[RISCV] Splat scalar to be of length VL instead of 1 for reductions

[• pcwang-thead]

Summary

Correct me if I am wrong, I think there is minmal microarchitecture
difference between vmv.s.x with vl==1 and v(f)mv.v.(ixf) with
vl==avl. So we can just splat scalar to be of length VL instead
of 1 to reduce context switch of vtype.

For VP reductions, we can do this iff we can prove AVL is non-zero.

To not increase register pressure, we do this iff LMUL <= 1.

[• pcwang-thead]

llvm/test/CodeGen/RISCV/rvv/fixed-vectors-reduction-fp.ll

41

Floating-point regressions here.

1366

Spilling because of increasing of register pressure.

[• pcwang-thead]

Only for LMUL <= 1.

[• pcwang-thead]

Make sure operand's types are of LMUL=1 types

[craig.topper]

llvm/lib/Target/RISCV/RISCVISelLowering.cpp

5724

Move this into an else. We shouldn't create nodes if they are going to end up dead.

[• pcwang-thead]

• Address comment.
• Fix floating-point regressions since this change will prevent combining fadd to reduce.

[reames]

Comment inline with a problematic case.

Note as well that InsertVSETVLI should already be eliding the vsetvli when we can prove the avl non-zero.

llvm/test/CodeGen/RISCV/rvv/fixed-vectors-reduction-fp-vp.ll

12–14

There's a subtle semantic distinction between the old and proposed code here where a0 is zero.

The old code would unconditionally insert the neutral element into v9, and then the vfredusum would see a VL=0, and not update the destination register. As a result, the final return value is the neutral element.

The new code leaves v9 unchanged, and thus the result is whatever lane 0 of the v9 register happened to contain previously.

[• pcwang-thead]

For VP reductions, we can do this iff we can prove AVL is non-zero.

[• pcwang-thead]

Add tests for VP reductions with known non-zero avl.

[• pcwang-thead]

Add missing tests.

[• pcwang-thead]

Ping :-)

[• pcwang-thead]

Gentle ping. :-)

[craig.topper]

llvm/lib/Target/RISCV/RISCVISelLowering.cpp

1881

AVL is unsigned. Why getSExtValue?

5727

Add curly braces. LLVM coding standards say that if/else should both uses braces if one does.

5797

Curly braces

5873

Curly braces

llvm/test/CodeGen/RISCV/rvv/fixed-vectors-reduction-fp.ll

370

Why is this e32,m1 instead of e32, mf2?

[• pcwang-thead]

• Address comments.
• Fix incorrect EVT when combining to reductions.

[• pcwang-thead]

llvm/test/CodeGen/RISCV/rvv/fixed-vectors-reduction-fp.ll

370

Good catch! Fixed.

[reames]

I'm looking at your test changes - not the code - and it really seems like the sole benefit of this here is vsetvli insertion. I'd default to expecting this to belong inside RISCVInsertVSETVLI.cpp not during ISEL.

From what I can tell glancing at the test deltas, you're effectively proposing the following rules:

• A scalar insert w/TA demands the "non-zeroness" of the AVL.
• A splat or broadcast load w/TA demands the an equal or greater AVL (with the same LMUL).

From what I can tell, we have no reason to prefer the v.x/v.i/v.f form over the s.x/s.f forms. The changes in your diff appear to be an artifact of how you implemented the above rules, not a benefit on their own. Unless you disagree with this, I would ask that you rework the patch and patch description to make that obvious.

Worth noting is that even if you decide to keep this in ISEL, there's nothing about the rules above which are specific to reductions.

[• pcwang-thead]

In D137530#3978568, @reames wrote:

I'm looking at your test changes - not the code - and it really seems like the sole benefit of this here is vsetvli insertion.

Yes, you're right. My intention is to reduce vsetvlis.

I'd default to expecting this to belong inside RISCVInsertVSETVLI.cpp not during ISEL.

Thanks for your suggestions, I will do some works on RISCVInsertVSETVLI.cpp to see if it is beneficial.

From what I can tell glancing at the test deltas, you're effectively proposing the following rules:

• A scalar insert w/TA demands the "non-zeroness" of the AVL.
• A splat or broadcast load w/TA demands the an equal or greater AVL (with the same LMUL).

From what I can tell, we have no reason to prefer the v.x/v.i/v.f form over the s.x/s.f forms. The changes in your diff appear to be an artifact of how you implemented the above rules, not a benefit on their own. Unless you disagree with this, I would ask that you rework the patch and patch description to make that obvious.

Worth noting is that even if you decide to keep this in ISEL, there's nothing about the rules above which are specific to reductions.

I have no strong opinion on this. Let's see what we can do in RISCVInsertVSETVLI.cpp first. :-)

[reames]

I sat down yesterday to prototype the approach I had suggested, and basically end up convincing myself I was wrong. Not from a theoretical standpoint on the vsetvli toggles, but from the fact we really do want to change instruction selection in some cases as well.

In that vein, I wrote up a small patch series (D139656) which tries to work towards this objective, and apply the same logic more broadly. I also landed two small changes to InsertVSETVLI to better handle vmv.s.x idioms exposed by these patches. I'm curious what you think of those.

[reames]

FYI, D139747 implements a strict subset of this patch.

[reames]

Can you rebase? I think that most if not all, of the changes from this have now landed in other patches. If you think there's anything left, a rebase would help make that much more obvious.

[• pcwang-thead]

I think there is no left changes.
Thanks. @reames

llvm/lib/Target/RISCV/RISCVISelLowering.cpp

getDefaultScalableVLOps(MVT VecVT, SDLoc DL, SelectionDAG &DAG,
   return getDefaultVLOps(VecVT, VecVT, DL, DAG, Subtarget);
 }
 
 // The state of RVV BUILD_VECTOR and VECTOR_SHUFFLE lowering is that very few
 // of either is (currently) supported. This can get us into an infinite loop
 // where we try to lower a BUILD_VECTOR as a VECTOR_SHUFFLE as a BUILD_VECTOR
 // as a ..., etc.
 // Until either (or both) of these can reliably lower any node, reporting that
 // we don't want to expand BUILD_VECTORs via VECTOR_SHUFFLEs at least breaks
 // the infinite loop. Note that this lowers BUILD_VECTOR through the stack,
 // which is not desirable.
 bool RISCVTargetLowering::shouldExpandBuildVectorWithShuffles(
     EVT VT, unsigned DefinedValues) const {
   return false;
 }
 
 static SDValue lowerFP_TO_INT_SAT(SDValue Op, SelectionDAG &DAG,
SDValue RISCVTargetLowering::lowerVECREDUCE(SDValue Op,
   MVT M1VT = getLMUL1VT(ContainerVT);
   MVT XLenVT = Subtarget.getXLenVT();
 
   auto [Mask, VL] = getDefaultVLOps(VecVT, ContainerVT, DL, DAG, Subtarget);
 
   SDValue NeutralElem =
       DAG.getNeutralElement(BaseOpc, DL, VecEltVT, SDNodeFlags());
   SDValue IdentitySplat =
       lowerScalarSplat(SDValue(), NeutralElem, DAG.getConstant(1, DL, XLenVT),
                        M1VT, DL, DAG, Subtarget);
+  if (ElementCount::isKnownLE(ContainerVT.getVectorElementCount(),
+                              M1VT.getVectorElementCount())) {
+    IdentitySplat = lowerScalarSplat(SDValue(), NeutralElem, VL, ContainerVT,
+                                     DL, DAG, Subtarget);
+    IdentitySplat =
+        DAG.getNode(ISD::INSERT_SUBVECTOR, DL, M1VT, DAG.getUNDEF(M1VT),
+                    IdentitySplat, DAG.getConstant(0, DL, XLenVT));
+  }
   SDValue Reduction = DAG.getNode(RVVOpcode, DL, M1VT, DAG.getUNDEF(M1VT), Vec,
                                   IdentitySplat, Mask, VL);
   SDValue Elt0 = DAG.getNode(ISD::EXTRACT_VECTOR_ELT, DL, VecEltVT, Reduction,
                              DAG.getConstant(0, DL, XLenVT));
   return DAG.getSExtOrTrunc(Elt0, DL, Op.getValueType());
 }
 
 // Given a reduction op, this function returns the matching reduction opcode,
SDValue RISCVTargetLowering::lowerFPVECREDUCE(SDValue Op,
   MVT VecVT = VectorVal.getSimpleValueType();
 
   MVT ContainerVT = VecVT;
   if (VecVT.isFixedLengthVector()) {
     ContainerVT = getContainerForFixedLengthVector(VecVT);
     VectorVal = convertToScalableVector(ContainerVT, VectorVal, DAG, Subtarget);
   }
 
-  MVT M1VT = getLMUL1VT(VectorVal.getSimpleValueType());
+  MVT M1VT = getLMUL1VT(ContainerVT);
   MVT XLenVT = Subtarget.getXLenVT();
 
   auto [Mask, VL] = getDefaultVLOps(VecVT, ContainerVT, DL, DAG, Subtarget);
 
   SDValue ScalarSplat =
       lowerScalarSplat(SDValue(), ScalarVal, DAG.getConstant(1, DL, XLenVT),
                        M1VT, DL, DAG, Subtarget);
+  if (ElementCount::isKnownLE(ContainerVT.getVectorElementCount(),
+                              M1VT.getVectorElementCount())) {
+    ScalarVal = lowerScalarSplat(SDValue(), ScalarVal, VL, ContainerVT, DL, DAG,
+                                 Subtarget);
+    ScalarVal = DAG.getNode(ISD::INSERT_SUBVECTOR, DL, M1VT, DAG.getUNDEF(M1VT),
+                            ScalarVal, DAG.getConstant(0, DL, XLenVT));
+  }
   SDValue Reduction = DAG.getNode(RVVOpcode, DL, M1VT, DAG.getUNDEF(M1VT),
                                   VectorVal, ScalarSplat, Mask, VL);
   return DAG.getNode(ISD::EXTRACT_VECTOR_ELT, DL, VecEltVT, Reduction,
                      DAG.getConstant(0, DL, XLenVT));
 }
 
 static unsigned getRVVVPReductionOp(unsigned ISDOpcode) {
   switch (ISDOpcode) {
SDValue RISCVTargetLowering::lowerVPREDUCE(SDValue Op,
 
   MVT M1VT = getLMUL1VT(ContainerVT);
   MVT XLenVT = Subtarget.getXLenVT();
   MVT ResVT = !VecVT.isInteger() || VecEltVT.bitsGE(XLenVT) ? VecEltVT : XLenVT;
 
   SDValue StartSplat = lowerScalarSplat(SDValue(), Op.getOperand(0),
                                         DAG.getConstant(1, DL, XLenVT), M1VT,
                                         DL, DAG, Subtarget);
+  if (ElementCount::isKnownLE(ContainerVT.getVectorElementCount(),
+                              M1VT.getVectorElementCount())) {
+    StartSplat = lowerScalarSplat(SDValue(), Op.getOperand(0), VL, ContainerVT,
+                                  DL, DAG, Subtarget);
+    StartSplat =
+        DAG.getNode(ISD::INSERT_SUBVECTOR, DL, M1VT, DAG.getUNDEF(M1VT),
+                    StartSplat, DAG.getConstant(0, DL, XLenVT));
+  }
   SDValue Reduction =
       DAG.getNode(RVVOpcode, DL, M1VT, StartSplat, Vec, StartSplat, Mask, VL);
   SDValue Elt0 = DAG.getNode(ISD::EXTRACT_VECTOR_ELT, DL, ResVT, Reduction,
                              DAG.getConstant(0, DL, XLenVT));
   if (!VecVT.isInteger())
     return Elt0;
   return DAG.getSExtOrTrunc(Elt0, DL, Op.getValueType());
 }

llvm/test/CodeGen/RISCV/rvv/fixed-vectors-reduction-fp-vp.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+experimental-zvfh,+v -target-abi=ilp32d -riscv-v-vector-bits-min=128 \
 ; RUN:   -verify-machineinstrs < %s | FileCheck %s
 ; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+experimental-zvfh,+v -target-abi=lp64d -riscv-v-vector-bits-min=128 \
 ; RUN:   -verify-machineinstrs < %s | FileCheck %s
 
 declare half @llvm.vp.reduce.fadd.v2f16(half, <2 x half>, <2 x i1>, i32)
 
 define half @vpreduce_fadd_v2f16(half %s, <2 x half> %v, <2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_fadd_v2f16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vfmv.s.f v9, fa0
-; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, tu, ma
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, ma
+; CHECK-NEXT:    vfmv.v.f v9, fa0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf4, tu, ma
 ; CHECK-NEXT:    vfredusum.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vfmv.f.s fa0, v9
 ; CHECK-NEXT:    ret
   %r = call reassoc half @llvm.vp.reduce.fadd.v2f16(half %s, <2 x half> %v, <2 x i1> %m, i32 %evl)
   ret half %r
 }
 
 define half @vpreduce_ord_fadd_v2f16(half %s, <2 x half> %v, <2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_ord_fadd_v2f16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vfmv.s.f v9, fa0
-; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, tu, ma
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, ma
+; CHECK-NEXT:    vfmv.v.f v9, fa0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf4, tu, ma
 ; CHECK-NEXT:    vfredosum.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vfmv.f.s fa0, v9
 ; CHECK-NEXT:    ret
   %r = call half @llvm.vp.reduce.fadd.v2f16(half %s, <2 x half> %v, <2 x i1> %m, i32 %evl)
   ret half %r
 }
 
 declare half @llvm.vp.reduce.fadd.v4f16(half, <4 x half>, <4 x i1>, i32)
 
 define half @vpreduce_fadd_v4f16(half %s, <4 x half> %v, <4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_fadd_v4f16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vfmv.s.f v9, fa0
-; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, tu, ma
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, ma
+; CHECK-NEXT:    vfmv.v.f v9, fa0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf2, tu, ma
 ; CHECK-NEXT:    vfredusum.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vfmv.f.s fa0, v9
 ; CHECK-NEXT:    ret
   %r = call reassoc half @llvm.vp.reduce.fadd.v4f16(half %s, <4 x half> %v, <4 x i1> %m, i32 %evl)
   ret half %r
 }
 
 define half @vpreduce_ord_fadd_v4f16(half %s, <4 x half> %v, <4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_ord_fadd_v4f16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vfmv.s.f v9, fa0
-; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, tu, ma
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, ma
+; CHECK-NEXT:    vfmv.v.f v9, fa0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf2, tu, ma
 ; CHECK-NEXT:    vfredosum.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vfmv.f.s fa0, v9
 ; CHECK-NEXT:    ret
   %r = call half @llvm.vp.reduce.fadd.v4f16(half %s, <4 x half> %v, <4 x i1> %m, i32 %evl)
   ret half %r
 }
 
 declare float @llvm.vp.reduce.fadd.v2f32(float, <2 x float>, <2 x i1>, i32)
 
 define float @vpreduce_fadd_v2f32(float %s, <2 x float> %v, <2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_fadd_v2f32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vfmv.s.f v9, fa0
-; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, tu, ma
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    vfmv.v.f v9, fa0
+; CHECK-NEXT:    vsetvli zero, zero, e32, mf2, tu, ma
 ; CHECK-NEXT:    vfredusum.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vfmv.f.s fa0, v9
 ; CHECK-NEXT:    ret
   %r = call reassoc float @llvm.vp.reduce.fadd.v2f32(float %s, <2 x float> %v, <2 x i1> %m, i32 %evl)
   ret float %r
 }
 
 define float @vpreduce_ord_fadd_v2f32(float %s, <2 x float> %v, <2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_ord_fadd_v2f32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vfmv.s.f v9, fa0
-; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, tu, ma
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    vfmv.v.f v9, fa0
+; CHECK-NEXT:    vsetvli zero, zero, e32, mf2, tu, ma
 ; CHECK-NEXT:    vfredosum.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vfmv.f.s fa0, v9
 ; CHECK-NEXT:    ret
   %r = call float @llvm.vp.reduce.fadd.v2f32(float %s, <2 x float> %v, <2 x i1> %m, i32 %evl)
   ret float %r
 }
 
 declare float @llvm.vp.reduce.fadd.v4f32(float, <4 x float>, <4 x i1>, i32)
 
 define float @vpreduce_fadd_v4f32(float %s, <4 x float> %v, <4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_fadd_v4f32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vfmv.s.f v9, fa0
-; CHECK-NEXT:    vsetvli zero, a0, e32, m1, tu, ma
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, ma
+; CHECK-NEXT:    vfmv.v.f v9, fa0
+; CHECK-NEXT:    vsetvli zero, zero, e32, m1, tu, ma
 ; CHECK-NEXT:    vfredusum.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vfmv.f.s fa0, v9
 ; CHECK-NEXT:    ret
   %r = call reassoc float @llvm.vp.reduce.fadd.v4f32(float %s, <4 x float> %v, <4 x i1> %m, i32 %evl)
   ret float %r
 }
 
 define float @vpreduce_ord_fadd_v4f32(float %s, <4 x float> %v, <4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_ord_fadd_v4f32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vfmv.s.f v9, fa0
-; CHECK-NEXT:    vsetvli zero, a0, e32, m1, tu, ma
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, ma
+; CHECK-NEXT:    vfmv.v.f v9, fa0
+; CHECK-NEXT:    vsetvli zero, zero, e32, m1, tu, ma
 ; CHECK-NEXT:    vfredosum.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vfmv.f.s fa0, v9
 ; CHECK-NEXT:    ret
   %r = call float @llvm.vp.reduce.fadd.v4f32(float %s, <4 x float> %v, <4 x i1> %m, i32 %evl)
   ret float %r
 }
 
 declare float @llvm.vp.reduce.fadd.v64f32(float, <64 x float>, <64 x i1>, i32)
; CHECK-NEXT:    ret
   ret float %r
 }
 
 declare double @llvm.vp.reduce.fadd.v2f64(double, <2 x double>, <2 x i1>, i32)
 
 define double @vpreduce_fadd_v2f64(double %s, <2 x double> %v, <2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_fadd_v2f64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
-; CHECK-NEXT:    vfmv.s.f v9, fa0
-; CHECK-NEXT:    vsetvli zero, a0, e64, m1, tu, ma
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
+; CHECK-NEXT:    vfmv.v.f v9, fa0
+; CHECK-NEXT:    vsetvli zero, zero, e64, m1, tu, ma
 ; CHECK-NEXT:    vfredusum.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vfmv.f.s fa0, v9
 ; CHECK-NEXT:    ret
   %r = call reassoc double @llvm.vp.reduce.fadd.v2f64(double %s, <2 x double> %v, <2 x i1> %m, i32 %evl)
   ret double %r
 }
 
 define double @vpreduce_ord_fadd_v2f64(double %s, <2 x double> %v, <2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_ord_fadd_v2f64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
-; CHECK-NEXT:    vfmv.s.f v9, fa0
-; CHECK-NEXT:    vsetvli zero, a0, e64, m1, tu, ma
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
+; CHECK-NEXT:    vfmv.v.f v9, fa0
+; CHECK-NEXT:    vsetvli zero, zero, e64, m1, tu, ma
 ; CHECK-NEXT:    vfredosum.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vfmv.f.s fa0, v9
 ; CHECK-NEXT:    ret
   %r = call double @llvm.vp.reduce.fadd.v2f64(double %s, <2 x double> %v, <2 x i1> %m, i32 %evl)
   ret double %r
 }
 
 declare double @llvm.vp.reduce.fadd.v3f64(double, <3 x double>, <3 x i1>, i32)

llvm/test/CodeGen/RISCV/rvv/fixed-vectors-reduction-fp.ll

 
 declare half @llvm.vector.reduce.fadd.v2f16(half, <2 x half>)
 
 define half @vreduce_fadd_v2f16(<2 x half>* %x, half %s) {
 ; CHECK-LABEL: vreduce_fadd_v2f16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, ma
 ; CHECK-NEXT:    vle16.v v8, (a0)
 ; CHECK-NEXT:    vfmv.s.f v9, fa0
 ; CHECK-NEXT:    vfredusum.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <2 x half>, <2 x half>* %x
   %red = call reassoc half @llvm.vector.reduce.fadd.v2f16(half %s, <2 x half> %v)
   ret half %red
 }
 
; CHECK-NEXT:    ret
   ret float %red
 }
 
 define float @vreduce_fwadd_v2f32(<2 x half>* %x, float %s) {
 ; CHECK-LABEL: vreduce_fwadd_v2f32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, ma
 ; CHECK-NEXT:    vle16.v v8, (a0)
 ; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
 ; CHECK-NEXT:    vfmv.s.f v9, fa0
 ; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, ma
 ; CHECK-NEXT:    vfwredusum.vs v8, v8, v9
 ; CHECK-NEXT:    vsetivli zero, 0, e32, m1, ta, ma
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <2 x half>, <2 x half>* %x
   %e = fpext <2 x half> %v to <2 x float>

 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <32 x double>, <32 x double>* %x
   %red = call double @llvm.vector.reduce.fadd.v32f64(double %s, <32 x double> %v)
   ret double %red
 }
 
 define double @vreduce_fwadd_v32f64(<32 x float>* %x, double %s) {
-; RV32-LABEL: vreduce_fwadd_v32f64:
-; RV32:       # %bb.0:
-; RV32-NEXT:    li a1, 32
-; RV32-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
-; RV32-NEXT:    vle32.v v8, (a0)
-; RV32-NEXT:    vsetivli zero, 16, e32, m8, ta, mu
-; RV32-NEXT:    vslidedown.vi v16, v8, 16
-; RV32-NEXT:    vsetivli zero, 16, e32, m4, ta, mu
-; RV32-NEXT:    vfwadd.vv v24, v8, v16
-; RV32-NEXT:    fcvt.d.w ft0, zero
-; RV32-NEXT:    fneg.d ft0, ft0
-; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
-; RV32-NEXT:    vfmv.s.f v8, ft0
-; RV32-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
-; RV32-NEXT:    vfredusum.vs v8, v24, v8
-; RV32-NEXT:    vfmv.f.s ft0, v8
-; RV32-NEXT:    fadd.d fa0, fa0, ft0
-; RV32-NEXT:    ret
-;
-; RV64-LABEL: vreduce_fwadd_v32f64:
-; RV64:       # %bb.0:
-; RV64-NEXT:    li a1, 32
-; RV64-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
-; RV64-NEXT:    vle32.v v8, (a0)
-; RV64-NEXT:    vsetivli zero, 16, e32, m8, ta, mu
-; RV64-NEXT:    vslidedown.vi v16, v8, 16
-; RV64-NEXT:    vsetivli zero, 16, e32, m4, ta, mu
-; RV64-NEXT:    vfwadd.vv v24, v8, v16
-; RV64-NEXT:    fmv.d.x ft0, zero
-; RV64-NEXT:    fneg.d ft0, ft0
-; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
-; RV64-NEXT:    vfmv.s.f v8, ft0
-; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
-; RV64-NEXT:    vfredusum.vs v8, v24, v8
-; RV64-NEXT:    vfmv.f.s ft0, v8
-; RV64-NEXT:    fadd.d fa0, fa0, ft0
-; RV64-NEXT:    ret
 ; CHECK-LABEL: vreduce_fwadd_v32f64:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, ma
 ; CHECK-NEXT:    vle32.v v8, (a0)
 ; CHECK-NEXT:    vsetivli zero, 16, e32, m8, ta, ma
 ; CHECK-NEXT:    vslidedown.vi v16, v8, 16
 ; CHECK-NEXT:    vsetivli zero, 16, e32, m4, ta, ma

