>)
+
+define signext i16 @intrinsic_vmv.x.s_s_nxv16i16(<vscale x 16 x i16> %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.x.s_s_nxv16i16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, zero, e16,m4,ta,mu
+; CHECK-NEXT:    vmv.x.s a0, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call i16 @llvm.riscv.vmv.x.s.nxv16i16( <vscale x 16 x i16> %0)
+  ret i16 %a
+}
+
+declare i16 @llvm.riscv.vmv.x.s.nxv32i16( <vscale x 32 x i16>)
+
+define signext i16 @intrinsic_vmv.x.s_s_nxv32i16(<vscale x 32 x i16> %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.x.s_s_nxv32i16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, zero, e16,m8,ta,mu
+; CHECK-NEXT:    vmv.x.s a0, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call i16 @llvm.riscv.vmv.x.s.nxv32i16( <vscale x 32 x i16> %0)
+  ret i16 %a
+}
+
+declare i32 @llvm.riscv.vmv.x.s.nxv1i32( <vscale x 1 x i32>)
+
+define i32 @intrinsic_vmv.x.s_s_nxv1i32(<vscale x 1 x i32> %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.x.s_s_nxv1i32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, zero, e32,mf2,ta,mu
+; CHECK-NEXT:    vmv.x.s a0, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call i32 @llvm.riscv.vmv.x.s.nxv1i32( <vscale x 1 x i32> %0)
+  ret i32 %a
+}
+
+declare i32 @llvm.riscv.vmv.x.s.nxv2i32( <vscale x 2 x i32>)
+
+define i32 @intrinsic_vmv.x.s_s_nxv2i32(<vscale x 2 x i32> %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.x.s_s_nxv2i32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, zero, e32,m1,ta,mu
+; CHECK-NEXT:    vmv.x.s a0, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call i32 @llvm.riscv.vmv.x.s.nxv2i32( <vscale x 2 x i32> %0)
+  ret i32 %a
+}
+
+declare i32 @llvm.riscv.vmv.x.s.nxv4i32( <vscale x 4 x i32>)
+
+define i32 @intrinsic_vmv.x.s_s_nxv4i32(<vscale x 4 x i32> %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.x.s_s_nxv4i32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, zero, e32,m2,ta,mu
+; CHECK-NEXT:    vmv.x.s a0, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call i32 @llvm.riscv.vmv.x.s.nxv4i32( <vscale x 4 x i32> %0)
+  ret i32 %a
+}
+
+declare i32 @llvm.riscv.vmv.x.s.nxv8i32( <vscale x 8 x i32>)
+
+define i32 @intrinsic_vmv.x.s_s_nxv8i32(<vscale x 8 x i32> %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.x.s_s_nxv8i32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, zero, e32,m4,ta,mu
+; CHECK-NEXT:    vmv.x.s a0, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call i32 @llvm.riscv.vmv.x.s.nxv8i32( <vscale x 8 x i32> %0)
+  ret i32 %a
+}
+
+declare i32 @llvm.riscv.vmv.x.s.nxv16i32( <vscale x 16 x i32>)
+
+define i32 @intrinsic_vmv.x.s_s_nxv16i32(<vscale x 16 x i32> %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.x.s_s_nxv16i32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, zero, e32,m8,ta,mu
+; CHECK-NEXT:    vmv.x.s a0, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call i32 @llvm.riscv.vmv.x.s.nxv16i32( <vscale x 16 x i32> %0)
+  ret i32 %a
+}

llvm/test/CodeGen/RISCV/rvv/vmv.x.s-rv64.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=riscv64 -mattr=+experimental-v -verify-machineinstrs < %s | FileCheck %s
+
+declare i8 @llvm.riscv.vmv.x.s.nxv1i8(<vscale x 1 x i8>)
+
+define signext i8 @intrinsic_vmv.x.s_s_nxv1i8(<vscale x 1 x i8> %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.x.s_s_nxv1i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, zero, e8,mf8,ta,mu
+; CHECK-NEXT:    vmv.x.s a0, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call i8 @llvm.riscv.vmv.x.s.nxv1i8(<vscale x 1 x i8> %0)
+  ret i8 %a
+}
+
+declare i8 @llvm.riscv.vmv.x.s.nxv2i8(<vscale x 2 x i8>)
+
+define signext i8 @intrinsic_vmv.x.s_s_nxv2i8(<vscale x 2 x i8> %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.x.s_s_nxv2i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, zero, e8,mf4,ta,mu
+; CHECK-NEXT:    vmv.x.s a0, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call i8 @llvm.riscv.vmv.x.s.nxv2i8(<vscale x 2 x i8> %0)
+  ret i8 %a
+}
+
+declare i8 @llvm.riscv.vmv.x.s.nxv4i8(<vscale x 4 x i8>)
+
+define signext i8 @intrinsic_vmv.x.s_s_nxv4i8(<vscale x 4 x i8> %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.x.s_s_nxv4i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, zero, e8,mf2,ta,mu