 ; CHECK-LABEL: vreduce_fmin_v128f32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, ma
 ; CHECK-NEXT:    vle32.v v8, (a0)
 ; CHECK-NEXT:    addi a2, a0, 384
 ; CHECK-NEXT:    vle32.v v16, (a2)
 ; CHECK-NEXT:    addi a2, a0, 256
 ; CHECK-NEXT:    addi a0, a0, 128
 ; CHECK-NEXT:    vle32.v v24, (a0)
 ; CHECK-NEXT:    vle32.v v0, (a2)
 ; CHECK-NEXT:    vfmin.vv v16, v24, v16
 ; CHECK-NEXT:    vfmin.vv v8, v8, v0
 ; CHECK-NEXT:    vfmin.vv v8, v8, v16
 ; CHECK-NEXT:    lui a0, %hi(.LCPI77_0)
 ; CHECK-NEXT:    addi a0, a0, %lo(.LCPI77_0)
 ; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma

llvm/test/CodeGen/RISCV/rvv/fixed-vectors-reduction-int-vp.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc -mtriple=riscv32 -mattr=+v -riscv-v-vector-bits-min=128 -verify-machineinstrs < %s \
 ; RUN:   | FileCheck %s --check-prefixes=CHECK,RV32
 ; RUN: llc -mtriple=riscv64 -mattr=+v -riscv-v-vector-bits-min=128 -verify-machineinstrs < %s \
 ; RUN:   | FileCheck %s --check-prefixes=CHECK,RV64
 
 declare i8 @llvm.vp.reduce.add.v2i8(i8, <2 x i8>, <2 x i1>, i32)
 
 define signext i8 @vpreduce_add_v2i8(i8 signext %s, <2 x i8> %v, <2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_add_v2i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf8, tu, ma
 ; CHECK-NEXT:    vredsum.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.add.v2i8(i8 %s, <2 x i8> %v, <2 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i8 @llvm.vp.reduce.umax.v2i8(i8, <2 x i8>, <2 x i1>, i32)
 
 define signext i8 @vpreduce_umax_v2i8(i8 signext %s, <2 x i8> %v, <2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_umax_v2i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    andi a0, a0, 255
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf8, tu, ma
 ; CHECK-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.umax.v2i8(i8 %s, <2 x i8> %v, <2 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i8 @llvm.vp.reduce.smax.v2i8(i8, <2 x i8>, <2 x i1>, i32)
 
 define signext i8 @vpreduce_smax_v2i8(i8 signext %s, <2 x i8> %v, <2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_smax_v2i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf8, tu, ma
 ; CHECK-NEXT:    vredmax.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.smax.v2i8(i8 %s, <2 x i8> %v, <2 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i8 @llvm.vp.reduce.umin.v2i8(i8, <2 x i8>, <2 x i1>, i32)
 
 define signext i8 @vpreduce_umin_v2i8(i8 signext %s, <2 x i8> %v, <2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_umin_v2i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    andi a0, a0, 255
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf8, tu, ma
 ; CHECK-NEXT:    vredminu.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.umin.v2i8(i8 %s, <2 x i8> %v, <2 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i8 @llvm.vp.reduce.smin.v2i8(i8, <2 x i8>, <2 x i1>, i32)
 
 define signext i8 @vpreduce_smin_v2i8(i8 signext %s, <2 x i8> %v, <2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_smin_v2i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf8, tu, ma
 ; CHECK-NEXT:    vredmin.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.smin.v2i8(i8 %s, <2 x i8> %v, <2 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i8 @llvm.vp.reduce.and.v2i8(i8, <2 x i8>, <2 x i1>, i32)
 
 define signext i8 @vpreduce_and_v2i8(i8 signext %s, <2 x i8> %v, <2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_and_v2i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf8, tu, ma
 ; CHECK-NEXT:    vredand.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.and.v2i8(i8 %s, <2 x i8> %v, <2 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i8 @llvm.vp.reduce.or.v2i8(i8, <2 x i8>, <2 x i1>, i32)
 
 define signext i8 @vpreduce_or_v2i8(i8 signext %s, <2 x i8> %v, <2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_or_v2i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf8, tu, ma
 ; CHECK-NEXT:    vredor.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.or.v2i8(i8 %s, <2 x i8> %v, <2 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i8 @llvm.vp.reduce.xor.v2i8(i8, <2 x i8>, <2 x i1>, i32)
 
 define signext i8 @vpreduce_xor_v2i8(i8 signext %s, <2 x i8> %v, <2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_xor_v2i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf8, tu, ma
 ; CHECK-NEXT:    vredxor.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.xor.v2i8(i8 %s, <2 x i8> %v, <2 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i8 @llvm.vp.reduce.umin.v3i8(i8, <3 x i8>, <3 x i1>, i32)
 
 define signext i8 @vpreduce_umin_v3i8(i8 signext %s, <3 x i8> %v, <3 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_umin_v3i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    andi a0, a0, 255
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf4, tu, ma
 ; CHECK-NEXT:    vredminu.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.umin.v3i8(i8 %s, <3 x i8> %v, <3 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i8 @llvm.vp.reduce.add.v4i8(i8, <4 x i8>, <4 x i1>, i32)
 
 define signext i8 @vpreduce_add_v4i8(i8 signext %s, <4 x i8> %v, <4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_add_v4i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf4, tu, ma
 ; CHECK-NEXT:    vredsum.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.add.v4i8(i8 %s, <4 x i8> %v, <4 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i8 @llvm.vp.reduce.umax.v4i8(i8, <4 x i8>, <4 x i1>, i32)
 
 define signext i8 @vpreduce_umax_v4i8(i8 signext %s, <4 x i8> %v, <4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_umax_v4i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    andi a0, a0, 255
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf4, tu, ma
 ; CHECK-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.umax.v4i8(i8 %s, <4 x i8> %v, <4 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i8 @llvm.vp.reduce.smax.v4i8(i8, <4 x i8>, <4 x i1>, i32)
 
 define signext i8 @vpreduce_smax_v4i8(i8 signext %s, <4 x i8> %v, <4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_smax_v4i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf4, tu, ma
 ; CHECK-NEXT:    vredmax.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.smax.v4i8(i8 %s, <4 x i8> %v, <4 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i8 @llvm.vp.reduce.umin.v4i8(i8, <4 x i8>, <4 x i1>, i32)
 
 define signext i8 @vpreduce_umin_v4i8(i8 signext %s, <4 x i8> %v, <4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_umin_v4i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    andi a0, a0, 255
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf4, tu, ma
 ; CHECK-NEXT:    vredminu.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.umin.v4i8(i8 %s, <4 x i8> %v, <4 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i8 @llvm.vp.reduce.smin.v4i8(i8, <4 x i8>, <4 x i1>, i32)
 
 define signext i8 @vpreduce_smin_v4i8(i8 signext %s, <4 x i8> %v, <4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_smin_v4i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf4, tu, ma
 ; CHECK-NEXT:    vredmin.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.smin.v4i8(i8 %s, <4 x i8> %v, <4 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i8 @llvm.vp.reduce.and.v4i8(i8, <4 x i8>, <4 x i1>, i32)
 
 define signext i8 @vpreduce_and_v4i8(i8 signext %s, <4 x i8> %v, <4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_and_v4i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf4, tu, ma
 ; CHECK-NEXT:    vredand.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.and.v4i8(i8 %s, <4 x i8> %v, <4 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i8 @llvm.vp.reduce.or.v4i8(i8, <4 x i8>, <4 x i1>, i32)
 
 define signext i8 @vpreduce_or_v4i8(i8 signext %s, <4 x i8> %v, <4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_or_v4i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf4, tu, ma
 ; CHECK-NEXT:    vredor.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.or.v4i8(i8 %s, <4 x i8> %v, <4 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i8 @llvm.vp.reduce.xor.v4i8(i8, <4 x i8>, <4 x i1>, i32)
 
 define signext i8 @vpreduce_xor_v4i8(i8 signext %s, <4 x i8> %v, <4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_xor_v4i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf4, tu, ma
 ; CHECK-NEXT:    vredxor.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.xor.v4i8(i8 %s, <4 x i8> %v, <4 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i16 @llvm.vp.reduce.add.v2i16(i16, <2 x i16>, <2 x i1>, i32)
 
 define signext i16 @vpreduce_add_v2i16(i16 signext %s, <2 x i16> %v, <2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_add_v2i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e16, mf4, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e16, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf4, tu, ma
 ; CHECK-NEXT:    vredsum.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i16 @llvm.vp.reduce.add.v2i16(i16 %s, <2 x i16> %v, <2 x i1> %m, i32 %evl)
   ret i16 %r
 }
 
 declare i16 @llvm.vp.reduce.umax.v2i16(i16, <2 x i16>, <2 x i1>, i32)
 
 define signext i16 @vpreduce_umax_v2i16(i16 signext %s, <2 x i16> %v, <2 x i1> %m, i32 zeroext %evl) {
-; RV32-LABEL: vpreduce_umax_v2i16:
-; RV32:       # %bb.0:
-; RV32-NEXT:    slli a0, a0, 16
-; RV32-NEXT:    srli a0, a0, 16
-; RV32-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; RV32-NEXT:    vmv.s.x v9, a0
-; RV32-NEXT:    vsetvli zero, a1, e16, mf4, tu, ma
-; RV32-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
-; RV32-NEXT:    vmv.x.s a0, v9
-; RV32-NEXT:    ret
-;
-; RV64-LABEL: vpreduce_umax_v2i16:
-; RV64:       # %bb.0:
-; RV64-NEXT:    slli a0, a0, 48
-; RV64-NEXT:    srli a0, a0, 48
-; RV64-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli zero, a1, e16, mf4, tu, ma
-; RV64-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
-; RV64-NEXT:    vmv.x.s a0, v9
-; RV64-NEXT:    ret
+; CHECK-LABEL: vpreduce_umax_v2i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e16, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf4, tu, ma
+; CHECK-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
+; CHECK-NEXT:    vmv.x.s a0, v9
+; CHECK-NEXT:    ret
   %r = call i16 @llvm.vp.reduce.umax.v2i16(i16 %s, <2 x i16> %v, <2 x i1> %m, i32 %evl)
   ret i16 %r
 }
 
 declare i16 @llvm.vp.reduce.smax.v2i16(i16, <2 x i16>, <2 x i1>, i32)
 
 define signext i16 @vpreduce_smax_v2i16(i16 signext %s, <2 x i16> %v, <2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_smax_v2i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e16, mf4, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e16, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf4, tu, ma
 ; CHECK-NEXT:    vredmax.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i16 @llvm.vp.reduce.smax.v2i16(i16 %s, <2 x i16> %v, <2 x i1> %m, i32 %evl)
   ret i16 %r
 }
 
 declare i16 @llvm.vp.reduce.umin.v2i16(i16, <2 x i16>, <2 x i1>, i32)
 
 define signext i16 @vpreduce_umin_v2i16(i16 signext %s, <2 x i16> %v, <2 x i1> %m, i32 zeroext %evl) {
-; RV32-LABEL: vpreduce_umin_v2i16:
-; RV32:       # %bb.0:
-; RV32-NEXT:    slli a0, a0, 16
-; RV32-NEXT:    srli a0, a0, 16
-; RV32-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; RV32-NEXT:    vmv.s.x v9, a0
-; RV32-NEXT:    vsetvli zero, a1, e16, mf4, tu, ma
-; RV32-NEXT:    vredminu.vs v9, v8, v9, v0.t
-; RV32-NEXT:    vmv.x.s a0, v9
-; RV32-NEXT:    ret
-;
-; RV64-LABEL: vpreduce_umin_v2i16:
-; RV64:       # %bb.0:
-; RV64-NEXT:    slli a0, a0, 48
-; RV64-NEXT:    srli a0, a0, 48
-; RV64-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli zero, a1, e16, mf4, tu, ma
-; RV64-NEXT:    vredminu.vs v9, v8, v9, v0.t
-; RV64-NEXT:    vmv.x.s a0, v9
-; RV64-NEXT:    ret
+; CHECK-LABEL: vpreduce_umin_v2i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e16, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf4, tu, ma
+; CHECK-NEXT:    vredminu.vs v9, v8, v9, v0.t
+; CHECK-NEXT:    vmv.x.s a0, v9
+; CHECK-NEXT:    ret
   %r = call i16 @llvm.vp.reduce.umin.v2i16(i16 %s, <2 x i16> %v, <2 x i1> %m, i32 %evl)
   ret i16 %r
 }
 
 declare i16 @llvm.vp.reduce.smin.v2i16(i16, <2 x i16>, <2 x i1>, i32)
 
 define signext i16 @vpreduce_smin_v2i16(i16 signext %s, <2 x i16> %v, <2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_smin_v2i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e16, mf4, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e16, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf4, tu, ma
 ; CHECK-NEXT:    vredmin.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i16 @llvm.vp.reduce.smin.v2i16(i16 %s, <2 x i16> %v, <2 x i1> %m, i32 %evl)
   ret i16 %r
 }
 
 declare i16 @llvm.vp.reduce.and.v2i16(i16, <2 x i16>, <2 x i1>, i32)
 
 define signext i16 @vpreduce_and_v2i16(i16 signext %s, <2 x i16> %v, <2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_and_v2i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e16, mf4, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e16, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf4, tu, ma
 ; CHECK-NEXT:    vredand.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i16 @llvm.vp.reduce.and.v2i16(i16 %s, <2 x i16> %v, <2 x i1> %m, i32 %evl)
   ret i16 %r
 }
 
 declare i16 @llvm.vp.reduce.or.v2i16(i16, <2 x i16>, <2 x i1>, i32)
 
 define signext i16 @vpreduce_or_v2i16(i16 signext %s, <2 x i16> %v, <2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_or_v2i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e16, mf4, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e16, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf4, tu, ma
 ; CHECK-NEXT:    vredor.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i16 @llvm.vp.reduce.or.v2i16(i16 %s, <2 x i16> %v, <2 x i1> %m, i32 %evl)
   ret i16 %r
 }
 
 declare i16 @llvm.vp.reduce.xor.v2i16(i16, <2 x i16>, <2 x i1>, i32)
 
 define signext i16 @vpreduce_xor_v2i16(i16 signext %s, <2 x i16> %v, <2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_xor_v2i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e16, mf4, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e16, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf4, tu, ma
 ; CHECK-NEXT:    vredxor.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i16 @llvm.vp.reduce.xor.v2i16(i16 %s, <2 x i16> %v, <2 x i1> %m, i32 %evl)
   ret i16 %r
 }
 
 declare i16 @llvm.vp.reduce.add.v4i16(i16, <4 x i16>, <4 x i1>, i32)
 
 define signext i16 @vpreduce_add_v4i16(i16 signext %s, <4 x i16> %v, <4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_add_v4i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e16, mf2, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e16, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf2, tu, ma
 ; CHECK-NEXT:    vredsum.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i16 @llvm.vp.reduce.add.v4i16(i16 %s, <4 x i16> %v, <4 x i1> %m, i32 %evl)
   ret i16 %r
 }
 
 declare i16 @llvm.vp.reduce.umax.v4i16(i16, <4 x i16>, <4 x i1>, i32)
 
 define signext i16 @vpreduce_umax_v4i16(i16 signext %s, <4 x i16> %v, <4 x i1> %m, i32 zeroext %evl) {
-; RV32-LABEL: vpreduce_umax_v4i16:
-; RV32:       # %bb.0:
-; RV32-NEXT:    slli a0, a0, 16
-; RV32-NEXT:    srli a0, a0, 16
-; RV32-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; RV32-NEXT:    vmv.s.x v9, a0
-; RV32-NEXT:    vsetvli zero, a1, e16, mf2, tu, ma
-; RV32-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
-; RV32-NEXT:    vmv.x.s a0, v9
-; RV32-NEXT:    ret
-;
-; RV64-LABEL: vpreduce_umax_v4i16:
-; RV64:       # %bb.0:
-; RV64-NEXT:    slli a0, a0, 48
-; RV64-NEXT:    srli a0, a0, 48
-; RV64-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli zero, a1, e16, mf2, tu, ma
-; RV64-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
-; RV64-NEXT:    vmv.x.s a0, v9
-; RV64-NEXT:    ret
+; CHECK-LABEL: vpreduce_umax_v4i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e16, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf2, tu, ma
+; CHECK-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
+; CHECK-NEXT:    vmv.x.s a0, v9
+; CHECK-NEXT:    ret
   %r = call i16 @llvm.vp.reduce.umax.v4i16(i16 %s, <4 x i16> %v, <4 x i1> %m, i32 %evl)
   ret i16 %r
 }
 
 declare i16 @llvm.vp.reduce.smax.v4i16(i16, <4 x i16>, <4 x i1>, i32)
 
 define signext i16 @vpreduce_smax_v4i16(i16 signext %s, <4 x i16> %v, <4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_smax_v4i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e16, mf2, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e16, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf2, tu, ma
 ; CHECK-NEXT:    vredmax.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i16 @llvm.vp.reduce.smax.v4i16(i16 %s, <4 x i16> %v, <4 x i1> %m, i32 %evl)
   ret i16 %r
 }
 
 declare i16 @llvm.vp.reduce.umin.v4i16(i16, <4 x i16>, <4 x i1>, i32)
 
 define signext i16 @vpreduce_umin_v4i16(i16 signext %s, <4 x i16> %v, <4 x i1> %m, i32 zeroext %evl) {
-; RV32-LABEL: vpreduce_umin_v4i16:
-; RV32:       # %bb.0:
-; RV32-NEXT:    slli a0, a0, 16
-; RV32-NEXT:    srli a0, a0, 16
-; RV32-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; RV32-NEXT:    vmv.s.x v9, a0
-; RV32-NEXT:    vsetvli zero, a1, e16, mf2, tu, ma
-; RV32-NEXT:    vredminu.vs v9, v8, v9, v0.t
-; RV32-NEXT:    vmv.x.s a0, v9
-; RV32-NEXT:    ret
-;
-; RV64-LABEL: vpreduce_umin_v4i16:
-; RV64:       # %bb.0:
-; RV64-NEXT:    slli a0, a0, 48
-; RV64-NEXT:    srli a0, a0, 48
-; RV64-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli zero, a1, e16, mf2, tu, ma
-; RV64-NEXT:    vredminu.vs v9, v8, v9, v0.t
-; RV64-NEXT:    vmv.x.s a0, v9
-; RV64-NEXT:    ret
+; CHECK-LABEL: vpreduce_umin_v4i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e16, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf2, tu, ma
+; CHECK-NEXT:    vredminu.vs v9, v8, v9, v0.t
+; CHECK-NEXT:    vmv.x.s a0, v9
+; CHECK-NEXT:    ret
   %r = call i16 @llvm.vp.reduce.umin.v4i16(i16 %s, <4 x i16> %v, <4 x i1> %m, i32 %evl)
   ret i16 %r
 }
 
 declare i16 @llvm.vp.reduce.smin.v4i16(i16, <4 x i16>, <4 x i1>, i32)
 
 define signext i16 @vpreduce_smin_v4i16(i16 signext %s, <4 x i16> %v, <4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_smin_v4i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e16, mf2, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e16, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf2, tu, ma
 ; CHECK-NEXT:    vredmin.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i16 @llvm.vp.reduce.smin.v4i16(i16 %s, <4 x i16> %v, <4 x i1> %m, i32 %evl)
   ret i16 %r
 }
 
 declare i16 @llvm.vp.reduce.and.v4i16(i16, <4 x i16>, <4 x i1>, i32)
 
 define signext i16 @vpreduce_and_v4i16(i16 signext %s, <4 x i16> %v, <4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_and_v4i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e16, mf2, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e16, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf2, tu, ma
 ; CHECK-NEXT:    vredand.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i16 @llvm.vp.reduce.and.v4i16(i16 %s, <4 x i16> %v, <4 x i1> %m, i32 %evl)
   ret i16 %r
 }
 
 declare i16 @llvm.vp.reduce.or.v4i16(i16, <4 x i16>, <4 x i1>, i32)
 
 define signext i16 @vpreduce_or_v4i16(i16 signext %s, <4 x i16> %v, <4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_or_v4i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e16, mf2, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e16, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf2, tu, ma
 ; CHECK-NEXT:    vredor.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i16 @llvm.vp.reduce.or.v4i16(i16 %s, <4 x i16> %v, <4 x i1> %m, i32 %evl)
   ret i16 %r
 }
 
 declare i16 @llvm.vp.reduce.xor.v4i16(i16, <4 x i16>, <4 x i1>, i32)
 
 define signext i16 @vpreduce_xor_v4i16(i16 signext %s, <4 x i16> %v, <4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_xor_v4i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e16, mf2, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e16, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf2, tu, ma
 ; CHECK-NEXT:    vredxor.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i16 @llvm.vp.reduce.xor.v4i16(i16 %s, <4 x i16> %v, <4 x i1> %m, i32 %evl)
   ret i16 %r
 }
 
 declare i32 @llvm.vp.reduce.add.v2i32(i32, <2 x i32>, <2 x i1>, i32)
 
 define signext i32 @vpreduce_add_v2i32(i32 signext %s, <2 x i32> %v, <2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_add_v2i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e32, mf2, tu, ma
 ; CHECK-NEXT:    vredsum.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i32 @llvm.vp.reduce.add.v2i32(i32 %s, <2 x i32> %v, <2 x i1> %m, i32 %evl)
   ret i32 %r
 }
 
 declare i32 @llvm.vp.reduce.umax.v2i32(i32, <2 x i32>, <2 x i1>, i32)
 
 define signext i32 @vpreduce_umax_v2i32(i32 signext %s, <2 x i32> %v, <2 x i1> %m, i32 zeroext %evl) {
-; RV32-LABEL: vpreduce_umax_v2i32:
-; RV32:       # %bb.0:
-; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; RV32-NEXT:    vmv.s.x v9, a0
-; RV32-NEXT:    vsetvli zero, a1, e32, mf2, tu, ma
-; RV32-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
-; RV32-NEXT:    vmv.x.s a0, v9
-; RV32-NEXT:    ret
-;
-; RV64-LABEL: vpreduce_umax_v2i32:
-; RV64:       # %bb.0:
-; RV64-NEXT:    slli a0, a0, 32
-; RV64-NEXT:    srli a0, a0, 32
-; RV64-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli zero, a1, e32, mf2, tu, ma
-; RV64-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
-; RV64-NEXT:    vmv.x.s a0, v9
-; RV64-NEXT:    ret
+; CHECK-LABEL: vpreduce_umax_v2i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e32, mf2, tu, ma
+; CHECK-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
+; CHECK-NEXT:    vmv.x.s a0, v9
+; CHECK-NEXT:    ret
   %r = call i32 @llvm.vp.reduce.umax.v2i32(i32 %s, <2 x i32> %v, <2 x i1> %m, i32 %evl)
   ret i32 %r
 }
 
 declare i32 @llvm.vp.reduce.smax.v2i32(i32, <2 x i32>, <2 x i1>, i32)
 
 define signext i32 @vpreduce_smax_v2i32(i32 signext %s, <2 x i32> %v, <2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_smax_v2i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e32, mf2, tu, ma
 ; CHECK-NEXT:    vredmax.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i32 @llvm.vp.reduce.smax.v2i32(i32 %s, <2 x i32> %v, <2 x i1> %m, i32 %evl)
   ret i32 %r
 }
 
 declare i32 @llvm.vp.reduce.umin.v2i32(i32, <2 x i32>, <2 x i1>, i32)
 
 define signext i32 @vpreduce_umin_v2i32(i32 signext %s, <2 x i32> %v, <2 x i1> %m, i32 zeroext %evl) {
-; RV32-LABEL: vpreduce_umin_v2i32:
-; RV32:       # %bb.0:
-; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; RV32-NEXT:    vmv.s.x v9, a0
-; RV32-NEXT:    vsetvli zero, a1, e32, mf2, tu, ma
-; RV32-NEXT:    vredminu.vs v9, v8, v9, v0.t
-; RV32-NEXT:    vmv.x.s a0, v9
-; RV32-NEXT:    ret
-;
-; RV64-LABEL: vpreduce_umin_v2i32:
-; RV64:       # %bb.0:
-; RV64-NEXT:    slli a0, a0, 32
-; RV64-NEXT:    srli a0, a0, 32
-; RV64-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli zero, a1, e32, mf2, tu, ma
-; RV64-NEXT:    vredminu.vs v9, v8, v9, v0.t
-; RV64-NEXT:    vmv.x.s a0, v9
-; RV64-NEXT:    ret
+; CHECK-LABEL: vpreduce_umin_v2i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e32, mf2, tu, ma
+; CHECK-NEXT:    vredminu.vs v9, v8, v9, v0.t
+; CHECK-NEXT:    vmv.x.s a0, v9
+; CHECK-NEXT:    ret
   %r = call i32 @llvm.vp.reduce.umin.v2i32(i32 %s, <2 x i32> %v, <2 x i1> %m, i32 %evl)
   ret i32 %r
 }
 
 declare i32 @llvm.vp.reduce.smin.v2i32(i32, <2 x i32>, <2 x i1>, i32)
 
 define signext i32 @vpreduce_smin_v2i32(i32 signext %s, <2 x i32> %v, <2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_smin_v2i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e32, mf2, tu, ma
 ; CHECK-NEXT:    vredmin.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i32 @llvm.vp.reduce.smin.v2i32(i32 %s, <2 x i32> %v, <2 x i1> %m, i32 %evl)
   ret i32 %r
 }
 
 declare i32 @llvm.vp.reduce.and.v2i32(i32, <2 x i32>, <2 x i1>, i32)
 
 define signext i32 @vpreduce_and_v2i32(i32 signext %s, <2 x i32> %v, <2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_and_v2i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e32, mf2, tu, ma
 ; CHECK-NEXT:    vredand.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i32 @llvm.vp.reduce.and.v2i32(i32 %s, <2 x i32> %v, <2 x i1> %m, i32 %evl)
   ret i32 %r
 }
 
 declare i32 @llvm.vp.reduce.or.v2i32(i32, <2 x i32>, <2 x i1>, i32)
 
 define signext i32 @vpreduce_or_v2i32(i32 signext %s, <2 x i32> %v, <2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_or_v2i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e32, mf2, tu, ma
 ; CHECK-NEXT:    vredor.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i32 @llvm.vp.reduce.or.v2i32(i32 %s, <2 x i32> %v, <2 x i1> %m, i32 %evl)
   ret i32 %r
 }
 
 declare i32 @llvm.vp.reduce.xor.v2i32(i32, <2 x i32>, <2 x i1>, i32)
 
 define signext i32 @vpreduce_xor_v2i32(i32 signext %s, <2 x i32> %v, <2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_xor_v2i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e32, mf2, tu, ma
 ; CHECK-NEXT:    vredxor.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i32 @llvm.vp.reduce.xor.v2i32(i32 %s, <2 x i32> %v, <2 x i1> %m, i32 %evl)
   ret i32 %r
 }
 
 declare i32 @llvm.vp.reduce.add.v4i32(i32, <4 x i32>, <4 x i1>, i32)
 
 define signext i32 @vpreduce_add_v4i32(i32 signext %s, <4 x i32> %v, <4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_add_v4i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e32, m1, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e32, m1, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e32, m1, tu, ma
 ; CHECK-NEXT:    vredsum.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i32 @llvm.vp.reduce.add.v4i32(i32 %s, <4 x i32> %v, <4 x i1> %m, i32 %evl)
   ret i32 %r
 }
 
 declare i32 @llvm.vp.reduce.umax.v4i32(i32, <4 x i32>, <4 x i1>, i32)
 
 define signext i32 @vpreduce_umax_v4i32(i32 signext %s, <4 x i32> %v, <4 x i1> %m, i32 zeroext %evl) {
-; RV32-LABEL: vpreduce_umax_v4i32:
-; RV32:       # %bb.0:
-; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; RV32-NEXT:    vmv.s.x v9, a0
-; RV32-NEXT:    vsetvli zero, a1, e32, m1, tu, ma
-; RV32-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
-; RV32-NEXT:    vmv.x.s a0, v9
-; RV32-NEXT:    ret
-;
-; RV64-LABEL: vpreduce_umax_v4i32:
-; RV64:       # %bb.0:
-; RV64-NEXT:    slli a0, a0, 32
-; RV64-NEXT:    srli a0, a0, 32
-; RV64-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli zero, a1, e32, m1, tu, ma
-; RV64-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
-; RV64-NEXT:    vmv.x.s a0, v9
-; RV64-NEXT:    ret
+; CHECK-LABEL: vpreduce_umax_v4i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, m1, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e32, m1, tu, ma
+; CHECK-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
+; CHECK-NEXT:    vmv.x.s a0, v9
+; CHECK-NEXT:    ret
   %r = call i32 @llvm.vp.reduce.umax.v4i32(i32 %s, <4 x i32> %v, <4 x i1> %m, i32 %evl)
   ret i32 %r
 }
 
 declare i32 @llvm.vp.reduce.smax.v4i32(i32, <4 x i32>, <4 x i1>, i32)
 
 define signext i32 @vpreduce_smax_v4i32(i32 signext %s, <4 x i32> %v, <4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_smax_v4i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e32, m1, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e32, m1, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e32, m1, tu, ma
 ; CHECK-NEXT:    vredmax.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i32 @llvm.vp.reduce.smax.v4i32(i32 %s, <4 x i32> %v, <4 x i1> %m, i32 %evl)
   ret i32 %r
 }
 
 declare i32 @llvm.vp.reduce.umin.v4i32(i32, <4 x i32>, <4 x i1>, i32)
 
 define signext i32 @vpreduce_umin_v4i32(i32 signext %s, <4 x i32> %v, <4 x i1> %m, i32 zeroext %evl) {
-; RV32-LABEL: vpreduce_umin_v4i32:
-; RV32:       # %bb.0:
-; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; RV32-NEXT:    vmv.s.x v9, a0
-; RV32-NEXT:    vsetvli zero, a1, e32, m1, tu, ma
-; RV32-NEXT:    vredminu.vs v9, v8, v9, v0.t
-; RV32-NEXT:    vmv.x.s a0, v9
-; RV32-NEXT:    ret
-;
-; RV64-LABEL: vpreduce_umin_v4i32:
-; RV64:       # %bb.0:
-; RV64-NEXT:    slli a0, a0, 32
-; RV64-NEXT:    srli a0, a0, 32
-; RV64-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli zero, a1, e32, m1, tu, ma
-; RV64-NEXT:    vredminu.vs v9, v8, v9, v0.t
-; RV64-NEXT:    vmv.x.s a0, v9
-; RV64-NEXT:    ret
+; CHECK-LABEL: vpreduce_umin_v4i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, m1, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e32, m1, tu, ma
+; CHECK-NEXT:    vredminu.vs v9, v8, v9, v0.t
+; CHECK-NEXT:    vmv.x.s a0, v9
+; CHECK-NEXT:    ret
   %r = call i32 @llvm.vp.reduce.umin.v4i32(i32 %s, <4 x i32> %v, <4 x i1> %m, i32 %evl)
   ret i32 %r
 }
 
 declare i32 @llvm.vp.reduce.smin.v4i32(i32, <4 x i32>, <4 x i1>, i32)
 
 define signext i32 @vpreduce_smin_v4i32(i32 signext %s, <4 x i32> %v, <4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_smin_v4i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e32, m1, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e32, m1, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e32, m1, tu, ma
 ; CHECK-NEXT:    vredmin.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i32 @llvm.vp.reduce.smin.v4i32(i32 %s, <4 x i32> %v, <4 x i1> %m, i32 %evl)
   ret i32 %r
 }
 
 declare i32 @llvm.vp.reduce.and.v4i32(i32, <4 x i32>, <4 x i1>, i32)
 
 define signext i32 @vpreduce_and_v4i32(i32 signext %s, <4 x i32> %v, <4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_and_v4i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e32, m1, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e32, m1, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e32, m1, tu, ma
 ; CHECK-NEXT:    vredand.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i32 @llvm.vp.reduce.and.v4i32(i32 %s, <4 x i32> %v, <4 x i1> %m, i32 %evl)
   ret i32 %r
 }
 
 declare i32 @llvm.vp.reduce.or.v4i32(i32, <4 x i32>, <4 x i1>, i32)
 
 define signext i32 @vpreduce_or_v4i32(i32 signext %s, <4 x i32> %v, <4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_or_v4i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e32, m1, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e32, m1, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e32, m1, tu, ma
 ; CHECK-NEXT:    vredor.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i32 @llvm.vp.reduce.or.v4i32(i32 %s, <4 x i32> %v, <4 x i1> %m, i32 %evl)
   ret i32 %r
 }
 
 declare i32 @llvm.vp.reduce.xor.v4i32(i32, <4 x i32>, <4 x i1>, i32)
 
 define signext i32 @vpreduce_xor_v4i32(i32 signext %s, <4 x i32> %v, <4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_xor_v4i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e32, m1, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e32, m1, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e32, m1, tu, ma
 ; CHECK-NEXT:    vredxor.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i32 @llvm.vp.reduce.xor.v4i32(i32 %s, <4 x i32> %v, <4 x i1> %m, i32 %evl)
   ret i32 %r
 }
 
 declare i32 @llvm.vp.reduce.xor.v64i32(i32, <64 x i32>, <64 x i1>, i32)

 define signext i64 @vpreduce_add_v2i64(i64 signext %s, <2 x i64> %v, <2 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_add_v2i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    addi sp, sp, -16
 ; RV32-NEXT:    .cfi_def_cfa_offset 16
 ; RV32-NEXT:    sw a1, 12(sp)
 ; RV32-NEXT:    sw a0, 8(sp)
 ; RV32-NEXT:    addi a0, sp, 8
-; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV32-NEXT:    vsetvli zero, a2, e64, m1, ta, ma
 ; RV32-NEXT:    vlse64.v v9, (a0), zero
-; RV32-NEXT:    vsetvli zero, a2, e64, m1, tu, ma
+; RV32-NEXT:    vsetvli zero, zero, e64, m1, tu, ma
 ; RV32-NEXT:    vredsum.vs v9, v8, v9, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v9
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vsrl.vx v8, v9, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_add_v2i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli zero, a1, e64, m1, tu, ma
+; RV64-NEXT:    vsetvli zero, a1, e64, m1, ta, ma
+; RV64-NEXT:    vmv.v.x v9, a0
+; RV64-NEXT:    vsetvli zero, zero, e64, m1, tu, ma
 ; RV64-NEXT:    vredsum.vs v9, v8, v9, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v9
 ; RV64-NEXT:    ret
   %r = call i64 @llvm.vp.reduce.add.v2i64(i64 %s, <2 x i64> %v, <2 x i1> %m, i32 %evl)
   ret i64 %r
 }
 
 declare i64 @llvm.vp.reduce.umax.v2i64(i64, <2 x i64>, <2 x i1>, i32)
 
 define signext i64 @vpreduce_umax_v2i64(i64 signext %s, <2 x i64> %v, <2 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_umax_v2i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    addi sp, sp, -16
 ; RV32-NEXT:    .cfi_def_cfa_offset 16
 ; RV32-NEXT:    sw a1, 12(sp)
 ; RV32-NEXT:    sw a0, 8(sp)
 ; RV32-NEXT:    addi a0, sp, 8
-; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV32-NEXT:    vsetvli zero, a2, e64, m1, ta, ma
 ; RV32-NEXT:    vlse64.v v9, (a0), zero
-; RV32-NEXT:    vsetvli zero, a2, e64, m1, tu, ma
+; RV32-NEXT:    vsetvli zero, zero, e64, m1, tu, ma
 ; RV32-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v9
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vsrl.vx v8, v9, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_umax_v2i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli zero, a1, e64, m1, tu, ma
+; RV64-NEXT:    vsetvli zero, a1, e64, m1, ta, ma
+; RV64-NEXT:    vmv.v.x v9, a0
+; RV64-NEXT:    vsetvli zero, zero, e64, m1, tu, ma
 ; RV64-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v9
 ; RV64-NEXT:    ret
   %r = call i64 @llvm.vp.reduce.umax.v2i64(i64 %s, <2 x i64> %v, <2 x i1> %m, i32 %evl)
   ret i64 %r
 }
 
 declare i64 @llvm.vp.reduce.smax.v2i64(i64, <2 x i64>, <2 x i1>, i32)
 
 define signext i64 @vpreduce_smax_v2i64(i64 signext %s, <2 x i64> %v, <2 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_smax_v2i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    addi sp, sp, -16
 ; RV32-NEXT:    .cfi_def_cfa_offset 16
 ; RV32-NEXT:    sw a1, 12(sp)
 ; RV32-NEXT:    sw a0, 8(sp)
 ; RV32-NEXT:    addi a0, sp, 8
-; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV32-NEXT:    vsetvli zero, a2, e64, m1, ta, ma
 ; RV32-NEXT:    vlse64.v v9, (a0), zero
-; RV32-NEXT:    vsetvli zero, a2, e64, m1, tu, ma
+; RV32-NEXT:    vsetvli zero, zero, e64, m1, tu, ma
 ; RV32-NEXT:    vredmax.vs v9, v8, v9, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v9
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vsrl.vx v8, v9, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_smax_v2i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli zero, a1, e64, m1, tu, ma
+; RV64-NEXT:    vsetvli zero, a1, e64, m1, ta, ma
+; RV64-NEXT:    vmv.v.x v9, a0
+; RV64-NEXT:    vsetvli zero, zero, e64, m1, tu, ma
 ; RV64-NEXT:    vredmax.vs v9, v8, v9, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v9
 ; RV64-NEXT:    ret
   %r = call i64 @llvm.vp.reduce.smax.v2i64(i64 %s, <2 x i64> %v, <2 x i1> %m, i32 %evl)
   ret i64 %r
 }
 
 declare i64 @llvm.vp.reduce.umin.v2i64(i64, <2 x i64>, <2 x i1>, i32)
 
 define signext i64 @vpreduce_umin_v2i64(i64 signext %s, <2 x i64> %v, <2 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_umin_v2i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    addi sp, sp, -16
 ; RV32-NEXT:    .cfi_def_cfa_offset 16
 ; RV32-NEXT:    sw a1, 12(sp)
 ; RV32-NEXT:    sw a0, 8(sp)
 ; RV32-NEXT:    addi a0, sp, 8
-; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV32-NEXT:    vsetvli zero, a2, e64, m1, ta, ma
 ; RV32-NEXT:    vlse64.v v9, (a0), zero
-; RV32-NEXT:    vsetvli zero, a2, e64, m1, tu, ma
+; RV32-NEXT:    vsetvli zero, zero, e64, m1, tu, ma
 ; RV32-NEXT:    vredminu.vs v9, v8, v9, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v9
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vsrl.vx v8, v9, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_umin_v2i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli zero, a1, e64, m1, tu, ma
+; RV64-NEXT:    vsetvli zero, a1, e64, m1, ta, ma
+; RV64-NEXT:    vmv.v.x v9, a0
+; RV64-NEXT:    vsetvli zero, zero, e64, m1, tu, ma
 ; RV64-NEXT:    vredminu.vs v9, v8, v9, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v9
 ; RV64-NEXT:    ret
   %r = call i64 @llvm.vp.reduce.umin.v2i64(i64 %s, <2 x i64> %v, <2 x i1> %m, i32 %evl)
   ret i64 %r
 }
 
 declare i64 @llvm.vp.reduce.smin.v2i64(i64, <2 x i64>, <2 x i1>, i32)
 
 define signext i64 @vpreduce_smin_v2i64(i64 signext %s, <2 x i64> %v, <2 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_smin_v2i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    addi sp, sp, -16
 ; RV32-NEXT:    .cfi_def_cfa_offset 16
 ; RV32-NEXT:    sw a1, 12(sp)
 ; RV32-NEXT:    sw a0, 8(sp)
 ; RV32-NEXT:    addi a0, sp, 8
-; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV32-NEXT:    vsetvli zero, a2, e64, m1, ta, ma
 ; RV32-NEXT:    vlse64.v v9, (a0), zero
-; RV32-NEXT:    vsetvli zero, a2, e64, m1, tu, ma
+; RV32-NEXT:    vsetvli zero, zero, e64, m1, tu, ma
 ; RV32-NEXT:    vredmin.vs v9, v8, v9, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v9
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vsrl.vx v8, v9, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_smin_v2i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli zero, a1, e64, m1, tu, ma
+; RV64-NEXT:    vsetvli zero, a1, e64, m1, ta, ma
+; RV64-NEXT:    vmv.v.x v9, a0
+; RV64-NEXT:    vsetvli zero, zero, e64, m1, tu, ma
 ; RV64-NEXT:    vredmin.vs v9, v8, v9, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v9
 ; RV64-NEXT:    ret
   %r = call i64 @llvm.vp.reduce.smin.v2i64(i64 %s, <2 x i64> %v, <2 x i1> %m, i32 %evl)
   ret i64 %r
 }
 
 declare i64 @llvm.vp.reduce.and.v2i64(i64, <2 x i64>, <2 x i1>, i32)
 
 define signext i64 @vpreduce_and_v2i64(i64 signext %s, <2 x i64> %v, <2 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_and_v2i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    addi sp, sp, -16
 ; RV32-NEXT:    .cfi_def_cfa_offset 16
 ; RV32-NEXT:    sw a1, 12(sp)
 ; RV32-NEXT:    sw a0, 8(sp)
 ; RV32-NEXT:    addi a0, sp, 8
-; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV32-NEXT:    vsetvli zero, a2, e64, m1, ta, ma
 ; RV32-NEXT:    vlse64.v v9, (a0), zero
-; RV32-NEXT:    vsetvli zero, a2, e64, m1, tu, ma
+; RV32-NEXT:    vsetvli zero, zero, e64, m1, tu, ma
 ; RV32-NEXT:    vredand.vs v9, v8, v9, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v9
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vsrl.vx v8, v9, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_and_v2i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli zero, a1, e64, m1, tu, ma
+; RV64-NEXT:    vsetvli zero, a1, e64, m1, ta, ma
+; RV64-NEXT:    vmv.v.x v9, a0
+; RV64-NEXT:    vsetvli zero, zero, e64, m1, tu, ma
 ; RV64-NEXT:    vredand.vs v9, v8, v9, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v9
 ; RV64-NEXT:    ret
   %r = call i64 @llvm.vp.reduce.and.v2i64(i64 %s, <2 x i64> %v, <2 x i1> %m, i32 %evl)
   ret i64 %r
 }
 
 declare i64 @llvm.vp.reduce.or.v2i64(i64, <2 x i64>, <2 x i1>, i32)
 
 define signext i64 @vpreduce_or_v2i64(i64 signext %s, <2 x i64> %v, <2 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_or_v2i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    addi sp, sp, -16
 ; RV32-NEXT:    .cfi_def_cfa_offset 16
 ; RV32-NEXT:    sw a1, 12(sp)
 ; RV32-NEXT:    sw a0, 8(sp)
 ; RV32-NEXT:    addi a0, sp, 8
-; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV32-NEXT:    vsetvli zero, a2, e64, m1, ta, ma
 ; RV32-NEXT:    vlse64.v v9, (a0), zero
-; RV32-NEXT:    vsetvli zero, a2, e64, m1, tu, ma
+; RV32-NEXT:    vsetvli zero, zero, e64, m1, tu, ma
 ; RV32-NEXT:    vredor.vs v9, v8, v9, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v9
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vsrl.vx v8, v9, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_or_v2i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli zero, a1, e64, m1, tu, ma
+; RV64-NEXT:    vsetvli zero, a1, e64, m1, ta, ma
+; RV64-NEXT:    vmv.v.x v9, a0
+; RV64-NEXT:    vsetvli zero, zero, e64, m1, tu, ma
 ; RV64-NEXT:    vredor.vs v9, v8, v9, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v9
 ; RV64-NEXT:    ret
   %r = call i64 @llvm.vp.reduce.or.v2i64(i64 %s, <2 x i64> %v, <2 x i1> %m, i32 %evl)
   ret i64 %r
 }
 
 declare i64 @llvm.vp.reduce.xor.v2i64(i64, <2 x i64>, <2 x i1>, i32)
 
 define signext i64 @vpreduce_xor_v2i64(i64 signext %s, <2 x i64> %v, <2 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_xor_v2i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    addi sp, sp, -16
 ; RV32-NEXT:    .cfi_def_cfa_offset 16
 ; RV32-NEXT:    sw a1, 12(sp)
 ; RV32-NEXT:    sw a0, 8(sp)
 ; RV32-NEXT:    addi a0, sp, 8
-; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV32-NEXT:    vsetvli zero, a2, e64, m1, ta, ma
 ; RV32-NEXT:    vlse64.v v9, (a0), zero
-; RV32-NEXT:    vsetvli zero, a2, e64, m1, tu, ma
+; RV32-NEXT:    vsetvli zero, zero, e64, m1, tu, ma
 ; RV32-NEXT:    vredxor.vs v9, v8, v9, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v9
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vsrl.vx v8, v9, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_xor_v2i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli zero, a1, e64, m1, tu, ma
+; RV64-NEXT:    vsetvli zero, a1, e64, m1, ta, ma
+; RV64-NEXT:    vmv.v.x v9, a0
+; RV64-NEXT:    vsetvli zero, zero, e64, m1, tu, ma
 ; RV64-NEXT:    vredxor.vs v9, v8, v9, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v9
 ; RV64-NEXT:    ret
   %r = call i64 @llvm.vp.reduce.xor.v2i64(i64 %s, <2 x i64> %v, <2 x i1> %m, i32 %evl)
   ret i64 %r
 }
 
 declare i64 @llvm.vp.reduce.add.v4i64(i64, <4 x i64>, <4 x i1>, i32)

llvm/test/CodeGen/RISCV/rvv/fixed-vectors-reduction-int.ll

 
 declare i8 @llvm.vector.reduce.add.v2i8(<2 x i8>)
 
 define i8 @vreduce_add_v2i8(<2 x i8>* %x) {
 ; CHECK-LABEL: vreduce_add_v2i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 2, e8, mf8, ta, ma
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vmv.s.x v9, zero
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredsum.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <2 x i8>, <2 x i8>* %x
   %red = call i8 @llvm.vector.reduce.add.v2i8(<2 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.add.v4i8(<4 x i8>)
 
 define i8 @vreduce_add_v4i8(<4 x i8>* %x) {
 ; CHECK-LABEL: vreduce_add_v4i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 4, e8, mf4, ta, ma
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vmv.s.x v9, zero
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredsum.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x i8>, <4 x i8>* %x
   %red = call i8 @llvm.vector.reduce.add.v4i8(<4 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.add.v8i8(<8 x i8>)
 
 define i8 @vreduce_add_v8i8(<8 x i8>* %x) {
 ; CHECK-LABEL: vreduce_add_v8i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 8, e8, mf2, ta, ma
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vmv.s.x v9, zero
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredsum.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <8 x i8>, <8 x i8>* %x
   %red = call i8 @llvm.vector.reduce.add.v8i8(<8 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.add.v16i8(<16 x i8>)
 
 define i8 @vreduce_add_v16i8(<16 x i8>* %x) {
 ; CHECK-LABEL: vreduce_add_v16i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 16, e8, m1, ta, ma
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vmv.s.x v9, zero
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredsum.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <16 x i8>, <16 x i8>* %x
   %red = call i8 @llvm.vector.reduce.add.v16i8(<16 x i8> %v)
   ret i8 %red
 }
 

 
 declare i16 @llvm.vector.reduce.add.v2i16(<2 x i16>)
 
 define i16 @vreduce_add_v2i16(<2 x i16>* %x) {
 ; CHECK-LABEL: vreduce_add_v2i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, ma
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vmv.s.x v9, zero
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredsum.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <2 x i16>, <2 x i16>* %x
   %red = call i16 @llvm.vector.reduce.add.v2i16(<2 x i16> %v)
   ret i16 %red
 }
 
 define i16 @vwreduce_add_v2i16(<2 x i8>* %x) {
 ; CHECK-LABEL: vwreduce_add_v2i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 2, e8, mf8, ta, ma
+; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, ma
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
-; CHECK-NEXT:    vsetivli zero, 2, e8, mf8, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf8, ta, ma
 ; CHECK-NEXT:    vwredsum.vs v8, v8, v9
 ; CHECK-NEXT:    vsetivli zero, 0, e16, m1, ta, ma
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <2 x i8>, <2 x i8>* %x
   %e = sext <2 x i8> %v to <2 x i16>
   %red = call i16 @llvm.vector.reduce.add.v2i16(<2 x i16> %e)
   ret i16 %red
 }
 
 define i16 @vwreduce_uadd_v2i16(<2 x i8>* %x) {
 ; CHECK-LABEL: vwreduce_uadd_v2i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 2, e8, mf8, ta, ma
+; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, ma
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
-; CHECK-NEXT:    vsetivli zero, 2, e8, mf8, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf8, ta, ma
 ; CHECK-NEXT:    vwredsumu.vs v8, v8, v9
 ; CHECK-NEXT:    vsetivli zero, 0, e16, m1, ta, ma
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <2 x i8>, <2 x i8>* %x
   %e = zext <2 x i8> %v to <2 x i16>
   %red = call i16 @llvm.vector.reduce.add.v2i16(<2 x i16> %e)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.add.v4i16(<4 x i16>)
 
 define i16 @vreduce_add_v4i16(<4 x i16>* %x) {
 ; CHECK-LABEL: vreduce_add_v4i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, ma
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vmv.s.x v9, zero
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredsum.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x i16>, <4 x i16>* %x
   %red = call i16 @llvm.vector.reduce.add.v4i16(<4 x i16> %v)
   ret i16 %red
 }
 
 define i16 @vwreduce_add_v4i16(<4 x i8>* %x) {
 ; CHECK-LABEL: vwreduce_add_v4i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 4, e8, mf4, ta, ma
+; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, ma
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
-; CHECK-NEXT:    vsetivli zero, 4, e8, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf4, ta, ma
 ; CHECK-NEXT:    vwredsum.vs v8, v8, v9
 ; CHECK-NEXT:    vsetivli zero, 0, e16, m1, ta, ma
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x i8>, <4 x i8>* %x
   %e = sext <4 x i8> %v to <4 x i16>
   %red = call i16 @llvm.vector.reduce.add.v4i16(<4 x i16> %e)
   ret i16 %red
 }
 
 define i16 @vwreduce_uadd_v4i16(<4 x i8>* %x) {
 ; CHECK-LABEL: vwreduce_uadd_v4i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 4, e8, mf4, ta, ma
+; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, ma
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
-; CHECK-NEXT:    vsetivli zero, 4, e8, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf4, ta, ma
 ; CHECK-NEXT:    vwredsumu.vs v8, v8, v9
 ; CHECK-NEXT:    vsetivli zero, 0, e16, m1, ta, ma
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x i8>, <4 x i8>* %x
   %e = zext <4 x i8> %v to <4 x i16>
   %red = call i16 @llvm.vector.reduce.add.v4i16(<4 x i16> %e)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.add.v8i16(<8 x i16>)
 
 define i16 @vreduce_add_v8i16(<8 x i16>* %x) {
 ; CHECK-LABEL: vreduce_add_v8i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 8, e16, m1, ta, ma
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vmv.s.x v9, zero
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredsum.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <8 x i16>, <8 x i16>* %x
   %red = call i16 @llvm.vector.reduce.add.v8i16(<8 x i16> %v)
   ret i16 %red
 }
 
 define i16 @vwreduce_add_v8i16(<8 x i8>* %x) {
 ; CHECK-LABEL: vwreduce_add_v8i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 8, e16, m1, ta, ma
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vmv.s.x v9, zero
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli zero, zero, e8, mf2, ta, ma
 ; CHECK-NEXT:    vwredsum.vs v8, v8, v9
 ; CHECK-NEXT:    vsetvli zero, zero, e16, m1, ta, ma
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <8 x i8>, <8 x i8>* %x
   %e = sext <8 x i8> %v to <8 x i16>
   %red = call i16 @llvm.vector.reduce.add.v8i16(<8 x i16> %e)
   ret i16 %red
 }
 
 define i16 @vwreduce_uadd_v8i16(<8 x i8>* %x) {
 ; CHECK-LABEL: vwreduce_uadd_v8i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 8, e16, m1, ta, ma
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vmv.s.x v9, zero
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli zero, zero, e8, mf2, ta, ma
 ; CHECK-NEXT:    vwredsumu.vs v8, v8, v9
 ; CHECK-NEXT:    vsetvli zero, zero, e16, m1, ta, ma
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <8 x i8>, <8 x i8>* %x
   %e = zext <8 x i8> %v to <8 x i16>
   %red = call i16 @llvm.vector.reduce.add.v8i16(<8 x i16> %e)