+; CHECK-NEXT:    vmv.x.s a0, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call i8 @llvm.riscv.vmv.x.s.nxv4i8(<vscale x 4 x i8> %0)
+  ret i8 %a
+}
+
+declare i8 @llvm.riscv.vmv.x.s.nxv8i8(<vscale x 8 x i8>)
+
+define signext i8 @intrinsic_vmv.x.s_s_nxv8i8(<vscale x 8 x i8> %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.x.s_s_nxv8i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, zero, e8,m1,ta,mu
+; CHECK-NEXT:    vmv.x.s a0, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call i8 @llvm.riscv.vmv.x.s.nxv8i8(<vscale x 8 x i8> %0)
+  ret i8 %a
+}
+
+declare i8 @llvm.riscv.vmv.x.s.nxv16i8(<vscale x 16 x i8>)
+
+define signext i8 @intrinsic_vmv.x.s_s_nxv16i8(<vscale x 16 x i8> %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.x.s_s_nxv16i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, zero, e8,m2,ta,mu
+; CHECK-NEXT:    vmv.x.s a0, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call i8 @llvm.riscv.vmv.x.s.nxv16i8(<vscale x 16 x i8> %0)
+  ret i8 %a
+}
+
+declare i8 @llvm.riscv.vmv.x.s.nxv32i8(<vscale x 32 x i8>)
+
+define signext i8 @intrinsic_vmv.x.s_s_nxv32i8(<vscale x 32 x i8> %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.x.s_s_nxv32i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, zero, e8,m4,ta,mu
+; CHECK-NEXT:    vmv.x.s a0, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call i8 @llvm.riscv.vmv.x.s.nxv32i8(<vscale x 32 x i8> %0)
+  ret i8 %a
+}
+
+declare i8 @llvm.riscv.vmv.x.s.nxv64i8(<vscale x 64 x i8>)
+
+define signext i8 @intrinsic_vmv.x.s_s_nxv64i8(<vscale x 64 x i8> %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.x.s_s_nxv64i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, zero, e8,m8,ta,mu
+; CHECK-NEXT:    vmv.x.s a0, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call i8 @llvm.riscv.vmv.x.s.nxv64i8(<vscale x 64 x i8> %0)
+  ret i8 %a
+}
+
+declare i16 @llvm.riscv.vmv.x.s.nxv1i16(<vscale x 1 x i16>)
+
+define signext i16 @intrinsic_vmv.x.s_s_nxv1i16(<vscale x 1 x i16> %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.x.s_s_nxv1i16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, zero, e16,mf4,ta,mu
+; CHECK-NEXT:    vmv.x.s a0, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call i16 @llvm.riscv.vmv.x.s.nxv1i16(<vscale x 1 x i16> %0)
+  ret i16 %a
+}
+
+declare i16 @llvm.riscv.vmv.x.s.nxv2i16(<vscale x 2 x i16>)
+
+define signext i16 @intrinsic_vmv.x.s_s_nxv2i16(<vscale x 2 x i16> %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.x.s_s_nxv2i16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, zero, e16,mf2,ta,mu
+; CHECK-NEXT:    vmv.x.s a0, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call i16 @llvm.riscv.vmv.x.s.nxv2i16(<vscale x 2 x i16> %0)
+  ret i16 %a
+}
+
+declare i16 @llvm.riscv.vmv.x.s.nxv4i16(<vscale x 4 x i16>)
+
+define signext i16 @intrinsic_vmv.x.s_s_nxv4i16(<vscale x 4 x i16> %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.x.s_s_nxv4i16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, zero, e16,m1,ta,mu
+; CHECK-NEXT:    vmv.x.s a0, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call i16 @llvm.riscv.vmv.x.s.nxv4i16(<vscale x 4 x i16> %0)
+  ret i16 %a
+}
+
+declare i16 @llvm.riscv.vmv.x.s.nxv8i16(<vscale x 8 x i16>)
+
+define signext i16 @intrinsic_vmv.x.s_s_nxv8i16(<vscale x 8 x i16> %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.x.s_s_nxv8i16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, zero, e16,m2,ta,mu
+; CHECK-NEXT:    vmv.x.s a0, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call i16 @llvm.riscv.vmv.x.s.nxv8i16(<vscale x 8 x i16> %0)
+  ret i16 %a
+}
+
+declare i16 @llvm.riscv.vmv.x.s.nxv16i16(<vscale x 16 x i16>)
+
+define signext i16 @intrinsic_vmv.x.s_s_nxv16i16(<vscale x 16 x i16> %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.x.s_s_nxv16i16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, zero, e16,m4,ta,mu
+; CHECK-NEXT:    vmv.x.s a0, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call i16 @llvm.riscv.vmv.x.s.nxv16i16( <vscale x 16 x i16> %0)
+  ret i16 %a
+}
+
+declare i16 @llvm.riscv.vmv.x.s.nxv32i16( <vscale x 32 x i16>)
+