 
 declare i32 @llvm.vector.reduce.add.v2i32(<2 x i32>)
 
 define i32 @vreduce_add_v2i32(<2 x i32>* %x) {
 ; CHECK-LABEL: vreduce_add_v2i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, ma
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vmv.s.x v9, zero
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredsum.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <2 x i32>, <2 x i32>* %x
   %red = call i32 @llvm.vector.reduce.add.v2i32(<2 x i32> %v)
   ret i32 %red
 }
 
 define i32 @vwreduce_add_v2i32(<2 x i16>* %x) {
 ; CHECK-LABEL: vwreduce_add_v2i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, ma
+; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, ma
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
-; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf4, ta, ma
 ; CHECK-NEXT:    vwredsum.vs v8, v8, v9
 ; CHECK-NEXT:    vsetivli zero, 0, e32, m1, ta, ma
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <2 x i16>, <2 x i16>* %x
   %e = sext <2 x i16> %v to <2 x i32>
   %red = call i32 @llvm.vector.reduce.add.v2i32(<2 x i32> %e)
   ret i32 %red
 }
 
 define i32 @vwreduce_uadd_v2i32(<2 x i16>* %x) {
 ; CHECK-LABEL: vwreduce_uadd_v2i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, ma
+; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, ma
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
-; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf4, ta, ma
 ; CHECK-NEXT:    vwredsumu.vs v8, v8, v9
 ; CHECK-NEXT:    vsetivli zero, 0, e32, m1, ta, ma
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <2 x i16>, <2 x i16>* %x
   %e = zext <2 x i16> %v to <2 x i32>
   %red = call i32 @llvm.vector.reduce.add.v2i32(<2 x i32> %e)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.add.v4i32(<4 x i32>)
 
 define i32 @vreduce_add_v4i32(<4 x i32>* %x) {
 ; CHECK-LABEL: vreduce_add_v4i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, ma
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vmv.s.x v9, zero
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredsum.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x i32>, <4 x i32>* %x
   %red = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %v)
   ret i32 %red
 }
 
 define i32 @vwreduce_add_v4i32(<4 x i16>* %x) {
 ; CHECK-LABEL: vwreduce_add_v4i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, ma
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vmv.s.x v9, zero
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli zero, zero, e16, mf2, ta, ma
 ; CHECK-NEXT:    vwredsum.vs v8, v8, v9
 ; CHECK-NEXT:    vsetvli zero, zero, e32, m1, ta, ma
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x i16>, <4 x i16>* %x
   %e = sext <4 x i16> %v to <4 x i32>
   %red = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %e)
   ret i32 %red
 }
 
 define i32 @vwreduce_uadd_v4i32(<4 x i16>* %x) {
 ; CHECK-LABEL: vwreduce_uadd_v4i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, ma
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vmv.s.x v9, zero
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli zero, zero, e16, mf2, ta, ma
 ; CHECK-NEXT:    vwredsumu.vs v8, v8, v9
 ; CHECK-NEXT:    vsetvli zero, zero, e32, m1, ta, ma
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x i16>, <4 x i16>* %x
   %e = zext <4 x i16> %v to <4 x i32>
   %red = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> %e)

 
 declare i64 @llvm.vector.reduce.add.v2i64(<2 x i64>)
 
 define i64 @vreduce_add_v2i64(<2 x i64>* %x) {
 ; RV32-LABEL: vreduce_add_v2i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
 ; RV32-NEXT:    vle64.v v8, (a0)
-; RV32-NEXT:    vmv.s.x v9, zero
+; RV32-NEXT:    vmv.v.i v9, 0
 ; RV32-NEXT:    vredsum.vs v8, v8, v9
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_add_v2i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
 ; RV64-NEXT:    vle64.v v8, (a0)
-; RV64-NEXT:    vmv.s.x v9, zero
+; RV64-NEXT:    vmv.v.i v9, 0
 ; RV64-NEXT:    vredsum.vs v8, v8, v9
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <2 x i64>, <2 x i64>* %x
   %red = call i64 @llvm.vector.reduce.add.v2i64(<2 x i64> %v)
   ret i64 %red
 }
 
 define i64 @vwreduce_add_v2i64(<2 x i32>* %x) {
 ; RV32-LABEL: vwreduce_add_v2i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
 ; RV32-NEXT:    vle32.v v8, (a0)
-; RV32-NEXT:    vmv.s.x v9, zero
+; RV32-NEXT:    vmv.v.i v9, 0
 ; RV32-NEXT:    vsetvli zero, zero, e32, mf2, ta, ma
 ; RV32-NEXT:    vwredsum.vs v8, v8, v9
 ; RV32-NEXT:    vsetvli zero, zero, e64, m1, ta, ma
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vwreduce_add_v2i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
 ; RV64-NEXT:    vle32.v v8, (a0)
-; RV64-NEXT:    vmv.s.x v9, zero
+; RV64-NEXT:    vmv.v.i v9, 0
 ; RV64-NEXT:    vsetvli zero, zero, e32, mf2, ta, ma
 ; RV64-NEXT:    vwredsum.vs v8, v8, v9
 ; RV64-NEXT:    vsetvli zero, zero, e64, m1, ta, ma
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <2 x i32>, <2 x i32>* %x
   %e = sext <2 x i32> %v to <2 x i64>
   %red = call i64 @llvm.vector.reduce.add.v2i64(<2 x i64> %e)
   ret i64 %red
 }
 
 define i64 @vwreduce_uadd_v2i64(<2 x i32>* %x) {
 ; RV32-LABEL: vwreduce_uadd_v2i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
 ; RV32-NEXT:    vle32.v v8, (a0)
-; RV32-NEXT:    vmv.s.x v9, zero
+; RV32-NEXT:    vmv.v.i v9, 0
 ; RV32-NEXT:    vsetvli zero, zero, e32, mf2, ta, ma
 ; RV32-NEXT:    vwredsumu.vs v8, v8, v9
 ; RV32-NEXT:    vsetvli zero, zero, e64, m1, ta, ma
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vwreduce_uadd_v2i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
 ; RV64-NEXT:    vle32.v v8, (a0)
-; RV64-NEXT:    vmv.s.x v9, zero
+; RV64-NEXT:    vmv.v.i v9, 0
 ; RV64-NEXT:    vsetvli zero, zero, e32, mf2, ta, ma
 ; RV64-NEXT:    vwredsumu.vs v8, v8, v9
 ; RV64-NEXT:    vsetvli zero, zero, e64, m1, ta, ma
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <2 x i32>, <2 x i32>* %x
   %e = zext <2 x i32> %v to <2 x i64>
   %red = call i64 @llvm.vector.reduce.add.v2i64(<2 x i64> %e)

 
 declare i8 @llvm.vector.reduce.and.v2i8(<2 x i8>)
 
 define i8 @vreduce_and_v2i8(<2 x i8>* %x) {
 ; CHECK-LABEL: vreduce_and_v2i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 2, e8, mf8, ta, ma
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
 ; CHECK-NEXT:    vmv.v.i v9, -1
-; CHECK-NEXT:    vsetivli zero, 2, e8, mf8, ta, ma
 ; CHECK-NEXT:    vredand.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <2 x i8>, <2 x i8>* %x
   %red = call i8 @llvm.vector.reduce.and.v2i8(<2 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.and.v4i8(<4 x i8>)
 
 define i8 @vreduce_and_v4i8(<4 x i8>* %x) {
 ; CHECK-LABEL: vreduce_and_v4i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 4, e8, mf4, ta, ma
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
 ; CHECK-NEXT:    vmv.v.i v9, -1
-; CHECK-NEXT:    vsetivli zero, 4, e8, mf4, ta, ma
 ; CHECK-NEXT:    vredand.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x i8>, <4 x i8>* %x
   %red = call i8 @llvm.vector.reduce.and.v4i8(<4 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.and.v8i8(<8 x i8>)
 
 define i8 @vreduce_and_v8i8(<8 x i8>* %x) {
 ; CHECK-LABEL: vreduce_and_v8i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 8, e8, mf2, ta, ma
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
 ; CHECK-NEXT:    vmv.v.i v9, -1
-; CHECK-NEXT:    vsetivli zero, 8, e8, mf2, ta, ma
 ; CHECK-NEXT:    vredand.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <8 x i8>, <8 x i8>* %x
   %red = call i8 @llvm.vector.reduce.and.v8i8(<8 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.and.v16i8(<16 x i8>)
 
 define i8 @vreduce_and_v16i8(<16 x i8>* %x) {
 ; CHECK-LABEL: vreduce_and_v16i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 16, e8, m1, ta, ma
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
 ; CHECK-NEXT:    vmv.v.i v9, -1
-; CHECK-NEXT:    vsetivli zero, 16, e8, m1, ta, ma
 ; CHECK-NEXT:    vredand.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <16 x i8>, <16 x i8>* %x
   %red = call i8 @llvm.vector.reduce.and.v16i8(<16 x i8> %v)
   ret i8 %red
 }
 

 
 declare i16 @llvm.vector.reduce.and.v2i16(<2 x i16>)
 
 define i16 @vreduce_and_v2i16(<2 x i16>* %x) {
 ; CHECK-LABEL: vreduce_and_v2i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, ma
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
 ; CHECK-NEXT:    vmv.v.i v9, -1
-; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, ma
 ; CHECK-NEXT:    vredand.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <2 x i16>, <2 x i16>* %x
   %red = call i16 @llvm.vector.reduce.and.v2i16(<2 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.and.v4i16(<4 x i16>)
 
 define i16 @vreduce_and_v4i16(<4 x i16>* %x) {
 ; CHECK-LABEL: vreduce_and_v4i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, ma
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
 ; CHECK-NEXT:    vmv.v.i v9, -1
-; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, ma
 ; CHECK-NEXT:    vredand.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x i16>, <4 x i16>* %x
   %red = call i16 @llvm.vector.reduce.and.v4i16(<4 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.and.v8i16(<8 x i16>)
 
 define i16 @vreduce_and_v8i16(<8 x i16>* %x) {
 ; CHECK-LABEL: vreduce_and_v8i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 8, e16, m1, ta, ma
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
 ; CHECK-NEXT:    vmv.v.i v9, -1
-; CHECK-NEXT:    vsetivli zero, 8, e16, m1, ta, ma
 ; CHECK-NEXT:    vredand.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <8 x i16>, <8 x i16>* %x
   %red = call i16 @llvm.vector.reduce.and.v8i16(<8 x i16> %v)
   ret i16 %red
 }
 

 
 declare i32 @llvm.vector.reduce.and.v2i32(<2 x i32>)
 
 define i32 @vreduce_and_v2i32(<2 x i32>* %x) {
 ; CHECK-LABEL: vreduce_and_v2i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, ma
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
 ; CHECK-NEXT:    vmv.v.i v9, -1
-; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, ma
 ; CHECK-NEXT:    vredand.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <2 x i32>, <2 x i32>* %x
   %red = call i32 @llvm.vector.reduce.and.v2i32(<2 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.and.v4i32(<4 x i32>)
 
 define i32 @vreduce_and_v4i32(<4 x i32>* %x) {
 ; CHECK-LABEL: vreduce_and_v4i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, ma
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
 ; CHECK-NEXT:    vmv.v.i v9, -1
-; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, ma
 ; CHECK-NEXT:    vredand.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x i32>, <4 x i32>* %x
   %red = call i32 @llvm.vector.reduce.and.v4i32(<4 x i32> %v)
   ret i32 %red
 }
 

 
 declare i64 @llvm.vector.reduce.and.v2i64(<2 x i64>)
 
 define i64 @vreduce_and_v2i64(<2 x i64>* %x) {
 ; RV32-LABEL: vreduce_and_v2i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
 ; RV32-NEXT:    vle64.v v8, (a0)
-; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vmv.v.i v9, -1
-; RV32-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
 ; RV32-NEXT:    vredand.vs v8, v8, v9
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_and_v2i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
 ; RV64-NEXT:    vle64.v v8, (a0)
-; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV64-NEXT:    vmv.v.i v9, -1
-; RV64-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
 ; RV64-NEXT:    vredand.vs v8, v8, v9
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <2 x i64>, <2 x i64>* %x
   %red = call i64 @llvm.vector.reduce.and.v2i64(<2 x i64> %v)
   ret i64 %red
 }
 

 
 declare i8 @llvm.vector.reduce.or.v2i8(<2 x i8>)
 
 define i8 @vreduce_or_v2i8(<2 x i8>* %x) {
 ; CHECK-LABEL: vreduce_or_v2i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 2, e8, mf8, ta, ma
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vmv.s.x v9, zero
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredor.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <2 x i8>, <2 x i8>* %x
   %red = call i8 @llvm.vector.reduce.or.v2i8(<2 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.or.v4i8(<4 x i8>)
 
 define i8 @vreduce_or_v4i8(<4 x i8>* %x) {
 ; CHECK-LABEL: vreduce_or_v4i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 4, e8, mf4, ta, ma
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vmv.s.x v9, zero
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredor.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x i8>, <4 x i8>* %x
   %red = call i8 @llvm.vector.reduce.or.v4i8(<4 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.or.v8i8(<8 x i8>)
 
 define i8 @vreduce_or_v8i8(<8 x i8>* %x) {
 ; CHECK-LABEL: vreduce_or_v8i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 8, e8, mf2, ta, ma
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vmv.s.x v9, zero
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredor.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <8 x i8>, <8 x i8>* %x
   %red = call i8 @llvm.vector.reduce.or.v8i8(<8 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.or.v16i8(<16 x i8>)
 
 define i8 @vreduce_or_v16i8(<16 x i8>* %x) {
 ; CHECK-LABEL: vreduce_or_v16i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 16, e8, m1, ta, ma
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vmv.s.x v9, zero
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredor.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <16 x i8>, <16 x i8>* %x
   %red = call i8 @llvm.vector.reduce.or.v16i8(<16 x i8> %v)
   ret i8 %red
 }
 

 
 declare i16 @llvm.vector.reduce.or.v2i16(<2 x i16>)
 
 define i16 @vreduce_or_v2i16(<2 x i16>* %x) {
 ; CHECK-LABEL: vreduce_or_v2i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, ma
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vmv.s.x v9, zero
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredor.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <2 x i16>, <2 x i16>* %x
   %red = call i16 @llvm.vector.reduce.or.v2i16(<2 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.or.v4i16(<4 x i16>)
 
 define i16 @vreduce_or_v4i16(<4 x i16>* %x) {
 ; CHECK-LABEL: vreduce_or_v4i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, ma
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vmv.s.x v9, zero
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredor.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x i16>, <4 x i16>* %x
   %red = call i16 @llvm.vector.reduce.or.v4i16(<4 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.or.v8i16(<8 x i16>)
 
 define i16 @vreduce_or_v8i16(<8 x i16>* %x) {
 ; CHECK-LABEL: vreduce_or_v8i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 8, e16, m1, ta, ma
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vmv.s.x v9, zero
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredor.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <8 x i16>, <8 x i16>* %x
   %red = call i16 @llvm.vector.reduce.or.v8i16(<8 x i16> %v)
   ret i16 %red
 }
 

 
 declare i32 @llvm.vector.reduce.or.v2i32(<2 x i32>)
 
 define i32 @vreduce_or_v2i32(<2 x i32>* %x) {
 ; CHECK-LABEL: vreduce_or_v2i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, ma
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vmv.s.x v9, zero
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredor.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <2 x i32>, <2 x i32>* %x
   %red = call i32 @llvm.vector.reduce.or.v2i32(<2 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.or.v4i32(<4 x i32>)
 
 define i32 @vreduce_or_v4i32(<4 x i32>* %x) {
 ; CHECK-LABEL: vreduce_or_v4i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, ma
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vmv.s.x v9, zero
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredor.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x i32>, <4 x i32>* %x
   %red = call i32 @llvm.vector.reduce.or.v4i32(<4 x i32> %v)
   ret i32 %red
 }
 

 
 declare i64 @llvm.vector.reduce.or.v2i64(<2 x i64>)
 
 define i64 @vreduce_or_v2i64(<2 x i64>* %x) {
 ; RV32-LABEL: vreduce_or_v2i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
 ; RV32-NEXT:    vle64.v v8, (a0)
-; RV32-NEXT:    vmv.s.x v9, zero
+; RV32-NEXT:    vmv.v.i v9, 0
 ; RV32-NEXT:    vredor.vs v8, v8, v9
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_or_v2i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
 ; RV64-NEXT:    vle64.v v8, (a0)
-; RV64-NEXT:    vmv.s.x v9, zero
+; RV64-NEXT:    vmv.v.i v9, 0
 ; RV64-NEXT:    vredor.vs v8, v8, v9
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <2 x i64>, <2 x i64>* %x
   %red = call i64 @llvm.vector.reduce.or.v2i64(<2 x i64> %v)
   ret i64 %red
 }
 

 
 declare i8 @llvm.vector.reduce.xor.v2i8(<2 x i8>)
 
 define i8 @vreduce_xor_v2i8(<2 x i8>* %x) {
 ; CHECK-LABEL: vreduce_xor_v2i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 2, e8, mf8, ta, ma
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vmv.s.x v9, zero
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredxor.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <2 x i8>, <2 x i8>* %x
   %red = call i8 @llvm.vector.reduce.xor.v2i8(<2 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.xor.v4i8(<4 x i8>)
 
 define i8 @vreduce_xor_v4i8(<4 x i8>* %x) {
 ; CHECK-LABEL: vreduce_xor_v4i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 4, e8, mf4, ta, ma
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vmv.s.x v9, zero
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredxor.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x i8>, <4 x i8>* %x
   %red = call i8 @llvm.vector.reduce.xor.v4i8(<4 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.xor.v8i8(<8 x i8>)
 
 define i8 @vreduce_xor_v8i8(<8 x i8>* %x) {
 ; CHECK-LABEL: vreduce_xor_v8i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 8, e8, mf2, ta, ma
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vmv.s.x v9, zero
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredxor.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <8 x i8>, <8 x i8>* %x
   %red = call i8 @llvm.vector.reduce.xor.v8i8(<8 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.xor.v16i8(<16 x i8>)
 
 define i8 @vreduce_xor_v16i8(<16 x i8>* %x) {
 ; CHECK-LABEL: vreduce_xor_v16i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 16, e8, m1, ta, ma
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vmv.s.x v9, zero
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredxor.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <16 x i8>, <16 x i8>* %x
   %red = call i8 @llvm.vector.reduce.xor.v16i8(<16 x i8> %v)
   ret i8 %red
 }
 

 
 declare i16 @llvm.vector.reduce.xor.v2i16(<2 x i16>)
 
 define i16 @vreduce_xor_v2i16(<2 x i16>* %x) {
 ; CHECK-LABEL: vreduce_xor_v2i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, ma
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vmv.s.x v9, zero
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredxor.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <2 x i16>, <2 x i16>* %x
   %red = call i16 @llvm.vector.reduce.xor.v2i16(<2 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.xor.v4i16(<4 x i16>)
 
 define i16 @vreduce_xor_v4i16(<4 x i16>* %x) {
 ; CHECK-LABEL: vreduce_xor_v4i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, ma
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vmv.s.x v9, zero
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredxor.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x i16>, <4 x i16>* %x
   %red = call i16 @llvm.vector.reduce.xor.v4i16(<4 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.xor.v8i16(<8 x i16>)
 
 define i16 @vreduce_xor_v8i16(<8 x i16>* %x) {
 ; CHECK-LABEL: vreduce_xor_v8i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 8, e16, m1, ta, ma
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vmv.s.x v9, zero
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredxor.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <8 x i16>, <8 x i16>* %x
   %red = call i16 @llvm.vector.reduce.xor.v8i16(<8 x i16> %v)
   ret i16 %red
 }
 

 
 declare i32 @llvm.vector.reduce.xor.v2i32(<2 x i32>)
 
 define i32 @vreduce_xor_v2i32(<2 x i32>* %x) {
 ; CHECK-LABEL: vreduce_xor_v2i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, ma
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vmv.s.x v9, zero
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredxor.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <2 x i32>, <2 x i32>* %x
   %red = call i32 @llvm.vector.reduce.xor.v2i32(<2 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.xor.v4i32(<4 x i32>)
 
 define i32 @vreduce_xor_v4i32(<4 x i32>* %x) {
 ; CHECK-LABEL: vreduce_xor_v4i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, ma
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vmv.s.x v9, zero
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredxor.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x i32>, <4 x i32>* %x
   %red = call i32 @llvm.vector.reduce.xor.v4i32(<4 x i32> %v)
   ret i32 %red
 }
 

 
 declare i64 @llvm.vector.reduce.xor.v2i64(<2 x i64>)
 
 define i64 @vreduce_xor_v2i64(<2 x i64>* %x) {
 ; RV32-LABEL: vreduce_xor_v2i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
 ; RV32-NEXT:    vle64.v v8, (a0)
-; RV32-NEXT:    vmv.s.x v9, zero
+; RV32-NEXT:    vmv.v.i v9, 0
 ; RV32-NEXT:    vredxor.vs v8, v8, v9
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_xor_v2i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
 ; RV64-NEXT:    vle64.v v8, (a0)
-; RV64-NEXT:    vmv.s.x v9, zero
+; RV64-NEXT:    vmv.v.i v9, 0
 ; RV64-NEXT:    vredxor.vs v8, v8, v9
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <2 x i64>, <2 x i64>* %x
   %red = call i64 @llvm.vector.reduce.xor.v2i64(<2 x i64> %v)
   ret i64 %red
 }
 

 declare i8 @llvm.vector.reduce.smin.v2i8(<2 x i8>)
 
 define i8 @vreduce_smin_v2i8(<2 x i8>* %x) {
 ; CHECK-LABEL: vreduce_smin_v2i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 2, e8, mf8, ta, ma
 ; CHECK-NEXT:    vle8.v v8, (a0)
 ; CHECK-NEXT:    li a0, 127
-; CHECK-NEXT:    vmv.s.x v9, a0
+; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vredmin.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <2 x i8>, <2 x i8>* %x
   %red = call i8 @llvm.vector.reduce.smin.v2i8(<2 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.smin.v4i8(<4 x i8>)
 
 define i8 @vreduce_smin_v4i8(<4 x i8>* %x) {
 ; CHECK-LABEL: vreduce_smin_v4i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 4, e8, mf4, ta, ma
 ; CHECK-NEXT:    vle8.v v8, (a0)
 ; CHECK-NEXT:    li a0, 127
-; CHECK-NEXT:    vmv.s.x v9, a0
+; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vredmin.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x i8>, <4 x i8>* %x
   %red = call i8 @llvm.vector.reduce.smin.v4i8(<4 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.smin.v8i8(<8 x i8>)
 
 define i8 @vreduce_smin_v8i8(<8 x i8>* %x) {
 ; CHECK-LABEL: vreduce_smin_v8i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 8, e8, mf2, ta, ma
 ; CHECK-NEXT:    vle8.v v8, (a0)
 ; CHECK-NEXT:    li a0, 127
-; CHECK-NEXT:    vmv.s.x v9, a0
+; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vredmin.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <8 x i8>, <8 x i8>* %x
   %red = call i8 @llvm.vector.reduce.smin.v8i8(<8 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.smin.v16i8(<16 x i8>)
 
 define i8 @vreduce_smin_v16i8(<16 x i8>* %x) {
 ; CHECK-LABEL: vreduce_smin_v16i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 16, e8, m1, ta, ma
 ; CHECK-NEXT:    vle8.v v8, (a0)
 ; CHECK-NEXT:    li a0, 127
-; CHECK-NEXT:    vmv.s.x v9, a0
+; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vredmin.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <16 x i8>, <16 x i8>* %x
   %red = call i8 @llvm.vector.reduce.smin.v16i8(<16 x i8> %v)
   ret i8 %red
 }
 

 
 define i16 @vreduce_smin_v2i16(<2 x i16>* %x) {
 ; RV32-LABEL: vreduce_smin_v2i16:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    vsetivli zero, 2, e16, mf4, ta, ma
 ; RV32-NEXT:    vle16.v v8, (a0)
 ; RV32-NEXT:    lui a0, 8
 ; RV32-NEXT:    addi a0, a0, -1
-; RV32-NEXT:    vmv.s.x v9, a0
+; RV32-NEXT:    vmv.v.x v9, a0
 ; RV32-NEXT:    vredmin.vs v8, v8, v9
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_smin_v2i16:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 2, e16, mf4, ta, ma
 ; RV64-NEXT:    vle16.v v8, (a0)
 ; RV64-NEXT:    lui a0, 8
 ; RV64-NEXT:    addiw a0, a0, -1
-; RV64-NEXT:    vmv.s.x v9, a0
+; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vredmin.vs v8, v8, v9
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <2 x i16>, <2 x i16>* %x
   %red = call i16 @llvm.vector.reduce.smin.v2i16(<2 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.smin.v4i16(<4 x i16>)
 
 define i16 @vreduce_smin_v4i16(<4 x i16>* %x) {
 ; RV32-LABEL: vreduce_smin_v4i16:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    vsetivli zero, 4, e16, mf2, ta, ma
 ; RV32-NEXT:    vle16.v v8, (a0)
 ; RV32-NEXT:    lui a0, 8
 ; RV32-NEXT:    addi a0, a0, -1
-; RV32-NEXT:    vmv.s.x v9, a0
+; RV32-NEXT:    vmv.v.x v9, a0
 ; RV32-NEXT:    vredmin.vs v8, v8, v9
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_smin_v4i16:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 4, e16, mf2, ta, ma
 ; RV64-NEXT:    vle16.v v8, (a0)
 ; RV64-NEXT:    lui a0, 8
 ; RV64-NEXT:    addiw a0, a0, -1
-; RV64-NEXT:    vmv.s.x v9, a0
+; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vredmin.vs v8, v8, v9
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <4 x i16>, <4 x i16>* %x
   %red = call i16 @llvm.vector.reduce.smin.v4i16(<4 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.smin.v8i16(<8 x i16>)
 
 define i16 @vreduce_smin_v8i16(<8 x i16>* %x) {
 ; RV32-LABEL: vreduce_smin_v8i16:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    vsetivli zero, 8, e16, m1, ta, ma
 ; RV32-NEXT:    vle16.v v8, (a0)
 ; RV32-NEXT:    lui a0, 8
 ; RV32-NEXT:    addi a0, a0, -1
-; RV32-NEXT:    vmv.s.x v9, a0
+; RV32-NEXT:    vmv.v.x v9, a0
 ; RV32-NEXT:    vredmin.vs v8, v8, v9
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_smin_v8i16:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 8, e16, m1, ta, ma
 ; RV64-NEXT:    vle16.v v8, (a0)
 ; RV64-NEXT:    lui a0, 8
 ; RV64-NEXT:    addiw a0, a0, -1
-; RV64-NEXT:    vmv.s.x v9, a0
+; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vredmin.vs v8, v8, v9
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <8 x i16>, <8 x i16>* %x
   %red = call i16 @llvm.vector.reduce.smin.v8i16(<8 x i16> %v)
   ret i16 %red
 }
 

 
 define i32 @vreduce_smin_v2i32(<2 x i32>* %x) {
 ; RV32-LABEL: vreduce_smin_v2i32:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    vsetivli zero, 2, e32, mf2, ta, ma
 ; RV32-NEXT:    vle32.v v8, (a0)
 ; RV32-NEXT:    lui a0, 524288
 ; RV32-NEXT:    addi a0, a0, -1
-; RV32-NEXT:    vmv.s.x v9, a0
+; RV32-NEXT:    vmv.v.x v9, a0
 ; RV32-NEXT:    vredmin.vs v8, v8, v9
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_smin_v2i32:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 2, e32, mf2, ta, ma
 ; RV64-NEXT:    vle32.v v8, (a0)
 ; RV64-NEXT:    lui a0, 524288
 ; RV64-NEXT:    addiw a0, a0, -1
-; RV64-NEXT:    vmv.s.x v9, a0
+; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vredmin.vs v8, v8, v9
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <2 x i32>, <2 x i32>* %x
   %red = call i32 @llvm.vector.reduce.smin.v2i32(<2 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.smin.v4i32(<4 x i32>)
 
 define i32 @vreduce_smin_v4i32(<4 x i32>* %x) {
 ; RV32-LABEL: vreduce_smin_v4i32:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    vsetivli zero, 4, e32, m1, ta, ma
 ; RV32-NEXT:    vle32.v v8, (a0)
 ; RV32-NEXT:    lui a0, 524288
 ; RV32-NEXT:    addi a0, a0, -1
-; RV32-NEXT:    vmv.s.x v9, a0
+; RV32-NEXT:    vmv.v.x v9, a0
 ; RV32-NEXT:    vredmin.vs v8, v8, v9
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_smin_v4i32:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 4, e32, m1, ta, ma
 ; RV64-NEXT:    vle32.v v8, (a0)
 ; RV64-NEXT:    lui a0, 524288
 ; RV64-NEXT:    addiw a0, a0, -1
-; RV64-NEXT:    vmv.s.x v9, a0
+; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vredmin.vs v8, v8, v9
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <4 x i32>, <4 x i32>* %x
   %red = call i32 @llvm.vector.reduce.smin.v4i32(<4 x i32> %v)
   ret i32 %red
 }
 

 ; RV32-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
 ; RV32-NEXT:    vle64.v v8, (a0)
 ; RV32-NEXT:    li a0, -1
 ; RV32-NEXT:    sw a0, 8(sp)
 ; RV32-NEXT:    lui a0, 524288
 ; RV32-NEXT:    addi a0, a0, -1
 ; RV32-NEXT:    sw a0, 12(sp)
 ; RV32-NEXT:    addi a0, sp, 8
-; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vlse64.v v9, (a0), zero
-; RV32-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
 ; RV32-NEXT:    vredmin.vs v8, v8, v9
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_smin_v2i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
 ; RV64-NEXT:    vle64.v v8, (a0)
 ; RV64-NEXT:    li a0, -1
 ; RV64-NEXT:    srli a0, a0, 1
-; RV64-NEXT:    vmv.s.x v9, a0
+; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vredmin.vs v8, v8, v9
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <2 x i64>, <2 x i64>* %x
   %red = call i64 @llvm.vector.reduce.smin.v2i64(<2 x i64> %v)
   ret i64 %red
 }
 

 declare i8 @llvm.vector.reduce.smax.v2i8(<2 x i8>)
 
 define i8 @vreduce_smax_v2i8(<2 x i8>* %x) {
 ; CHECK-LABEL: vreduce_smax_v2i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 2, e8, mf8, ta, ma
 ; CHECK-NEXT:    vle8.v v8, (a0)
 ; CHECK-NEXT:    li a0, -128
-; CHECK-NEXT:    vmv.s.x v9, a0
+; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vredmax.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <2 x i8>, <2 x i8>* %x
   %red = call i8 @llvm.vector.reduce.smax.v2i8(<2 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.smax.v4i8(<4 x i8>)
 
 define i8 @vreduce_smax_v4i8(<4 x i8>* %x) {
 ; CHECK-LABEL: vreduce_smax_v4i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 4, e8, mf4, ta, ma
 ; CHECK-NEXT:    vle8.v v8, (a0)
 ; CHECK-NEXT:    li a0, -128
-; CHECK-NEXT:    vmv.s.x v9, a0
+; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vredmax.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x i8>, <4 x i8>* %x
   %red = call i8 @llvm.vector.reduce.smax.v4i8(<4 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.smax.v8i8(<8 x i8>)
 
 define i8 @vreduce_smax_v8i8(<8 x i8>* %x) {
 ; CHECK-LABEL: vreduce_smax_v8i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 8, e8, mf2, ta, ma
 ; CHECK-NEXT:    vle8.v v8, (a0)
 ; CHECK-NEXT:    li a0, -128
-; CHECK-NEXT:    vmv.s.x v9, a0
+; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vredmax.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <8 x i8>, <8 x i8>* %x
   %red = call i8 @llvm.vector.reduce.smax.v8i8(<8 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.smax.v16i8(<16 x i8>)
 
 define i8 @vreduce_smax_v16i8(<16 x i8>* %x) {
 ; CHECK-LABEL: vreduce_smax_v16i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 16, e8, m1, ta, ma
 ; CHECK-NEXT:    vle8.v v8, (a0)
 ; CHECK-NEXT:    li a0, -128
-; CHECK-NEXT:    vmv.s.x v9, a0
+; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vredmax.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <16 x i8>, <16 x i8>* %x
   %red = call i8 @llvm.vector.reduce.smax.v16i8(<16 x i8> %v)
   ret i8 %red
 }
 

 declare i16 @llvm.vector.reduce.smax.v2i16(<2 x i16>)
 
 define i16 @vreduce_smax_v2i16(<2 x i16>* %x) {
 ; CHECK-LABEL: vreduce_smax_v2i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, ma
 ; CHECK-NEXT:    vle16.v v8, (a0)
 ; CHECK-NEXT:    lui a0, 1048568
-; CHECK-NEXT:    vmv.s.x v9, a0
+; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vredmax.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <2 x i16>, <2 x i16>* %x
   %red = call i16 @llvm.vector.reduce.smax.v2i16(<2 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.smax.v4i16(<4 x i16>)
 
 define i16 @vreduce_smax_v4i16(<4 x i16>* %x) {
 ; CHECK-LABEL: vreduce_smax_v4i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, ma
 ; CHECK-NEXT:    vle16.v v8, (a0)
 ; CHECK-NEXT:    lui a0, 1048568
-; CHECK-NEXT:    vmv.s.x v9, a0
+; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vredmax.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x i16>, <4 x i16>* %x
   %red = call i16 @llvm.vector.reduce.smax.v4i16(<4 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.smax.v8i16(<8 x i16>)
 
 define i16 @vreduce_smax_v8i16(<8 x i16>* %x) {
 ; CHECK-LABEL: vreduce_smax_v8i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 8, e16, m1, ta, ma
 ; CHECK-NEXT:    vle16.v v8, (a0)
 ; CHECK-NEXT:    lui a0, 1048568
-; CHECK-NEXT:    vmv.s.x v9, a0
+; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vredmax.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <8 x i16>, <8 x i16>* %x
   %red = call i16 @llvm.vector.reduce.smax.v8i16(<8 x i16> %v)
   ret i16 %red
 }
 

 declare i32 @llvm.vector.reduce.smax.v2i32(<2 x i32>)
 
 define i32 @vreduce_smax_v2i32(<2 x i32>* %x) {
 ; CHECK-LABEL: vreduce_smax_v2i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, ma
 ; CHECK-NEXT:    vle32.v v8, (a0)
 ; CHECK-NEXT:    lui a0, 524288
-; CHECK-NEXT:    vmv.s.x v9, a0
+; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vredmax.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <2 x i32>, <2 x i32>* %x
   %red = call i32 @llvm.vector.reduce.smax.v2i32(<2 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.smax.v4i32(<4 x i32>)
 
 define i32 @vreduce_smax_v4i32(<4 x i32>* %x) {
 ; CHECK-LABEL: vreduce_smax_v4i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, ma
 ; CHECK-NEXT:    vle32.v v8, (a0)
 ; CHECK-NEXT:    lui a0, 524288
-; CHECK-NEXT:    vmv.s.x v9, a0
+; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vredmax.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x i32>, <4 x i32>* %x
   %red = call i32 @llvm.vector.reduce.smax.v4i32(<4 x i32> %v)
   ret i32 %red
 }
 

 ; RV32-NEXT:    addi sp, sp, -16
 ; RV32-NEXT:    .cfi_def_cfa_offset 16
 ; RV32-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
 ; RV32-NEXT:    vle64.v v8, (a0)
 ; RV32-NEXT:    lui a0, 524288
 ; RV32-NEXT:    sw a0, 12(sp)
 ; RV32-NEXT:    sw zero, 8(sp)
 ; RV32-NEXT:    addi a0, sp, 8
-; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vlse64.v v9, (a0), zero
-; RV32-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
 ; RV32-NEXT:    vredmax.vs v8, v8, v9
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_smax_v2i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
 ; RV64-NEXT:    vle64.v v8, (a0)
 ; RV64-NEXT:    li a0, -1
 ; RV64-NEXT:    slli a0, a0, 63
-; RV64-NEXT:    vmv.s.x v9, a0
+; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vredmax.vs v8, v8, v9
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <2 x i64>, <2 x i64>* %x
   %red = call i64 @llvm.vector.reduce.smax.v2i64(<2 x i64> %v)
   ret i64 %red
 }
 

 
 declare i8 @llvm.vector.reduce.umin.v2i8(<2 x i8>)
 
 define i8 @vreduce_umin_v2i8(<2 x i8>* %x) {
 ; CHECK-LABEL: vreduce_umin_v2i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 2, e8, mf8, ta, ma
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
 ; CHECK-NEXT:    vmv.v.i v9, -1
-; CHECK-NEXT:    vsetivli zero, 2, e8, mf8, ta, ma
 ; CHECK-NEXT:    vredminu.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <2 x i8>, <2 x i8>* %x
   %red = call i8 @llvm.vector.reduce.umin.v2i8(<2 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.umin.v4i8(<4 x i8>)
 
 define i8 @vreduce_umin_v4i8(<4 x i8>* %x) {
 ; CHECK-LABEL: vreduce_umin_v4i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 4, e8, mf4, ta, ma
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
 ; CHECK-NEXT:    vmv.v.i v9, -1
-; CHECK-NEXT:    vsetivli zero, 4, e8, mf4, ta, ma
 ; CHECK-NEXT:    vredminu.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x i8>, <4 x i8>* %x
   %red = call i8 @llvm.vector.reduce.umin.v4i8(<4 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.umin.v8i8(<8 x i8>)
 
 define i8 @vreduce_umin_v8i8(<8 x i8>* %x) {
 ; CHECK-LABEL: vreduce_umin_v8i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 8, e8, mf2, ta, ma
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
 ; CHECK-NEXT:    vmv.v.i v9, -1
-; CHECK-NEXT:    vsetivli zero, 8, e8, mf2, ta, ma
 ; CHECK-NEXT:    vredminu.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <8 x i8>, <8 x i8>* %x
   %red = call i8 @llvm.vector.reduce.umin.v8i8(<8 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.umin.v16i8(<16 x i8>)
 
 define i8 @vreduce_umin_v16i8(<16 x i8>* %x) {
 ; CHECK-LABEL: vreduce_umin_v16i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 16, e8, m1, ta, ma
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
 ; CHECK-NEXT:    vmv.v.i v9, -1
-; CHECK-NEXT:    vsetivli zero, 16, e8, m1, ta, ma
 ; CHECK-NEXT:    vredminu.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <16 x i8>, <16 x i8>* %x
   %red = call i8 @llvm.vector.reduce.umin.v16i8(<16 x i8> %v)
   ret i8 %red
 }
 

 
 declare i16 @llvm.vector.reduce.umin.v2i16(<2 x i16>)
 
 define i16 @vreduce_umin_v2i16(<2 x i16>* %x) {
 ; CHECK-LABEL: vreduce_umin_v2i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, ma
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
 ; CHECK-NEXT:    vmv.v.i v9, -1
-; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, ma
 ; CHECK-NEXT:    vredminu.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <2 x i16>, <2 x i16>* %x
   %red = call i16 @llvm.vector.reduce.umin.v2i16(<2 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.umin.v4i16(<4 x i16>)
 
 define i16 @vreduce_umin_v4i16(<4 x i16>* %x) {
 ; CHECK-LABEL: vreduce_umin_v4i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, ma
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
 ; CHECK-NEXT:    vmv.v.i v9, -1
-; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, ma
 ; CHECK-NEXT:    vredminu.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x i16>, <4 x i16>* %x
   %red = call i16 @llvm.vector.reduce.umin.v4i16(<4 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.umin.v8i16(<8 x i16>)
 
 define i16 @vreduce_umin_v8i16(<8 x i16>* %x) {
 ; CHECK-LABEL: vreduce_umin_v8i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 8, e16, m1, ta, ma
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
 ; CHECK-NEXT:    vmv.v.i v9, -1
-; CHECK-NEXT:    vsetivli zero, 8, e16, m1, ta, ma
 ; CHECK-NEXT:    vredminu.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <8 x i16>, <8 x i16>* %x
   %red = call i16 @llvm.vector.reduce.umin.v8i16(<8 x i16> %v)
   ret i16 %red
 }
 

 
 declare i32 @llvm.vector.reduce.umin.v2i32(<2 x i32>)
 
 define i32 @vreduce_umin_v2i32(<2 x i32>* %x) {
 ; CHECK-LABEL: vreduce_umin_v2i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, ma
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
 ; CHECK-NEXT:    vmv.v.i v9, -1
-; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, ma
 ; CHECK-NEXT:    vredminu.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <2 x i32>, <2 x i32>* %x
   %red = call i32 @llvm.vector.reduce.umin.v2i32(<2 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.umin.v4i32(<4 x i32>)
 
 define i32 @vreduce_umin_v4i32(<4 x i32>* %x) {
 ; CHECK-LABEL: vreduce_umin_v4i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, ma
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
 ; CHECK-NEXT:    vmv.v.i v9, -1
-; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, ma
 ; CHECK-NEXT:    vredminu.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x i32>, <4 x i32>* %x
   %red = call i32 @llvm.vector.reduce.umin.v4i32(<4 x i32> %v)
   ret i32 %red
 }
 

 
 declare i64 @llvm.vector.reduce.umin.v2i64(<2 x i64>)
 
 define i64 @vreduce_umin_v2i64(<2 x i64>* %x) {
 ; RV32-LABEL: vreduce_umin_v2i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
 ; RV32-NEXT:    vle64.v v8, (a0)
-; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vmv.v.i v9, -1
-; RV32-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
 ; RV32-NEXT:    vredminu.vs v8, v8, v9
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_umin_v2i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
 ; RV64-NEXT:    vle64.v v8, (a0)
-; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV64-NEXT:    vmv.v.i v9, -1
-; RV64-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
 ; RV64-NEXT:    vredminu.vs v8, v8, v9
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <2 x i64>, <2 x i64>* %x
   %red = call i64 @llvm.vector.reduce.umin.v2i64(<2 x i64> %v)
   ret i64 %red
 }
 

 
 declare i8 @llvm.vector.reduce.umax.v2i8(<2 x i8>)
 
 define i8 @vreduce_umax_v2i8(<2 x i8>* %x) {
 ; CHECK-LABEL: vreduce_umax_v2i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 2, e8, mf8, ta, ma
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vmv.s.x v9, zero
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredmaxu.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <2 x i8>, <2 x i8>* %x
   %red = call i8 @llvm.vector.reduce.umax.v2i8(<2 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.umax.v4i8(<4 x i8>)
 
 define i8 @vreduce_umax_v4i8(<4 x i8>* %x) {
 ; CHECK-LABEL: vreduce_umax_v4i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 4, e8, mf4, ta, ma
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vmv.s.x v9, zero
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredmaxu.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x i8>, <4 x i8>* %x
   %red = call i8 @llvm.vector.reduce.umax.v4i8(<4 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.umax.v8i8(<8 x i8>)
 
 define i8 @vreduce_umax_v8i8(<8 x i8>* %x) {
 ; CHECK-LABEL: vreduce_umax_v8i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 8, e8, mf2, ta, ma
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vmv.s.x v9, zero
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredmaxu.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <8 x i8>, <8 x i8>* %x
   %red = call i8 @llvm.vector.reduce.umax.v8i8(<8 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.umax.v16i8(<16 x i8>)
 
 define i8 @vreduce_umax_v16i8(<16 x i8>* %x) {
 ; CHECK-LABEL: vreduce_umax_v16i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 16, e8, m1, ta, ma
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vmv.s.x v9, zero
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredmaxu.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <16 x i8>, <16 x i8>* %x
   %red = call i8 @llvm.vector.reduce.umax.v16i8(<16 x i8> %v)
   ret i8 %red
 }
 

 
 declare i16 @llvm.vector.reduce.umax.v2i16(<2 x i16>)
 
 define i16 @vreduce_umax_v2i16(<2 x i16>* %x) {
 ; CHECK-LABEL: vreduce_umax_v2i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, ma
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vmv.s.x v9, zero
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredmaxu.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <2 x i16>, <2 x i16>* %x
   %red = call i16 @llvm.vector.reduce.umax.v2i16(<2 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.umax.v4i16(<4 x i16>)
 
 define i16 @vreduce_umax_v4i16(<4 x i16>* %x) {
 ; CHECK-LABEL: vreduce_umax_v4i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, ma
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vmv.s.x v9, zero
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredmaxu.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x i16>, <4 x i16>* %x
   %red = call i16 @llvm.vector.reduce.umax.v4i16(<4 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.umax.v8i16(<8 x i16>)
 
 define i16 @vreduce_umax_v8i16(<8 x i16>* %x) {
 ; CHECK-LABEL: vreduce_umax_v8i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 8, e16, m1, ta, ma
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vmv.s.x v9, zero
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredmaxu.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <8 x i16>, <8 x i16>* %x
   %red = call i16 @llvm.vector.reduce.umax.v8i16(<8 x i16> %v)
   ret i16 %red
 }
 

 
 declare i32 @llvm.vector.reduce.umax.v2i32(<2 x i32>)
 
 define i32 @vreduce_umax_v2i32(<2 x i32>* %x) {
 ; CHECK-LABEL: vreduce_umax_v2i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, ma
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vmv.s.x v9, zero
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredmaxu.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <2 x i32>, <2 x i32>* %x
   %red = call i32 @llvm.vector.reduce.umax.v2i32(<2 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.umax.v4i32(<4 x i32>)
 
 define i32 @vreduce_umax_v4i32(<4 x i32>* %x) {
 ; CHECK-LABEL: vreduce_umax_v4i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, ma
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vmv.s.x v9, zero
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredmaxu.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x i32>, <4 x i32>* %x
   %red = call i32 @llvm.vector.reduce.umax.v4i32(<4 x i32> %v)
   ret i32 %red
 }
 

 
 declare i64 @llvm.vector.reduce.umax.v2i64(<2 x i64>)
 
 define i64 @vreduce_umax_v2i64(<2 x i64>* %x) {
 ; RV32-LABEL: vreduce_umax_v2i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
 ; RV32-NEXT:    vle64.v v8, (a0)
-; RV32-NEXT:    vmv.s.x v9, zero
+; RV32-NEXT:    vmv.v.i v9, 0
 ; RV32-NEXT:    vredmaxu.vs v8, v8, v9
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_umax_v2i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 2, e64, m1, ta, ma
 ; RV64-NEXT:    vle64.v v8, (a0)
-; RV64-NEXT:    vmv.s.x v9, zero
+; RV64-NEXT:    vmv.v.i v9, 0
 ; RV64-NEXT:    vredmaxu.vs v8, v8, v9
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <2 x i64>, <2 x i64>* %x
   %red = call i64 @llvm.vector.reduce.umax.v2i64(<2 x i64> %v)
   ret i64 %red
 }
 

llvm/test/CodeGen/RISCV/rvv/vreductions-fp-vp.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+experimental-zvfh,+v -target-abi=ilp32d \
 ; RUN:     -verify-machineinstrs < %s | FileCheck %s
 ; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+experimental-zvfh,+v -target-abi=lp64d \
 ; RUN:     -verify-machineinstrs < %s | FileCheck %s
 
 declare half @llvm.vp.reduce.fadd.nxv1f16(half, <vscale x 1 x half>, <vscale x 1 x i1>, i32)
 
 define half @vpreduce_fadd_nxv1f16(half %s, <vscale x 1 x half> %v, <vscale x 1 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_fadd_nxv1f16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vfmv.s.f v9, fa0
-; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, tu, ma
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, ma
+; CHECK-NEXT:    vfmv.v.f v9, fa0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf4, tu, ma
 ; CHECK-NEXT:    vfredusum.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vfmv.f.s fa0, v9
 ; CHECK-NEXT:    ret
   %r = call reassoc half @llvm.vp.reduce.fadd.nxv1f16(half %s, <vscale x 1 x half> %v, <vscale x 1 x i1> %m, i32 %evl)
   ret half %r
 }
 
 define half @vpreduce_ord_fadd_nxv1f16(half %s, <vscale x 1 x half> %v, <vscale x 1 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_ord_fadd_nxv1f16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vfmv.s.f v9, fa0
-; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, tu, ma
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, ma
+; CHECK-NEXT:    vfmv.v.f v9, fa0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf4, tu, ma
 ; CHECK-NEXT:    vfredosum.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vfmv.f.s fa0, v9
 ; CHECK-NEXT:    ret
   %r = call half @llvm.vp.reduce.fadd.nxv1f16(half %s, <vscale x 1 x half> %v, <vscale x 1 x i1> %m, i32 %evl)
   ret half %r
 }
 
 declare half @llvm.vp.reduce.fadd.nxv2f16(half, <vscale x 2 x half>, <vscale x 2 x i1>, i32)
 
 define half @vpreduce_fadd_nxv2f16(half %s, <vscale x 2 x half> %v, <vscale x 2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_fadd_nxv2f16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vfmv.s.f v9, fa0
-; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, tu, ma
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, ma
+; CHECK-NEXT:    vfmv.v.f v9, fa0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf2, tu, ma
 ; CHECK-NEXT:    vfredusum.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vfmv.f.s fa0, v9
 ; CHECK-NEXT:    ret
   %r = call reassoc half @llvm.vp.reduce.fadd.nxv2f16(half %s, <vscale x 2 x half> %v, <vscale x 2 x i1> %m, i32 %evl)
   ret half %r
 }
 
 define half @vpreduce_ord_fadd_nxv2f16(half %s, <vscale x 2 x half> %v, <vscale x 2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_ord_fadd_nxv2f16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vfmv.s.f v9, fa0
-; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, tu, ma
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, ma
+; CHECK-NEXT:    vfmv.v.f v9, fa0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf2, tu, ma
 ; CHECK-NEXT:    vfredosum.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vfmv.f.s fa0, v9
 ; CHECK-NEXT:    ret
   %r = call half @llvm.vp.reduce.fadd.nxv2f16(half %s, <vscale x 2 x half> %v, <vscale x 2 x i1> %m, i32 %evl)
   ret half %r
 }
 
 declare half @llvm.vp.reduce.fadd.nxv4f16(half, <vscale x 4 x half>, <vscale x 4 x i1>, i32)
 