+define signext i16 @intrinsic_vmv.x.s_s_nxv32i16(<vscale x 32 x i16> %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.x.s_s_nxv32i16:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, zero, e16,m8,ta,mu
+; CHECK-NEXT:    vmv.x.s a0, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call i16 @llvm.riscv.vmv.x.s.nxv32i16( <vscale x 32 x i16> %0)
+  ret i16 %a
+}
+
+declare i32 @llvm.riscv.vmv.x.s.nxv1i32( <vscale x 1 x i32>)
+
+define signext i32 @intrinsic_vmv.x.s_s_nxv1i32(<vscale x 1 x i32> %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.x.s_s_nxv1i32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, zero, e32,mf2,ta,mu
+; CHECK-NEXT:    vmv.x.s a0, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call i32 @llvm.riscv.vmv.x.s.nxv1i32( <vscale x 1 x i32> %0)
+  ret i32 %a
+}
+
+declare i32 @llvm.riscv.vmv.x.s.nxv2i32( <vscale x 2 x i32>)
+
+define signext i32 @intrinsic_vmv.x.s_s_nxv2i32(<vscale x 2 x i32> %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.x.s_s_nxv2i32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, zero, e32,m1,ta,mu
+; CHECK-NEXT:    vmv.x.s a0, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call i32 @llvm.riscv.vmv.x.s.nxv2i32( <vscale x 2 x i32> %0)
+  ret i32 %a
+}
+
+declare i32 @llvm.riscv.vmv.x.s.nxv4i32( <vscale x 4 x i32>)
+
+define signext i32 @intrinsic_vmv.x.s_s_nxv4i32(<vscale x 4 x i32> %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.x.s_s_nxv4i32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, zero, e32,m2,ta,mu
+; CHECK-NEXT:    vmv.x.s a0, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call i32 @llvm.riscv.vmv.x.s.nxv4i32( <vscale x 4 x i32> %0)
+  ret i32 %a
+}
+
+declare i32 @llvm.riscv.vmv.x.s.nxv8i32( <vscale x 8 x i32>)
+
+define signext i32 @intrinsic_vmv.x.s_s_nxv8i32(<vscale x 8 x i32> %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.x.s_s_nxv8i32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, zero, e32,m4,ta,mu
+; CHECK-NEXT:    vmv.x.s a0, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call i32 @llvm.riscv.vmv.x.s.nxv8i32( <vscale x 8 x i32> %0)
+  ret i32 %a
+}
+
+declare i32 @llvm.riscv.vmv.x.s.nxv16i32( <vscale x 16 x i32>)
+
+define signext i32 @intrinsic_vmv.x.s_s_nxv16i32(<vscale x 16 x i32> %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.x.s_s_nxv16i32:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, zero, e32,m8,ta,mu
+; CHECK-NEXT:    vmv.x.s a0, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call i32 @llvm.riscv.vmv.x.s.nxv16i32( <vscale x 16 x i32> %0)
+  ret i32 %a
+}
+
+declare i64 @llvm.riscv.vmv.x.s.nxv1i64( <vscale x 1 x i64>)
+
+define i64 @intrinsic_vmv.x.s_s_nxv1i64(<vscale x 1 x i64> %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.x.s_s_nxv1i64:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, zero, e64,m1,ta,mu
+; CHECK-NEXT:    vmv.x.s a0, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call i64 @llvm.riscv.vmv.x.s.nxv1i64( <vscale x 1 x i64> %0)
+  ret i64 %a
+}
+
+declare i64 @llvm.riscv.vmv.x.s.nxv2i64( <vscale x 2 x i64>)
+
+define i64 @intrinsic_vmv.x.s_s_nxv2i64(<vscale x 2 x i64> %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.x.s_s_nxv2i64:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, zero, e64,m2,ta,mu
+; CHECK-NEXT:    vmv.x.s a0, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call i64 @llvm.riscv.vmv.x.s.nxv2i64( <vscale x 2 x i64> %0)
+  ret i64 %a
+}
+
+declare i64 @llvm.riscv.vmv.x.s.nxv4i64( <vscale x 4 x i64>)
+
+define i64 @intrinsic_vmv.x.s_s_nxv4i64(<vscale x 4 x i64> %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.x.s_s_nxv4i64:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, zero, e64,m4,ta,mu
+; CHECK-NEXT:    vmv.x.s a0, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call i64 @llvm.riscv.vmv.x.s.nxv4i64( <vscale x 4 x i64> %0)
+  ret i64 %a
+}
+
+declare i64 @llvm.riscv.vmv.x.s.nxv8i64(<vscale x 8 x i64>)
+
+define i64 @intrinsic_vmv.x.s_s_nxv8i64(<vscale x 8 x i64> %0) nounwind {
+; CHECK-LABEL: intrinsic_vmv.x.s_s_nxv8i64:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vsetvli zero, zero, e64,m8,ta,mu
+; CHECK-NEXT:    vmv.x.s a0, v16
+; CHECK-NEXT:    ret
+entry:
+  %a = call i64 @llvm.riscv.vmv.x.s.nxv8i64(<vscale x 8 x i64> %0)
+  ret i64 %a
+}