 define half @vpreduce_fadd_nxv4f16(half %s, <vscale x 4 x half> %v, <vscale x 4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_fadd_nxv4f16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vfmv.s.f v9, fa0
-; CHECK-NEXT:    vsetvli zero, a0, e16, m1, tu, ma
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, ma
+; CHECK-NEXT:    vfmv.v.f v9, fa0
+; CHECK-NEXT:    vsetvli zero, zero, e16, m1, tu, ma
 ; CHECK-NEXT:    vfredusum.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vfmv.f.s fa0, v9
 ; CHECK-NEXT:    ret
   %r = call reassoc half @llvm.vp.reduce.fadd.nxv4f16(half %s, <vscale x 4 x half> %v, <vscale x 4 x i1> %m, i32 %evl)
   ret half %r
 }
 
 define half @vpreduce_ord_fadd_nxv4f16(half %s, <vscale x 4 x half> %v, <vscale x 4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_ord_fadd_nxv4f16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vfmv.s.f v9, fa0
-; CHECK-NEXT:    vsetvli zero, a0, e16, m1, tu, ma
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, ma
+; CHECK-NEXT:    vfmv.v.f v9, fa0
+; CHECK-NEXT:    vsetvli zero, zero, e16, m1, tu, ma
 ; CHECK-NEXT:    vfredosum.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vfmv.f.s fa0, v9
 ; CHECK-NEXT:    ret
   %r = call half @llvm.vp.reduce.fadd.nxv4f16(half %s, <vscale x 4 x half> %v, <vscale x 4 x i1> %m, i32 %evl)
   ret half %r
 }
 
 declare half @llvm.vp.reduce.fadd.nxv64f16(half, <vscale x 64 x half>, <vscale x 64 x i1>, i32)
; CHECK-NEXT:    ret
   ret half %r
 }
 
 declare float @llvm.vp.reduce.fadd.nxv1f32(float, <vscale x 1 x float>, <vscale x 1 x i1>, i32)
 
 define float @vpreduce_fadd_nxv1f32(float %s, <vscale x 1 x float> %v, <vscale x 1 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_fadd_nxv1f32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vfmv.s.f v9, fa0
-; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, tu, ma
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    vfmv.v.f v9, fa0
+; CHECK-NEXT:    vsetvli zero, zero, e32, mf2, tu, ma
 ; CHECK-NEXT:    vfredusum.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vfmv.f.s fa0, v9
 ; CHECK-NEXT:    ret
   %r = call reassoc float @llvm.vp.reduce.fadd.nxv1f32(float %s, <vscale x 1 x float> %v, <vscale x 1 x i1> %m, i32 %evl)
   ret float %r
 }
 
 define float @vpreduce_ord_fadd_nxv1f32(float %s, <vscale x 1 x float> %v, <vscale x 1 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_ord_fadd_nxv1f32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vfmv.s.f v9, fa0
-; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, tu, ma
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, ma
+; CHECK-NEXT:    vfmv.v.f v9, fa0
+; CHECK-NEXT:    vsetvli zero, zero, e32, mf2, tu, ma
 ; CHECK-NEXT:    vfredosum.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vfmv.f.s fa0, v9
 ; CHECK-NEXT:    ret
   %r = call float @llvm.vp.reduce.fadd.nxv1f32(float %s, <vscale x 1 x float> %v, <vscale x 1 x i1> %m, i32 %evl)
   ret float %r
 }
 
 declare float @llvm.vp.reduce.fadd.nxv2f32(float, <vscale x 2 x float>, <vscale x 2 x i1>, i32)
 
 define float @vpreduce_fadd_nxv2f32(float %s, <vscale x 2 x float> %v, <vscale x 2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_fadd_nxv2f32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vfmv.s.f v9, fa0
-; CHECK-NEXT:    vsetvli zero, a0, e32, m1, tu, ma
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, ma
+; CHECK-NEXT:    vfmv.v.f v9, fa0
+; CHECK-NEXT:    vsetvli zero, zero, e32, m1, tu, ma
 ; CHECK-NEXT:    vfredusum.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vfmv.f.s fa0, v9
 ; CHECK-NEXT:    ret
   %r = call reassoc float @llvm.vp.reduce.fadd.nxv2f32(float %s, <vscale x 2 x float> %v, <vscale x 2 x i1> %m, i32 %evl)
   ret float %r
 }
 
 define float @vpreduce_ord_fadd_nxv2f32(float %s, <vscale x 2 x float> %v, <vscale x 2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_ord_fadd_nxv2f32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vfmv.s.f v9, fa0
-; CHECK-NEXT:    vsetvli zero, a0, e32, m1, tu, ma
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, ma
+; CHECK-NEXT:    vfmv.v.f v9, fa0
+; CHECK-NEXT:    vsetvli zero, zero, e32, m1, tu, ma
 ; CHECK-NEXT:    vfredosum.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vfmv.f.s fa0, v9
 ; CHECK-NEXT:    ret
   %r = call float @llvm.vp.reduce.fadd.nxv2f32(float %s, <vscale x 2 x float> %v, <vscale x 2 x i1> %m, i32 %evl)
   ret float %r
 }
 
 declare float @llvm.vp.reduce.fadd.nxv4f32(float, <vscale x 4 x float>, <vscale x 4 x i1>, i32)
; CHECK-NEXT:    ret
   ret float %r
 }
 
 declare double @llvm.vp.reduce.fadd.nxv1f64(double, <vscale x 1 x double>, <vscale x 1 x i1>, i32)
 
 define double @vpreduce_fadd_nxv1f64(double %s, <vscale x 1 x double> %v, <vscale x 1 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_fadd_nxv1f64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
-; CHECK-NEXT:    vfmv.s.f v9, fa0
-; CHECK-NEXT:    vsetvli zero, a0, e64, m1, tu, ma
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
+; CHECK-NEXT:    vfmv.v.f v9, fa0
+; CHECK-NEXT:    vsetvli zero, zero, e64, m1, tu, ma
 ; CHECK-NEXT:    vfredusum.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vfmv.f.s fa0, v9
 ; CHECK-NEXT:    ret
   %r = call reassoc double @llvm.vp.reduce.fadd.nxv1f64(double %s, <vscale x 1 x double> %v, <vscale x 1 x i1> %m, i32 %evl)
   ret double %r
 }
 
 define double @vpreduce_ord_fadd_nxv1f64(double %s, <vscale x 1 x double> %v, <vscale x 1 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_ord_fadd_nxv1f64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
-; CHECK-NEXT:    vfmv.s.f v9, fa0
-; CHECK-NEXT:    vsetvli zero, a0, e64, m1, tu, ma
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
+; CHECK-NEXT:    vfmv.v.f v9, fa0
+; CHECK-NEXT:    vsetvli zero, zero, e64, m1, tu, ma
 ; CHECK-NEXT:    vfredosum.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vfmv.f.s fa0, v9
 ; CHECK-NEXT:    ret
   %r = call double @llvm.vp.reduce.fadd.nxv1f64(double %s, <vscale x 1 x double> %v, <vscale x 1 x i1> %m, i32 %evl)
   ret double %r
 }
 
 declare double @llvm.vp.reduce.fadd.nxv2f64(double, <vscale x 2 x double>, <vscale x 2 x i1>, i32)

llvm/test/CodeGen/RISCV/rvv/vreductions-int-vp.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc -mtriple=riscv32 -mattr=+v -verify-machineinstrs < %s \
 ; RUN:   | FileCheck %s --check-prefixes=CHECK,RV32
 ; RUN: llc -mtriple=riscv64 -mattr=+v -verify-machineinstrs < %s \
 ; RUN:   | FileCheck %s --check-prefixes=CHECK,RV64
 
 declare i8 @llvm.vp.reduce.add.nxv1i8(i8, <vscale x 1 x i8>, <vscale x 1 x i1>, i32)
 
 define signext i8 @vpreduce_add_nxv1i8(i8 signext %s, <vscale x 1 x i8> %v, <vscale x 1 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_add_nxv1i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf8, tu, ma
 ; CHECK-NEXT:    vredsum.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.add.nxv1i8(i8 %s, <vscale x 1 x i8> %v, <vscale x 1 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i8 @llvm.vp.reduce.umax.nxv1i8(i8, <vscale x 1 x i8>, <vscale x 1 x i1>, i32)
 
 define signext i8 @vpreduce_umax_nxv1i8(i8 signext %s, <vscale x 1 x i8> %v, <vscale x 1 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_umax_nxv1i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    andi a0, a0, 255
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf8, tu, ma
 ; CHECK-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.umax.nxv1i8(i8 %s, <vscale x 1 x i8> %v, <vscale x 1 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i8 @llvm.vp.reduce.smax.nxv1i8(i8, <vscale x 1 x i8>, <vscale x 1 x i1>, i32)
 
 define signext i8 @vpreduce_smax_nxv1i8(i8 signext %s, <vscale x 1 x i8> %v, <vscale x 1 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_smax_nxv1i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf8, tu, ma
 ; CHECK-NEXT:    vredmax.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.smax.nxv1i8(i8 %s, <vscale x 1 x i8> %v, <vscale x 1 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i8 @llvm.vp.reduce.umin.nxv1i8(i8, <vscale x 1 x i8>, <vscale x 1 x i1>, i32)
 
 define signext i8 @vpreduce_umin_nxv1i8(i8 signext %s, <vscale x 1 x i8> %v, <vscale x 1 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_umin_nxv1i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    andi a0, a0, 255
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf8, tu, ma
 ; CHECK-NEXT:    vredminu.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.umin.nxv1i8(i8 %s, <vscale x 1 x i8> %v, <vscale x 1 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i8 @llvm.vp.reduce.smin.nxv1i8(i8, <vscale x 1 x i8>, <vscale x 1 x i1>, i32)
 
 define signext i8 @vpreduce_smin_nxv1i8(i8 signext %s, <vscale x 1 x i8> %v, <vscale x 1 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_smin_nxv1i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf8, tu, ma
 ; CHECK-NEXT:    vredmin.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.smin.nxv1i8(i8 %s, <vscale x 1 x i8> %v, <vscale x 1 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i8 @llvm.vp.reduce.and.nxv1i8(i8, <vscale x 1 x i8>, <vscale x 1 x i1>, i32)
 
 define signext i8 @vpreduce_and_nxv1i8(i8 signext %s, <vscale x 1 x i8> %v, <vscale x 1 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_and_nxv1i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf8, tu, ma
 ; CHECK-NEXT:    vredand.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.and.nxv1i8(i8 %s, <vscale x 1 x i8> %v, <vscale x 1 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i8 @llvm.vp.reduce.or.nxv1i8(i8, <vscale x 1 x i8>, <vscale x 1 x i1>, i32)
 
 define signext i8 @vpreduce_or_nxv1i8(i8 signext %s, <vscale x 1 x i8> %v, <vscale x 1 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_or_nxv1i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf8, tu, ma
 ; CHECK-NEXT:    vredor.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.or.nxv1i8(i8 %s, <vscale x 1 x i8> %v, <vscale x 1 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i8 @llvm.vp.reduce.xor.nxv1i8(i8, <vscale x 1 x i8>, <vscale x 1 x i1>, i32)
 
 define signext i8 @vpreduce_xor_nxv1i8(i8 signext %s, <vscale x 1 x i8> %v, <vscale x 1 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_xor_nxv1i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf8, tu, ma
 ; CHECK-NEXT:    vredxor.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.xor.nxv1i8(i8 %s, <vscale x 1 x i8> %v, <vscale x 1 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i8 @llvm.vp.reduce.add.nxv2i8(i8, <vscale x 2 x i8>, <vscale x 2 x i1>, i32)
 
 define signext i8 @vpreduce_add_nxv2i8(i8 signext %s, <vscale x 2 x i8> %v, <vscale x 2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_add_nxv2i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf4, tu, ma
 ; CHECK-NEXT:    vredsum.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.add.nxv2i8(i8 %s, <vscale x 2 x i8> %v, <vscale x 2 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i8 @llvm.vp.reduce.umax.nxv2i8(i8, <vscale x 2 x i8>, <vscale x 2 x i1>, i32)
 
 define signext i8 @vpreduce_umax_nxv2i8(i8 signext %s, <vscale x 2 x i8> %v, <vscale x 2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_umax_nxv2i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    andi a0, a0, 255
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf4, tu, ma
 ; CHECK-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.umax.nxv2i8(i8 %s, <vscale x 2 x i8> %v, <vscale x 2 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i8 @llvm.vp.reduce.smax.nxv2i8(i8, <vscale x 2 x i8>, <vscale x 2 x i1>, i32)
 
 define signext i8 @vpreduce_smax_nxv2i8(i8 signext %s, <vscale x 2 x i8> %v, <vscale x 2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_smax_nxv2i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf4, tu, ma
 ; CHECK-NEXT:    vredmax.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.smax.nxv2i8(i8 %s, <vscale x 2 x i8> %v, <vscale x 2 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i8 @llvm.vp.reduce.umin.nxv2i8(i8, <vscale x 2 x i8>, <vscale x 2 x i1>, i32)
 
 define signext i8 @vpreduce_umin_nxv2i8(i8 signext %s, <vscale x 2 x i8> %v, <vscale x 2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_umin_nxv2i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    andi a0, a0, 255
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf4, tu, ma
 ; CHECK-NEXT:    vredminu.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.umin.nxv2i8(i8 %s, <vscale x 2 x i8> %v, <vscale x 2 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i8 @llvm.vp.reduce.smin.nxv2i8(i8, <vscale x 2 x i8>, <vscale x 2 x i1>, i32)
 
 define signext i8 @vpreduce_smin_nxv2i8(i8 signext %s, <vscale x 2 x i8> %v, <vscale x 2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_smin_nxv2i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf4, tu, ma
 ; CHECK-NEXT:    vredmin.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.smin.nxv2i8(i8 %s, <vscale x 2 x i8> %v, <vscale x 2 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i8 @llvm.vp.reduce.and.nxv2i8(i8, <vscale x 2 x i8>, <vscale x 2 x i1>, i32)
 
 define signext i8 @vpreduce_and_nxv2i8(i8 signext %s, <vscale x 2 x i8> %v, <vscale x 2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_and_nxv2i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf4, tu, ma
 ; CHECK-NEXT:    vredand.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.and.nxv2i8(i8 %s, <vscale x 2 x i8> %v, <vscale x 2 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i8 @llvm.vp.reduce.or.nxv2i8(i8, <vscale x 2 x i8>, <vscale x 2 x i1>, i32)
 
 define signext i8 @vpreduce_or_nxv2i8(i8 signext %s, <vscale x 2 x i8> %v, <vscale x 2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_or_nxv2i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf4, tu, ma
 ; CHECK-NEXT:    vredor.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.or.nxv2i8(i8 %s, <vscale x 2 x i8> %v, <vscale x 2 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i8 @llvm.vp.reduce.xor.nxv2i8(i8, <vscale x 2 x i8>, <vscale x 2 x i1>, i32)
 
 define signext i8 @vpreduce_xor_nxv2i8(i8 signext %s, <vscale x 2 x i8> %v, <vscale x 2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_xor_nxv2i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf4, tu, ma
 ; CHECK-NEXT:    vredxor.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.xor.nxv2i8(i8 %s, <vscale x 2 x i8> %v, <vscale x 2 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i8 @llvm.vp.reduce.smax.nxv3i8(i8, <vscale x 3 x i8>, <vscale x 3 x i1>, i32)
 
 define signext i8 @vpreduce_smax_nxv3i8(i8 signext %s, <vscale x 3 x i8> %v, <vscale x 3 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_smax_nxv3i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf2, tu, ma
 ; CHECK-NEXT:    vredmax.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.smax.nxv3i8(i8 %s, <vscale x 3 x i8> %v, <vscale x 3 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i8 @llvm.vp.reduce.add.nxv4i8(i8, <vscale x 4 x i8>, <vscale x 4 x i1>, i32)
 
 define signext i8 @vpreduce_add_nxv4i8(i8 signext %s, <vscale x 4 x i8> %v, <vscale x 4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_add_nxv4i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf2, tu, ma
 ; CHECK-NEXT:    vredsum.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.add.nxv4i8(i8 %s, <vscale x 4 x i8> %v, <vscale x 4 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i8 @llvm.vp.reduce.umax.nxv4i8(i8, <vscale x 4 x i8>, <vscale x 4 x i1>, i32)
 
 define signext i8 @vpreduce_umax_nxv4i8(i8 signext %s, <vscale x 4 x i8> %v, <vscale x 4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_umax_nxv4i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    andi a0, a0, 255
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf2, tu, ma
 ; CHECK-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.umax.nxv4i8(i8 %s, <vscale x 4 x i8> %v, <vscale x 4 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i8 @llvm.vp.reduce.smax.nxv4i8(i8, <vscale x 4 x i8>, <vscale x 4 x i1>, i32)
 
 define signext i8 @vpreduce_smax_nxv4i8(i8 signext %s, <vscale x 4 x i8> %v, <vscale x 4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_smax_nxv4i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf2, tu, ma
 ; CHECK-NEXT:    vredmax.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.smax.nxv4i8(i8 %s, <vscale x 4 x i8> %v, <vscale x 4 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i8 @llvm.vp.reduce.umin.nxv4i8(i8, <vscale x 4 x i8>, <vscale x 4 x i1>, i32)
 
 define signext i8 @vpreduce_umin_nxv4i8(i8 signext %s, <vscale x 4 x i8> %v, <vscale x 4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_umin_nxv4i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    andi a0, a0, 255
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf2, tu, ma
 ; CHECK-NEXT:    vredminu.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.umin.nxv4i8(i8 %s, <vscale x 4 x i8> %v, <vscale x 4 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i8 @llvm.vp.reduce.smin.nxv4i8(i8, <vscale x 4 x i8>, <vscale x 4 x i1>, i32)
 
 define signext i8 @vpreduce_smin_nxv4i8(i8 signext %s, <vscale x 4 x i8> %v, <vscale x 4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_smin_nxv4i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf2, tu, ma
 ; CHECK-NEXT:    vredmin.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.smin.nxv4i8(i8 %s, <vscale x 4 x i8> %v, <vscale x 4 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i8 @llvm.vp.reduce.and.nxv4i8(i8, <vscale x 4 x i8>, <vscale x 4 x i1>, i32)
 
 define signext i8 @vpreduce_and_nxv4i8(i8 signext %s, <vscale x 4 x i8> %v, <vscale x 4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_and_nxv4i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf2, tu, ma
 ; CHECK-NEXT:    vredand.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.and.nxv4i8(i8 %s, <vscale x 4 x i8> %v, <vscale x 4 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i8 @llvm.vp.reduce.or.nxv4i8(i8, <vscale x 4 x i8>, <vscale x 4 x i1>, i32)
 
 define signext i8 @vpreduce_or_nxv4i8(i8 signext %s, <vscale x 4 x i8> %v, <vscale x 4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_or_nxv4i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf2, tu, ma
 ; CHECK-NEXT:    vredor.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.or.nxv4i8(i8 %s, <vscale x 4 x i8> %v, <vscale x 4 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i8 @llvm.vp.reduce.xor.nxv4i8(i8, <vscale x 4 x i8>, <vscale x 4 x i1>, i32)
 
 define signext i8 @vpreduce_xor_nxv4i8(i8 signext %s, <vscale x 4 x i8> %v, <vscale x 4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_xor_nxv4i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf2, tu, ma
 ; CHECK-NEXT:    vredxor.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i8 @llvm.vp.reduce.xor.nxv4i8(i8 %s, <vscale x 4 x i8> %v, <vscale x 4 x i1> %m, i32 %evl)
   ret i8 %r
 }
 
 declare i16 @llvm.vp.reduce.add.nxv1i16(i16, <vscale x 1 x i16>, <vscale x 1 x i1>, i32)
 
 define signext i16 @vpreduce_add_nxv1i16(i16 signext %s, <vscale x 1 x i16> %v, <vscale x 1 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_add_nxv1i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e16, mf4, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e16, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf4, tu, ma
 ; CHECK-NEXT:    vredsum.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i16 @llvm.vp.reduce.add.nxv1i16(i16 %s, <vscale x 1 x i16> %v, <vscale x 1 x i1> %m, i32 %evl)
   ret i16 %r
 }
 
 declare i16 @llvm.vp.reduce.umax.nxv1i16(i16, <vscale x 1 x i16>, <vscale x 1 x i1>, i32)
 
 define signext i16 @vpreduce_umax_nxv1i16(i16 signext %s, <vscale x 1 x i16> %v, <vscale x 1 x i1> %m, i32 zeroext %evl) {
-; RV32-LABEL: vpreduce_umax_nxv1i16:
-; RV32:       # %bb.0:
-; RV32-NEXT:    slli a0, a0, 16
-; RV32-NEXT:    srli a0, a0, 16
-; RV32-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; RV32-NEXT:    vmv.s.x v9, a0
-; RV32-NEXT:    vsetvli zero, a1, e16, mf4, tu, ma
-; RV32-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
-; RV32-NEXT:    vmv.x.s a0, v9
-; RV32-NEXT:    ret
-;
-; RV64-LABEL: vpreduce_umax_nxv1i16:
-; RV64:       # %bb.0:
-; RV64-NEXT:    slli a0, a0, 48
-; RV64-NEXT:    srli a0, a0, 48
-; RV64-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli zero, a1, e16, mf4, tu, ma
-; RV64-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
-; RV64-NEXT:    vmv.x.s a0, v9
-; RV64-NEXT:    ret
+; CHECK-LABEL: vpreduce_umax_nxv1i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e16, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf4, tu, ma
+; CHECK-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
+; CHECK-NEXT:    vmv.x.s a0, v9
+; CHECK-NEXT:    ret
   %r = call i16 @llvm.vp.reduce.umax.nxv1i16(i16 %s, <vscale x 1 x i16> %v, <vscale x 1 x i1> %m, i32 %evl)
   ret i16 %r
 }
 
 declare i16 @llvm.vp.reduce.smax.nxv1i16(i16, <vscale x 1 x i16>, <vscale x 1 x i1>, i32)
 
 define signext i16 @vpreduce_smax_nxv1i16(i16 signext %s, <vscale x 1 x i16> %v, <vscale x 1 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_smax_nxv1i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e16, mf4, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e16, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf4, tu, ma
 ; CHECK-NEXT:    vredmax.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i16 @llvm.vp.reduce.smax.nxv1i16(i16 %s, <vscale x 1 x i16> %v, <vscale x 1 x i1> %m, i32 %evl)
   ret i16 %r
 }
 
 declare i16 @llvm.vp.reduce.umin.nxv1i16(i16, <vscale x 1 x i16>, <vscale x 1 x i1>, i32)
 
 define signext i16 @vpreduce_umin_nxv1i16(i16 signext %s, <vscale x 1 x i16> %v, <vscale x 1 x i1> %m, i32 zeroext %evl) {
-; RV32-LABEL: vpreduce_umin_nxv1i16:
-; RV32:       # %bb.0:
-; RV32-NEXT:    slli a0, a0, 16
-; RV32-NEXT:    srli a0, a0, 16
-; RV32-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; RV32-NEXT:    vmv.s.x v9, a0
-; RV32-NEXT:    vsetvli zero, a1, e16, mf4, tu, ma
-; RV32-NEXT:    vredminu.vs v9, v8, v9, v0.t
-; RV32-NEXT:    vmv.x.s a0, v9
-; RV32-NEXT:    ret
-;
-; RV64-LABEL: vpreduce_umin_nxv1i16:
-; RV64:       # %bb.0:
-; RV64-NEXT:    slli a0, a0, 48
-; RV64-NEXT:    srli a0, a0, 48
-; RV64-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli zero, a1, e16, mf4, tu, ma
-; RV64-NEXT:    vredminu.vs v9, v8, v9, v0.t
-; RV64-NEXT:    vmv.x.s a0, v9
-; RV64-NEXT:    ret
+; CHECK-LABEL: vpreduce_umin_nxv1i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e16, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf4, tu, ma
+; CHECK-NEXT:    vredminu.vs v9, v8, v9, v0.t
+; CHECK-NEXT:    vmv.x.s a0, v9
+; CHECK-NEXT:    ret
   %r = call i16 @llvm.vp.reduce.umin.nxv1i16(i16 %s, <vscale x 1 x i16> %v, <vscale x 1 x i1> %m, i32 %evl)
   ret i16 %r
 }
 
 declare i16 @llvm.vp.reduce.smin.nxv1i16(i16, <vscale x 1 x i16>, <vscale x 1 x i1>, i32)
 
 define signext i16 @vpreduce_smin_nxv1i16(i16 signext %s, <vscale x 1 x i16> %v, <vscale x 1 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_smin_nxv1i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e16, mf4, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e16, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf4, tu, ma
 ; CHECK-NEXT:    vredmin.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i16 @llvm.vp.reduce.smin.nxv1i16(i16 %s, <vscale x 1 x i16> %v, <vscale x 1 x i1> %m, i32 %evl)
   ret i16 %r
 }
 
 declare i16 @llvm.vp.reduce.and.nxv1i16(i16, <vscale x 1 x i16>, <vscale x 1 x i1>, i32)
 
 define signext i16 @vpreduce_and_nxv1i16(i16 signext %s, <vscale x 1 x i16> %v, <vscale x 1 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_and_nxv1i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e16, mf4, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e16, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf4, tu, ma
 ; CHECK-NEXT:    vredand.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i16 @llvm.vp.reduce.and.nxv1i16(i16 %s, <vscale x 1 x i16> %v, <vscale x 1 x i1> %m, i32 %evl)
   ret i16 %r
 }
 
 declare i16 @llvm.vp.reduce.or.nxv1i16(i16, <vscale x 1 x i16>, <vscale x 1 x i1>, i32)
 
 define signext i16 @vpreduce_or_nxv1i16(i16 signext %s, <vscale x 1 x i16> %v, <vscale x 1 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_or_nxv1i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e16, mf4, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e16, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf4, tu, ma
 ; CHECK-NEXT:    vredor.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i16 @llvm.vp.reduce.or.nxv1i16(i16 %s, <vscale x 1 x i16> %v, <vscale x 1 x i1> %m, i32 %evl)
   ret i16 %r
 }
 
 declare i16 @llvm.vp.reduce.xor.nxv1i16(i16, <vscale x 1 x i16>, <vscale x 1 x i1>, i32)
 
 define signext i16 @vpreduce_xor_nxv1i16(i16 signext %s, <vscale x 1 x i16> %v, <vscale x 1 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_xor_nxv1i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e16, mf4, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e16, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf4, tu, ma
 ; CHECK-NEXT:    vredxor.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i16 @llvm.vp.reduce.xor.nxv1i16(i16 %s, <vscale x 1 x i16> %v, <vscale x 1 x i1> %m, i32 %evl)
   ret i16 %r
 }
 
 declare i16 @llvm.vp.reduce.add.nxv2i16(i16, <vscale x 2 x i16>, <vscale x 2 x i1>, i32)
 
 define signext i16 @vpreduce_add_nxv2i16(i16 signext %s, <vscale x 2 x i16> %v, <vscale x 2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_add_nxv2i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e16, mf2, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e16, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf2, tu, ma
 ; CHECK-NEXT:    vredsum.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i16 @llvm.vp.reduce.add.nxv2i16(i16 %s, <vscale x 2 x i16> %v, <vscale x 2 x i1> %m, i32 %evl)
   ret i16 %r
 }
 
 declare i16 @llvm.vp.reduce.umax.nxv2i16(i16, <vscale x 2 x i16>, <vscale x 2 x i1>, i32)
 
 define signext i16 @vpreduce_umax_nxv2i16(i16 signext %s, <vscale x 2 x i16> %v, <vscale x 2 x i1> %m, i32 zeroext %evl) {
-; RV32-LABEL: vpreduce_umax_nxv2i16:
-; RV32:       # %bb.0:
-; RV32-NEXT:    slli a0, a0, 16
-; RV32-NEXT:    srli a0, a0, 16
-; RV32-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; RV32-NEXT:    vmv.s.x v9, a0
-; RV32-NEXT:    vsetvli zero, a1, e16, mf2, tu, ma
-; RV32-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
-; RV32-NEXT:    vmv.x.s a0, v9
-; RV32-NEXT:    ret
-;
-; RV64-LABEL: vpreduce_umax_nxv2i16:
-; RV64:       # %bb.0:
-; RV64-NEXT:    slli a0, a0, 48
-; RV64-NEXT:    srli a0, a0, 48
-; RV64-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli zero, a1, e16, mf2, tu, ma
-; RV64-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
-; RV64-NEXT:    vmv.x.s a0, v9
-; RV64-NEXT:    ret
+; CHECK-LABEL: vpreduce_umax_nxv2i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e16, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf2, tu, ma
+; CHECK-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
+; CHECK-NEXT:    vmv.x.s a0, v9
+; CHECK-NEXT:    ret
   %r = call i16 @llvm.vp.reduce.umax.nxv2i16(i16 %s, <vscale x 2 x i16> %v, <vscale x 2 x i1> %m, i32 %evl)
   ret i16 %r
 }
 
 declare i16 @llvm.vp.reduce.smax.nxv2i16(i16, <vscale x 2 x i16>, <vscale x 2 x i1>, i32)
 
 define signext i16 @vpreduce_smax_nxv2i16(i16 signext %s, <vscale x 2 x i16> %v, <vscale x 2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_smax_nxv2i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e16, mf2, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e16, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf2, tu, ma
 ; CHECK-NEXT:    vredmax.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i16 @llvm.vp.reduce.smax.nxv2i16(i16 %s, <vscale x 2 x i16> %v, <vscale x 2 x i1> %m, i32 %evl)
   ret i16 %r
 }
 
 declare i16 @llvm.vp.reduce.umin.nxv2i16(i16, <vscale x 2 x i16>, <vscale x 2 x i1>, i32)
 
 define signext i16 @vpreduce_umin_nxv2i16(i16 signext %s, <vscale x 2 x i16> %v, <vscale x 2 x i1> %m, i32 zeroext %evl) {
-; RV32-LABEL: vpreduce_umin_nxv2i16:
-; RV32:       # %bb.0:
-; RV32-NEXT:    slli a0, a0, 16
-; RV32-NEXT:    srli a0, a0, 16
-; RV32-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; RV32-NEXT:    vmv.s.x v9, a0
-; RV32-NEXT:    vsetvli zero, a1, e16, mf2, tu, ma
-; RV32-NEXT:    vredminu.vs v9, v8, v9, v0.t
-; RV32-NEXT:    vmv.x.s a0, v9
-; RV32-NEXT:    ret
-;
-; RV64-LABEL: vpreduce_umin_nxv2i16:
-; RV64:       # %bb.0:
-; RV64-NEXT:    slli a0, a0, 48
-; RV64-NEXT:    srli a0, a0, 48
-; RV64-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli zero, a1, e16, mf2, tu, ma
-; RV64-NEXT:    vredminu.vs v9, v8, v9, v0.t
-; RV64-NEXT:    vmv.x.s a0, v9
-; RV64-NEXT:    ret
+; CHECK-LABEL: vpreduce_umin_nxv2i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e16, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf2, tu, ma
+; CHECK-NEXT:    vredminu.vs v9, v8, v9, v0.t
+; CHECK-NEXT:    vmv.x.s a0, v9
+; CHECK-NEXT:    ret
   %r = call i16 @llvm.vp.reduce.umin.nxv2i16(i16 %s, <vscale x 2 x i16> %v, <vscale x 2 x i1> %m, i32 %evl)
   ret i16 %r
 }
 
 declare i16 @llvm.vp.reduce.smin.nxv2i16(i16, <vscale x 2 x i16>, <vscale x 2 x i1>, i32)
 
 define signext i16 @vpreduce_smin_nxv2i16(i16 signext %s, <vscale x 2 x i16> %v, <vscale x 2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_smin_nxv2i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e16, mf2, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e16, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf2, tu, ma
 ; CHECK-NEXT:    vredmin.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i16 @llvm.vp.reduce.smin.nxv2i16(i16 %s, <vscale x 2 x i16> %v, <vscale x 2 x i1> %m, i32 %evl)
   ret i16 %r
 }
 
 declare i16 @llvm.vp.reduce.and.nxv2i16(i16, <vscale x 2 x i16>, <vscale x 2 x i1>, i32)
 
 define signext i16 @vpreduce_and_nxv2i16(i16 signext %s, <vscale x 2 x i16> %v, <vscale x 2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_and_nxv2i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e16, mf2, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e16, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf2, tu, ma
 ; CHECK-NEXT:    vredand.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i16 @llvm.vp.reduce.and.nxv2i16(i16 %s, <vscale x 2 x i16> %v, <vscale x 2 x i1> %m, i32 %evl)
   ret i16 %r
 }
 
 declare i16 @llvm.vp.reduce.or.nxv2i16(i16, <vscale x 2 x i16>, <vscale x 2 x i1>, i32)
 
 define signext i16 @vpreduce_or_nxv2i16(i16 signext %s, <vscale x 2 x i16> %v, <vscale x 2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_or_nxv2i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e16, mf2, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e16, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf2, tu, ma
 ; CHECK-NEXT:    vredor.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i16 @llvm.vp.reduce.or.nxv2i16(i16 %s, <vscale x 2 x i16> %v, <vscale x 2 x i1> %m, i32 %evl)
   ret i16 %r
 }
 
 declare i16 @llvm.vp.reduce.xor.nxv2i16(i16, <vscale x 2 x i16>, <vscale x 2 x i1>, i32)
 
 define signext i16 @vpreduce_xor_nxv2i16(i16 signext %s, <vscale x 2 x i16> %v, <vscale x 2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_xor_nxv2i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e16, mf2, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e16, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf2, tu, ma
 ; CHECK-NEXT:    vredxor.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i16 @llvm.vp.reduce.xor.nxv2i16(i16 %s, <vscale x 2 x i16> %v, <vscale x 2 x i1> %m, i32 %evl)
   ret i16 %r
 }
 
 declare i16 @llvm.vp.reduce.add.nxv4i16(i16, <vscale x 4 x i16>, <vscale x 4 x i1>, i32)
 
 define signext i16 @vpreduce_add_nxv4i16(i16 signext %s, <vscale x 4 x i16> %v, <vscale x 4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_add_nxv4i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e16, m1, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e16, m1, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e16, m1, tu, ma
 ; CHECK-NEXT:    vredsum.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i16 @llvm.vp.reduce.add.nxv4i16(i16 %s, <vscale x 4 x i16> %v, <vscale x 4 x i1> %m, i32 %evl)
   ret i16 %r
 }
 
 declare i16 @llvm.vp.reduce.umax.nxv4i16(i16, <vscale x 4 x i16>, <vscale x 4 x i1>, i32)
 
 define signext i16 @vpreduce_umax_nxv4i16(i16 signext %s, <vscale x 4 x i16> %v, <vscale x 4 x i1> %m, i32 zeroext %evl) {
-; RV32-LABEL: vpreduce_umax_nxv4i16:
-; RV32:       # %bb.0:
-; RV32-NEXT:    slli a0, a0, 16
-; RV32-NEXT:    srli a0, a0, 16
-; RV32-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; RV32-NEXT:    vmv.s.x v9, a0
-; RV32-NEXT:    vsetvli zero, a1, e16, m1, tu, ma
-; RV32-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
-; RV32-NEXT:    vmv.x.s a0, v9
-; RV32-NEXT:    ret
-;
-; RV64-LABEL: vpreduce_umax_nxv4i16:
-; RV64:       # %bb.0:
-; RV64-NEXT:    slli a0, a0, 48
-; RV64-NEXT:    srli a0, a0, 48
-; RV64-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli zero, a1, e16, m1, tu, ma
-; RV64-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
-; RV64-NEXT:    vmv.x.s a0, v9
-; RV64-NEXT:    ret
+; CHECK-LABEL: vpreduce_umax_nxv4i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e16, m1, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e16, m1, tu, ma
+; CHECK-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
+; CHECK-NEXT:    vmv.x.s a0, v9
+; CHECK-NEXT:    ret
   %r = call i16 @llvm.vp.reduce.umax.nxv4i16(i16 %s, <vscale x 4 x i16> %v, <vscale x 4 x i1> %m, i32 %evl)
   ret i16 %r
 }
 
 declare i16 @llvm.vp.reduce.smax.nxv4i16(i16, <vscale x 4 x i16>, <vscale x 4 x i1>, i32)
 
 define signext i16 @vpreduce_smax_nxv4i16(i16 signext %s, <vscale x 4 x i16> %v, <vscale x 4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_smax_nxv4i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e16, m1, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e16, m1, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e16, m1, tu, ma
 ; CHECK-NEXT:    vredmax.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i16 @llvm.vp.reduce.smax.nxv4i16(i16 %s, <vscale x 4 x i16> %v, <vscale x 4 x i1> %m, i32 %evl)
   ret i16 %r
 }
 
 declare i16 @llvm.vp.reduce.umin.nxv4i16(i16, <vscale x 4 x i16>, <vscale x 4 x i1>, i32)
 
 define signext i16 @vpreduce_umin_nxv4i16(i16 signext %s, <vscale x 4 x i16> %v, <vscale x 4 x i1> %m, i32 zeroext %evl) {
-; RV32-LABEL: vpreduce_umin_nxv4i16:
-; RV32:       # %bb.0:
-; RV32-NEXT:    slli a0, a0, 16
-; RV32-NEXT:    srli a0, a0, 16
-; RV32-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; RV32-NEXT:    vmv.s.x v9, a0
-; RV32-NEXT:    vsetvli zero, a1, e16, m1, tu, ma
-; RV32-NEXT:    vredminu.vs v9, v8, v9, v0.t
-; RV32-NEXT:    vmv.x.s a0, v9
-; RV32-NEXT:    ret
-;
-; RV64-LABEL: vpreduce_umin_nxv4i16:
-; RV64:       # %bb.0:
-; RV64-NEXT:    slli a0, a0, 48
-; RV64-NEXT:    srli a0, a0, 48
-; RV64-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli zero, a1, e16, m1, tu, ma
-; RV64-NEXT:    vredminu.vs v9, v8, v9, v0.t
-; RV64-NEXT:    vmv.x.s a0, v9
-; RV64-NEXT:    ret
+; CHECK-LABEL: vpreduce_umin_nxv4i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e16, m1, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e16, m1, tu, ma
+; CHECK-NEXT:    vredminu.vs v9, v8, v9, v0.t
+; CHECK-NEXT:    vmv.x.s a0, v9
+; CHECK-NEXT:    ret
   %r = call i16 @llvm.vp.reduce.umin.nxv4i16(i16 %s, <vscale x 4 x i16> %v, <vscale x 4 x i1> %m, i32 %evl)
   ret i16 %r
 }
 
 declare i16 @llvm.vp.reduce.smin.nxv4i16(i16, <vscale x 4 x i16>, <vscale x 4 x i1>, i32)
 
 define signext i16 @vpreduce_smin_nxv4i16(i16 signext %s, <vscale x 4 x i16> %v, <vscale x 4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_smin_nxv4i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e16, m1, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e16, m1, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e16, m1, tu, ma
 ; CHECK-NEXT:    vredmin.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i16 @llvm.vp.reduce.smin.nxv4i16(i16 %s, <vscale x 4 x i16> %v, <vscale x 4 x i1> %m, i32 %evl)
   ret i16 %r
 }
 
 declare i16 @llvm.vp.reduce.and.nxv4i16(i16, <vscale x 4 x i16>, <vscale x 4 x i1>, i32)
 
 define signext i16 @vpreduce_and_nxv4i16(i16 signext %s, <vscale x 4 x i16> %v, <vscale x 4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_and_nxv4i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e16, m1, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e16, m1, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e16, m1, tu, ma
 ; CHECK-NEXT:    vredand.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i16 @llvm.vp.reduce.and.nxv4i16(i16 %s, <vscale x 4 x i16> %v, <vscale x 4 x i1> %m, i32 %evl)
   ret i16 %r
 }
 
 declare i16 @llvm.vp.reduce.or.nxv4i16(i16, <vscale x 4 x i16>, <vscale x 4 x i1>, i32)
 
 define signext i16 @vpreduce_or_nxv4i16(i16 signext %s, <vscale x 4 x i16> %v, <vscale x 4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_or_nxv4i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e16, m1, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e16, m1, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e16, m1, tu, ma
 ; CHECK-NEXT:    vredor.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i16 @llvm.vp.reduce.or.nxv4i16(i16 %s, <vscale x 4 x i16> %v, <vscale x 4 x i1> %m, i32 %evl)
   ret i16 %r
 }
 
 declare i16 @llvm.vp.reduce.xor.nxv4i16(i16, <vscale x 4 x i16>, <vscale x 4 x i1>, i32)
 
 define signext i16 @vpreduce_xor_nxv4i16(i16 signext %s, <vscale x 4 x i16> %v, <vscale x 4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_xor_nxv4i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e16, m1, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e16, m1, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e16, m1, tu, ma
 ; CHECK-NEXT:    vredxor.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i16 @llvm.vp.reduce.xor.nxv4i16(i16 %s, <vscale x 4 x i16> %v, <vscale x 4 x i1> %m, i32 %evl)
   ret i16 %r
 }
 
 declare i32 @llvm.vp.reduce.add.nxv1i32(i32, <vscale x 1 x i32>, <vscale x 1 x i1>, i32)
 
 define signext i32 @vpreduce_add_nxv1i32(i32 signext %s, <vscale x 1 x i32> %v, <vscale x 1 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_add_nxv1i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e32, mf2, tu, ma
 ; CHECK-NEXT:    vredsum.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i32 @llvm.vp.reduce.add.nxv1i32(i32 %s, <vscale x 1 x i32> %v, <vscale x 1 x i1> %m, i32 %evl)
   ret i32 %r
 }
 
 declare i32 @llvm.vp.reduce.umax.nxv1i32(i32, <vscale x 1 x i32>, <vscale x 1 x i1>, i32)
 
 define signext i32 @vpreduce_umax_nxv1i32(i32 signext %s, <vscale x 1 x i32> %v, <vscale x 1 x i1> %m, i32 zeroext %evl) {
-; RV32-LABEL: vpreduce_umax_nxv1i32:
-; RV32:       # %bb.0:
-; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; RV32-NEXT:    vmv.s.x v9, a0
-; RV32-NEXT:    vsetvli zero, a1, e32, mf2, tu, ma
-; RV32-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
-; RV32-NEXT:    vmv.x.s a0, v9
-; RV32-NEXT:    ret
-;
-; RV64-LABEL: vpreduce_umax_nxv1i32:
-; RV64:       # %bb.0:
-; RV64-NEXT:    slli a0, a0, 32
-; RV64-NEXT:    srli a0, a0, 32
-; RV64-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli zero, a1, e32, mf2, tu, ma
-; RV64-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
-; RV64-NEXT:    vmv.x.s a0, v9
-; RV64-NEXT:    ret
+; CHECK-LABEL: vpreduce_umax_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e32, mf2, tu, ma
+; CHECK-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
+; CHECK-NEXT:    vmv.x.s a0, v9
+; CHECK-NEXT:    ret
   %r = call i32 @llvm.vp.reduce.umax.nxv1i32(i32 %s, <vscale x 1 x i32> %v, <vscale x 1 x i1> %m, i32 %evl)
   ret i32 %r
 }
 
 declare i32 @llvm.vp.reduce.smax.nxv1i32(i32, <vscale x 1 x i32>, <vscale x 1 x i1>, i32)
 
 define signext i32 @vpreduce_smax_nxv1i32(i32 signext %s, <vscale x 1 x i32> %v, <vscale x 1 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_smax_nxv1i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e32, mf2, tu, ma
 ; CHECK-NEXT:    vredmax.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i32 @llvm.vp.reduce.smax.nxv1i32(i32 %s, <vscale x 1 x i32> %v, <vscale x 1 x i1> %m, i32 %evl)
   ret i32 %r
 }
 
 declare i32 @llvm.vp.reduce.umin.nxv1i32(i32, <vscale x 1 x i32>, <vscale x 1 x i1>, i32)
 
 define signext i32 @vpreduce_umin_nxv1i32(i32 signext %s, <vscale x 1 x i32> %v, <vscale x 1 x i1> %m, i32 zeroext %evl) {
-; RV32-LABEL: vpreduce_umin_nxv1i32:
-; RV32:       # %bb.0:
-; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; RV32-NEXT:    vmv.s.x v9, a0
-; RV32-NEXT:    vsetvli zero, a1, e32, mf2, tu, ma
-; RV32-NEXT:    vredminu.vs v9, v8, v9, v0.t
-; RV32-NEXT:    vmv.x.s a0, v9
-; RV32-NEXT:    ret
-;
-; RV64-LABEL: vpreduce_umin_nxv1i32:
-; RV64:       # %bb.0:
-; RV64-NEXT:    slli a0, a0, 32
-; RV64-NEXT:    srli a0, a0, 32
-; RV64-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli zero, a1, e32, mf2, tu, ma
-; RV64-NEXT:    vredminu.vs v9, v8, v9, v0.t
-; RV64-NEXT:    vmv.x.s a0, v9
-; RV64-NEXT:    ret
+; CHECK-LABEL: vpreduce_umin_nxv1i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e32, mf2, tu, ma
+; CHECK-NEXT:    vredminu.vs v9, v8, v9, v0.t
+; CHECK-NEXT:    vmv.x.s a0, v9
+; CHECK-NEXT:    ret
   %r = call i32 @llvm.vp.reduce.umin.nxv1i32(i32 %s, <vscale x 1 x i32> %v, <vscale x 1 x i1> %m, i32 %evl)
   ret i32 %r
 }
 
 declare i32 @llvm.vp.reduce.smin.nxv1i32(i32, <vscale x 1 x i32>, <vscale x 1 x i1>, i32)
 
 define signext i32 @vpreduce_smin_nxv1i32(i32 signext %s, <vscale x 1 x i32> %v, <vscale x 1 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_smin_nxv1i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e32, mf2, tu, ma
 ; CHECK-NEXT:    vredmin.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i32 @llvm.vp.reduce.smin.nxv1i32(i32 %s, <vscale x 1 x i32> %v, <vscale x 1 x i1> %m, i32 %evl)
   ret i32 %r
 }
 
 declare i32 @llvm.vp.reduce.and.nxv1i32(i32, <vscale x 1 x i32>, <vscale x 1 x i1>, i32)
 
 define signext i32 @vpreduce_and_nxv1i32(i32 signext %s, <vscale x 1 x i32> %v, <vscale x 1 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_and_nxv1i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e32, mf2, tu, ma
 ; CHECK-NEXT:    vredand.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i32 @llvm.vp.reduce.and.nxv1i32(i32 %s, <vscale x 1 x i32> %v, <vscale x 1 x i1> %m, i32 %evl)
   ret i32 %r
 }
 
 declare i32 @llvm.vp.reduce.or.nxv1i32(i32, <vscale x 1 x i32>, <vscale x 1 x i1>, i32)
 
 define signext i32 @vpreduce_or_nxv1i32(i32 signext %s, <vscale x 1 x i32> %v, <vscale x 1 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_or_nxv1i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e32, mf2, tu, ma
 ; CHECK-NEXT:    vredor.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i32 @llvm.vp.reduce.or.nxv1i32(i32 %s, <vscale x 1 x i32> %v, <vscale x 1 x i1> %m, i32 %evl)
   ret i32 %r
 }
 
 declare i32 @llvm.vp.reduce.xor.nxv1i32(i32, <vscale x 1 x i32>, <vscale x 1 x i1>, i32)
 
 define signext i32 @vpreduce_xor_nxv1i32(i32 signext %s, <vscale x 1 x i32> %v, <vscale x 1 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_xor_nxv1i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e32, mf2, tu, ma
 ; CHECK-NEXT:    vredxor.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i32 @llvm.vp.reduce.xor.nxv1i32(i32 %s, <vscale x 1 x i32> %v, <vscale x 1 x i1> %m, i32 %evl)
   ret i32 %r
 }
 
 declare i32 @llvm.vp.reduce.add.nxv2i32(i32, <vscale x 2 x i32>, <vscale x 2 x i1>, i32)
 
 define signext i32 @vpreduce_add_nxv2i32(i32 signext %s, <vscale x 2 x i32> %v, <vscale x 2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_add_nxv2i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e32, m1, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e32, m1, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e32, m1, tu, ma
 ; CHECK-NEXT:    vredsum.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i32 @llvm.vp.reduce.add.nxv2i32(i32 %s, <vscale x 2 x i32> %v, <vscale x 2 x i1> %m, i32 %evl)
   ret i32 %r
 }
 
 declare i32 @llvm.vp.reduce.umax.nxv2i32(i32, <vscale x 2 x i32>, <vscale x 2 x i1>, i32)
 
 define signext i32 @vpreduce_umax_nxv2i32(i32 signext %s, <vscale x 2 x i32> %v, <vscale x 2 x i1> %m, i32 zeroext %evl) {
-; RV32-LABEL: vpreduce_umax_nxv2i32:
-; RV32:       # %bb.0:
-; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; RV32-NEXT:    vmv.s.x v9, a0
-; RV32-NEXT:    vsetvli zero, a1, e32, m1, tu, ma
-; RV32-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
-; RV32-NEXT:    vmv.x.s a0, v9
-; RV32-NEXT:    ret
-;
-; RV64-LABEL: vpreduce_umax_nxv2i32:
-; RV64:       # %bb.0:
-; RV64-NEXT:    slli a0, a0, 32
-; RV64-NEXT:    srli a0, a0, 32
-; RV64-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli zero, a1, e32, m1, tu, ma
-; RV64-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
-; RV64-NEXT:    vmv.x.s a0, v9
-; RV64-NEXT:    ret
+; CHECK-LABEL: vpreduce_umax_nxv2i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, m1, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e32, m1, tu, ma
+; CHECK-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
+; CHECK-NEXT:    vmv.x.s a0, v9
+; CHECK-NEXT:    ret
   %r = call i32 @llvm.vp.reduce.umax.nxv2i32(i32 %s, <vscale x 2 x i32> %v, <vscale x 2 x i1> %m, i32 %evl)
   ret i32 %r
 }
 
 declare i32 @llvm.vp.reduce.smax.nxv2i32(i32, <vscale x 2 x i32>, <vscale x 2 x i1>, i32)
 
 define signext i32 @vpreduce_smax_nxv2i32(i32 signext %s, <vscale x 2 x i32> %v, <vscale x 2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_smax_nxv2i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e32, m1, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e32, m1, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e32, m1, tu, ma
 ; CHECK-NEXT:    vredmax.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i32 @llvm.vp.reduce.smax.nxv2i32(i32 %s, <vscale x 2 x i32> %v, <vscale x 2 x i1> %m, i32 %evl)
   ret i32 %r
 }
 
 declare i32 @llvm.vp.reduce.umin.nxv2i32(i32, <vscale x 2 x i32>, <vscale x 2 x i1>, i32)
 
 define signext i32 @vpreduce_umin_nxv2i32(i32 signext %s, <vscale x 2 x i32> %v, <vscale x 2 x i1> %m, i32 zeroext %evl) {
-; RV32-LABEL: vpreduce_umin_nxv2i32:
-; RV32:       # %bb.0:
-; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; RV32-NEXT:    vmv.s.x v9, a0
-; RV32-NEXT:    vsetvli zero, a1, e32, m1, tu, ma
-; RV32-NEXT:    vredminu.vs v9, v8, v9, v0.t
-; RV32-NEXT:    vmv.x.s a0, v9
-; RV32-NEXT:    ret
-;
-; RV64-LABEL: vpreduce_umin_nxv2i32:
-; RV64:       # %bb.0:
-; RV64-NEXT:    slli a0, a0, 32
-; RV64-NEXT:    srli a0, a0, 32
-; RV64-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli zero, a1, e32, m1, tu, ma
-; RV64-NEXT:    vredminu.vs v9, v8, v9, v0.t
-; RV64-NEXT:    vmv.x.s a0, v9
-; RV64-NEXT:    ret
+; CHECK-LABEL: vpreduce_umin_nxv2i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a1, e32, m1, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e32, m1, tu, ma
+; CHECK-NEXT:    vredminu.vs v9, v8, v9, v0.t
+; CHECK-NEXT:    vmv.x.s a0, v9
+; CHECK-NEXT:    ret
   %r = call i32 @llvm.vp.reduce.umin.nxv2i32(i32 %s, <vscale x 2 x i32> %v, <vscale x 2 x i1> %m, i32 %evl)
   ret i32 %r
 }
 
 declare i32 @llvm.vp.reduce.smin.nxv2i32(i32, <vscale x 2 x i32>, <vscale x 2 x i1>, i32)
 
 define signext i32 @vpreduce_smin_nxv2i32(i32 signext %s, <vscale x 2 x i32> %v, <vscale x 2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_smin_nxv2i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e32, m1, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e32, m1, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e32, m1, tu, ma
 ; CHECK-NEXT:    vredmin.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i32 @llvm.vp.reduce.smin.nxv2i32(i32 %s, <vscale x 2 x i32> %v, <vscale x 2 x i1> %m, i32 %evl)
   ret i32 %r
 }
 
 declare i32 @llvm.vp.reduce.and.nxv2i32(i32, <vscale x 2 x i32>, <vscale x 2 x i1>, i32)
 
 define signext i32 @vpreduce_and_nxv2i32(i32 signext %s, <vscale x 2 x i32> %v, <vscale x 2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_and_nxv2i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e32, m1, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e32, m1, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e32, m1, tu, ma
 ; CHECK-NEXT:    vredand.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i32 @llvm.vp.reduce.and.nxv2i32(i32 %s, <vscale x 2 x i32> %v, <vscale x 2 x i1> %m, i32 %evl)
   ret i32 %r
 }
 
 declare i32 @llvm.vp.reduce.or.nxv2i32(i32, <vscale x 2 x i32>, <vscale x 2 x i1>, i32)
 
 define signext i32 @vpreduce_or_nxv2i32(i32 signext %s, <vscale x 2 x i32> %v, <vscale x 2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_or_nxv2i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e32, m1, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e32, m1, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e32, m1, tu, ma
 ; CHECK-NEXT:    vredor.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i32 @llvm.vp.reduce.or.nxv2i32(i32 %s, <vscale x 2 x i32> %v, <vscale x 2 x i1> %m, i32 %evl)
   ret i32 %r
 }
 
 declare i32 @llvm.vp.reduce.xor.nxv2i32(i32, <vscale x 2 x i32>, <vscale x 2 x i1>, i32)
 
 define signext i32 @vpreduce_xor_nxv2i32(i32 signext %s, <vscale x 2 x i32> %v, <vscale x 2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_xor_nxv2i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli zero, a1, e32, m1, tu, ma
+; CHECK-NEXT:    vsetvli zero, a1, e32, m1, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli zero, zero, e32, m1, tu, ma
 ; CHECK-NEXT:    vredxor.vs v9, v8, v9, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v9
 ; CHECK-NEXT:    ret
   %r = call i32 @llvm.vp.reduce.xor.nxv2i32(i32 %s, <vscale x 2 x i32> %v, <vscale x 2 x i1> %m, i32 %evl)
   ret i32 %r
 }
 
 declare i32 @llvm.vp.reduce.add.nxv4i32(i32, <vscale x 4 x i32>, <vscale x 4 x i1>, i32)

 define signext i64 @vpreduce_add_nxv1i64(i64 signext %s, <vscale x 1 x i64> %v, <vscale x 1 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_add_nxv1i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    addi sp, sp, -16
 ; RV32-NEXT:    .cfi_def_cfa_offset 16
 ; RV32-NEXT:    sw a1, 12(sp)
 ; RV32-NEXT:    sw a0, 8(sp)
 ; RV32-NEXT:    addi a0, sp, 8
-; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV32-NEXT:    vsetvli zero, a2, e64, m1, ta, ma
 ; RV32-NEXT:    vlse64.v v9, (a0), zero
-; RV32-NEXT:    vsetvli zero, a2, e64, m1, tu, ma
+; RV32-NEXT:    vsetvli zero, zero, e64, m1, tu, ma
 ; RV32-NEXT:    vredsum.vs v9, v8, v9, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v9
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vsrl.vx v8, v9, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_add_nxv1i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli zero, a1, e64, m1, tu, ma
+; RV64-NEXT:    vsetvli zero, a1, e64, m1, ta, ma
+; RV64-NEXT:    vmv.v.x v9, a0
+; RV64-NEXT:    vsetvli zero, zero, e64, m1, tu, ma
 ; RV64-NEXT:    vredsum.vs v9, v8, v9, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v9
 ; RV64-NEXT:    ret
   %r = call i64 @llvm.vp.reduce.add.nxv1i64(i64 %s, <vscale x 1 x i64> %v, <vscale x 1 x i1> %m, i32 %evl)
   ret i64 %r
 }
 
 define signext i64 @vpwreduce_add_nxv1i32(i64 signext %s, <vscale x 1 x i32> %v, <vscale x 1 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpwreduce_add_nxv1i32:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    addi sp, sp, -16
 ; RV32-NEXT:    .cfi_def_cfa_offset 16
 ; RV32-NEXT:    sw a1, 12(sp)
 ; RV32-NEXT:    sw a0, 8(sp)
 ; RV32-NEXT:    addi a0, sp, 8
-; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV32-NEXT:    vsetvli zero, a2, e64, m1, ta, ma
 ; RV32-NEXT:    vlse64.v v9, (a0), zero
-; RV32-NEXT:    vsetvli zero, a2, e32, mf2, tu, ma
+; RV32-NEXT:    vsetvli zero, zero, e32, mf2, tu, ma
 ; RV32-NEXT:    vwredsum.vs v9, v8, v9, v0.t
 ; RV32-NEXT:    vsetvli zero, zero, e64, m1, ta, ma
 ; RV32-NEXT:    vmv.x.s a0, v9
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vsrl.vx v8, v9, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpwreduce_add_nxv1i32:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli zero, a1, e32, mf2, tu, ma
+; RV64-NEXT:    vsetvli zero, a1, e64, m1, ta, ma
+; RV64-NEXT:    vmv.v.x v9, a0
+; RV64-NEXT:    vsetvli zero, zero, e32, mf2, tu, ma
 ; RV64-NEXT:    vwredsum.vs v9, v8, v9, v0.t
 ; RV64-NEXT:    vsetvli zero, zero, e64, m1, ta, ma
 ; RV64-NEXT:    vmv.x.s a0, v9
 ; RV64-NEXT:    ret
   %e = sext <vscale x 1 x i32> %v to <vscale x 1 x i64>
   %r = call i64 @llvm.vp.reduce.add.nxv1i64(i64 %s, <vscale x 1 x i64> %e, <vscale x 1 x i1> %m, i32 %evl)
   ret i64 %r
 }
 
 define signext i64 @vpwreduce_uadd_nxv1i32(i64 signext %s, <vscale x 1 x i32> %v, <vscale x 1 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpwreduce_uadd_nxv1i32:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    addi sp, sp, -16
 ; RV32-NEXT:    .cfi_def_cfa_offset 16
 ; RV32-NEXT:    sw a1, 12(sp)
 ; RV32-NEXT:    sw a0, 8(sp)
 ; RV32-NEXT:    addi a0, sp, 8
-; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV32-NEXT:    vsetvli zero, a2, e64, m1, ta, ma
 ; RV32-NEXT:    vlse64.v v9, (a0), zero
-; RV32-NEXT:    vsetvli zero, a2, e32, mf2, tu, ma
+; RV32-NEXT:    vsetvli zero, zero, e32, mf2, tu, ma
 ; RV32-NEXT:    vwredsum.vs v9, v8, v9, v0.t
 ; RV32-NEXT:    vsetvli zero, zero, e64, m1, ta, ma
 ; RV32-NEXT:    vmv.x.s a0, v9
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vsrl.vx v8, v9, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpwreduce_uadd_nxv1i32:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli zero, a1, e32, mf2, tu, ma
+; RV64-NEXT:    vsetvli zero, a1, e64, m1, ta, ma
+; RV64-NEXT:    vmv.v.x v9, a0
+; RV64-NEXT:    vsetvli zero, zero, e32, mf2, tu, ma
 ; RV64-NEXT:    vwredsum.vs v9, v8, v9, v0.t
 ; RV64-NEXT:    vsetvli zero, zero, e64, m1, ta, ma
 ; RV64-NEXT:    vmv.x.s a0, v9
 ; RV64-NEXT:    ret
   %e = sext <vscale x 1 x i32> %v to <vscale x 1 x i64>
   %r = call i64 @llvm.vp.reduce.add.nxv1i64(i64 %s, <vscale x 1 x i64> %e, <vscale x 1 x i1> %m, i32 %evl)
   ret i64 %r
 }
 
 declare i64 @llvm.vp.reduce.umax.nxv1i64(i64, <vscale x 1 x i64>, <vscale x 1 x i1>, i32)
 
 define signext i64 @vpreduce_umax_nxv1i64(i64 signext %s, <vscale x 1 x i64> %v, <vscale x 1 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_umax_nxv1i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    addi sp, sp, -16
 ; RV32-NEXT:    .cfi_def_cfa_offset 16
 ; RV32-NEXT:    sw a1, 12(sp)
 ; RV32-NEXT:    sw a0, 8(sp)
 ; RV32-NEXT:    addi a0, sp, 8
-; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV32-NEXT:    vsetvli zero, a2, e64, m1, ta, ma
 ; RV32-NEXT:    vlse64.v v9, (a0), zero
-; RV32-NEXT:    vsetvli zero, a2, e64, m1, tu, ma
+; RV32-NEXT:    vsetvli zero, zero, e64, m1, tu, ma
 ; RV32-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v9
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vsrl.vx v8, v9, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_umax_nxv1i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli zero, a1, e64, m1, tu, ma
+; RV64-NEXT:    vsetvli zero, a1, e64, m1, ta, ma
+; RV64-NEXT:    vmv.v.x v9, a0
+; RV64-NEXT:    vsetvli zero, zero, e64, m1, tu, ma
 ; RV64-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v9
 ; RV64-NEXT:    ret
   %r = call i64 @llvm.vp.reduce.umax.nxv1i64(i64 %s, <vscale x 1 x i64> %v, <vscale x 1 x i1> %m, i32 %evl)
   ret i64 %r
 }
 
 declare i64 @llvm.vp.reduce.smax.nxv1i64(i64, <vscale x 1 x i64>, <vscale x 1 x i1>, i32)
 
 define signext i64 @vpreduce_smax_nxv1i64(i64 signext %s, <vscale x 1 x i64> %v, <vscale x 1 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_smax_nxv1i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    addi sp, sp, -16
 ; RV32-NEXT:    .cfi_def_cfa_offset 16
 ; RV32-NEXT:    sw a1, 12(sp)
 ; RV32-NEXT:    sw a0, 8(sp)
 ; RV32-NEXT:    addi a0, sp, 8
-; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV32-NEXT:    vsetvli zero, a2, e64, m1, ta, ma
 ; RV32-NEXT:    vlse64.v v9, (a0), zero
-; RV32-NEXT:    vsetvli zero, a2, e64, m1, tu, ma
+; RV32-NEXT:    vsetvli zero, zero, e64, m1, tu, ma
 ; RV32-NEXT:    vredmax.vs v9, v8, v9, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v9
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vsrl.vx v8, v9, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_smax_nxv1i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli zero, a1, e64, m1, tu, ma
+; RV64-NEXT:    vsetvli zero, a1, e64, m1, ta, ma
+; RV64-NEXT:    vmv.v.x v9, a0
+; RV64-NEXT:    vsetvli zero, zero, e64, m1, tu, ma
 ; RV64-NEXT:    vredmax.vs v9, v8, v9, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v9
 ; RV64-NEXT:    ret
   %r = call i64 @llvm.vp.reduce.smax.nxv1i64(i64 %s, <vscale x 1 x i64> %v, <vscale x 1 x i1> %m, i32 %evl)
   ret i64 %r
 }
 
 declare i64 @llvm.vp.reduce.umin.nxv1i64(i64, <vscale x 1 x i64>, <vscale x 1 x i1>, i32)
 
 define signext i64 @vpreduce_umin_nxv1i64(i64 signext %s, <vscale x 1 x i64> %v, <vscale x 1 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_umin_nxv1i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    addi sp, sp, -16
 ; RV32-NEXT:    .cfi_def_cfa_offset 16
 ; RV32-NEXT:    sw a1, 12(sp)
 ; RV32-NEXT:    sw a0, 8(sp)
 ; RV32-NEXT:    addi a0, sp, 8
-; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV32-NEXT:    vsetvli zero, a2, e64, m1, ta, ma
 ; RV32-NEXT:    vlse64.v v9, (a0), zero
-; RV32-NEXT:    vsetvli zero, a2, e64, m1, tu, ma
+; RV32-NEXT:    vsetvli zero, zero, e64, m1, tu, ma
 ; RV32-NEXT:    vredminu.vs v9, v8, v9, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v9
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vsrl.vx v8, v9, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_umin_nxv1i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli zero, a1, e64, m1, tu, ma
+; RV64-NEXT:    vsetvli zero, a1, e64, m1, ta, ma
+; RV64-NEXT:    vmv.v.x v9, a0
+; RV64-NEXT:    vsetvli zero, zero, e64, m1, tu, ma
 ; RV64-NEXT:    vredminu.vs v9, v8, v9, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v9
 ; RV64-NEXT:    ret
   %r = call i64 @llvm.vp.reduce.umin.nxv1i64(i64 %s, <vscale x 1 x i64> %v, <vscale x 1 x i1> %m, i32 %evl)
   ret i64 %r
 }
 
 declare i64 @llvm.vp.reduce.smin.nxv1i64(i64, <vscale x 1 x i64>, <vscale x 1 x i1>, i32)
 
 define signext i64 @vpreduce_smin_nxv1i64(i64 signext %s, <vscale x 1 x i64> %v, <vscale x 1 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_smin_nxv1i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    addi sp, sp, -16
 ; RV32-NEXT:    .cfi_def_cfa_offset 16
 ; RV32-NEXT:    sw a1, 12(sp)
 ; RV32-NEXT:    sw a0, 8(sp)
 ; RV32-NEXT:    addi a0, sp, 8
-; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV32-NEXT:    vsetvli zero, a2, e64, m1, ta, ma
 ; RV32-NEXT:    vlse64.v v9, (a0), zero
-; RV32-NEXT:    vsetvli zero, a2, e64, m1, tu, ma
+; RV32-NEXT:    vsetvli zero, zero, e64, m1, tu, ma
 ; RV32-NEXT:    vredmin.vs v9, v8, v9, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v9
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vsrl.vx v8, v9, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_smin_nxv1i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli zero, a1, e64, m1, tu, ma
+; RV64-NEXT:    vsetvli zero, a1, e64, m1, ta, ma
+; RV64-NEXT:    vmv.v.x v9, a0
+; RV64-NEXT:    vsetvli zero, zero, e64, m1, tu, ma
 ; RV64-NEXT:    vredmin.vs v9, v8, v9, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v9
 ; RV64-NEXT:    ret
   %r = call i64 @llvm.vp.reduce.smin.nxv1i64(i64 %s, <vscale x 1 x i64> %v, <vscale x 1 x i1> %m, i32 %evl)
   ret i64 %r
 }
 
 declare i64 @llvm.vp.reduce.and.nxv1i64(i64, <vscale x 1 x i64>, <vscale x 1 x i1>, i32)
 
 define signext i64 @vpreduce_and_nxv1i64(i64 signext %s, <vscale x 1 x i64> %v, <vscale x 1 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_and_nxv1i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    addi sp, sp, -16
 ; RV32-NEXT:    .cfi_def_cfa_offset 16
 ; RV32-NEXT:    sw a1, 12(sp)
 ; RV32-NEXT:    sw a0, 8(sp)
 ; RV32-NEXT:    addi a0, sp, 8
-; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV32-NEXT:    vsetvli zero, a2, e64, m1, ta, ma
 ; RV32-NEXT:    vlse64.v v9, (a0), zero
-; RV32-NEXT:    vsetvli zero, a2, e64, m1, tu, ma
+; RV32-NEXT:    vsetvli zero, zero, e64, m1, tu, ma
 ; RV32-NEXT:    vredand.vs v9, v8, v9, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v9
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vsrl.vx v8, v9, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_and_nxv1i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli zero, a1, e64, m1, tu, ma
+; RV64-NEXT:    vsetvli zero, a1, e64, m1, ta, ma
+; RV64-NEXT:    vmv.v.x v9, a0
+; RV64-NEXT:    vsetvli zero, zero, e64, m1, tu, ma
 ; RV64-NEXT:    vredand.vs v9, v8, v9, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v9
 ; RV64-NEXT:    ret
   %r = call i64 @llvm.vp.reduce.and.nxv1i64(i64 %s, <vscale x 1 x i64> %v, <vscale x 1 x i1> %m, i32 %evl)
   ret i64 %r
 }
 
 declare i64 @llvm.vp.reduce.or.nxv1i64(i64, <vscale x 1 x i64>, <vscale x 1 x i1>, i32)
 
 define signext i64 @vpreduce_or_nxv1i64(i64 signext %s, <vscale x 1 x i64> %v, <vscale x 1 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_or_nxv1i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    addi sp, sp, -16
 ; RV32-NEXT:    .cfi_def_cfa_offset 16
 ; RV32-NEXT:    sw a1, 12(sp)
 ; RV32-NEXT:    sw a0, 8(sp)
 ; RV32-NEXT:    addi a0, sp, 8
-; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV32-NEXT:    vsetvli zero, a2, e64, m1, ta, ma
 ; RV32-NEXT:    vlse64.v v9, (a0), zero
-; RV32-NEXT:    vsetvli zero, a2, e64, m1, tu, ma
+; RV32-NEXT:    vsetvli zero, zero, e64, m1, tu, ma
 ; RV32-NEXT:    vredor.vs v9, v8, v9, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v9
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vsrl.vx v8, v9, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_or_nxv1i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli zero, a1, e64, m1, tu, ma
+; RV64-NEXT:    vsetvli zero, a1, e64, m1, ta, ma
+; RV64-NEXT:    vmv.v.x v9, a0
+; RV64-NEXT:    vsetvli zero, zero, e64, m1, tu, ma
 ; RV64-NEXT:    vredor.vs v9, v8, v9, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v9
 ; RV64-NEXT:    ret
   %r = call i64 @llvm.vp.reduce.or.nxv1i64(i64 %s, <vscale x 1 x i64> %v, <vscale x 1 x i1> %m, i32 %evl)
   ret i64 %r
 }
 
 declare i64 @llvm.vp.reduce.xor.nxv1i64(i64, <vscale x 1 x i64>, <vscale x 1 x i1>, i32)
 
 define signext i64 @vpreduce_xor_nxv1i64(i64 signext %s, <vscale x 1 x i64> %v, <vscale x 1 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_xor_nxv1i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    addi sp, sp, -16
 ; RV32-NEXT:    .cfi_def_cfa_offset 16
 ; RV32-NEXT:    sw a1, 12(sp)
 ; RV32-NEXT:    sw a0, 8(sp)
 ; RV32-NEXT:    addi a0, sp, 8
-; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV32-NEXT:    vsetvli zero, a2, e64, m1, ta, ma
 ; RV32-NEXT:    vlse64.v v9, (a0), zero
-; RV32-NEXT:    vsetvli zero, a2, e64, m1, tu, ma
+; RV32-NEXT:    vsetvli zero, zero, e64, m1, tu, ma
 ; RV32-NEXT:    vredxor.vs v9, v8, v9, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v9
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vsrl.vx v8, v9, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_xor_nxv1i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli zero, a1, e64, m1, tu, ma
+; RV64-NEXT:    vsetvli zero, a1, e64, m1, ta, ma
+; RV64-NEXT:    vmv.v.x v9, a0
+; RV64-NEXT:    vsetvli zero, zero, e64, m1, tu, ma
 ; RV64-NEXT:    vredxor.vs v9, v8, v9, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v9
 ; RV64-NEXT:    ret
   %r = call i64 @llvm.vp.reduce.xor.nxv1i64(i64 %s, <vscale x 1 x i64> %v, <vscale x 1 x i1> %m, i32 %evl)
   ret i64 %r
 }
 
 declare i64 @llvm.vp.reduce.add.nxv2i64(i64, <vscale x 2 x i64>, <vscale x 2 x i1>, i32)

llvm/test/CodeGen/RISCV/rvv/vreductions-int.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: sed 's/iXLen/i32/g' %s | llc -mtriple=riscv32 -mattr=+v \
 ; RUN:   -verify-machineinstrs | FileCheck %s --check-prefixes=CHECK,RV32
 ; RUN: sed 's/iXLen/i64/g' %s | llc -mtriple=riscv64 -mattr=+v \
 ; RUN:   -verify-machineinstrs | FileCheck %s --check-prefixes=CHECK,RV64
 
 declare i8 @llvm.vector.reduce.add.nxv1i8(<vscale x 1 x i8>)
 
 define signext i8 @vreduce_add_nxv1i8(<vscale x 1 x i8> %v) {
 ; CHECK-LABEL: vreduce_add_nxv1i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf8, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredsum.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i8 @llvm.vector.reduce.add.nxv1i8(<vscale x 1 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.umax.nxv1i8(<vscale x 1 x i8>)
 
 define signext i8 @vreduce_umax_nxv1i8(<vscale x 1 x i8> %v) {
 ; CHECK-LABEL: vreduce_umax_nxv1i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf8, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredmaxu.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i8 @llvm.vector.reduce.umax.nxv1i8(<vscale x 1 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.smax.nxv1i8(<vscale x 1 x i8>)
 
 define signext i8 @vreduce_smax_nxv1i8(<vscale x 1 x i8> %v) {
 ; CHECK-LABEL: vreduce_smax_nxv1i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a0, -128
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli a0, zero, e8, mf8, ta, ma
+; CHECK-NEXT:    vsetvli a1, zero, e8, mf8, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vredmax.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i8 @llvm.vector.reduce.smax.nxv1i8(<vscale x 1 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.umin.nxv1i8(<vscale x 1 x i8>)
 
 define signext i8 @vreduce_umin_nxv1i8(<vscale x 1 x i8> %v) {
 ; CHECK-LABEL: vreduce_umin_nxv1i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf8, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vredminu.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i8 @llvm.vector.reduce.umin.nxv1i8(<vscale x 1 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.smin.nxv1i8(<vscale x 1 x i8>)
 
 define signext i8 @vreduce_smin_nxv1i8(<vscale x 1 x i8> %v) {
 ; CHECK-LABEL: vreduce_smin_nxv1i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a0, 127
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli a0, zero, e8, mf8, ta, ma
+; CHECK-NEXT:    vsetvli a1, zero, e8, mf8, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vredmin.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i8 @llvm.vector.reduce.smin.nxv1i8(<vscale x 1 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.and.nxv1i8(<vscale x 1 x i8>)
 
 define signext i8 @vreduce_and_nxv1i8(<vscale x 1 x i8> %v) {
 ; CHECK-LABEL: vreduce_and_nxv1i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf8, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vredand.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i8 @llvm.vector.reduce.and.nxv1i8(<vscale x 1 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.or.nxv1i8(<vscale x 1 x i8>)
 
 define signext i8 @vreduce_or_nxv1i8(<vscale x 1 x i8> %v) {
 ; CHECK-LABEL: vreduce_or_nxv1i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf8, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredor.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i8 @llvm.vector.reduce.or.nxv1i8(<vscale x 1 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.xor.nxv1i8(<vscale x 1 x i8>)
 
 define signext i8 @vreduce_xor_nxv1i8(<vscale x 1 x i8> %v) {
 ; CHECK-LABEL: vreduce_xor_nxv1i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf8, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredxor.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i8 @llvm.vector.reduce.xor.nxv1i8(<vscale x 1 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.add.nxv2i8(<vscale x 2 x i8>)
 
 define signext i8 @vreduce_add_nxv2i8(<vscale x 2 x i8> %v) {
 ; CHECK-LABEL: vreduce_add_nxv2i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredsum.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i8 @llvm.vector.reduce.add.nxv2i8(<vscale x 2 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.umax.nxv2i8(<vscale x 2 x i8>)
 
 define signext i8 @vreduce_umax_nxv2i8(<vscale x 2 x i8> %v) {
 ; CHECK-LABEL: vreduce_umax_nxv2i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredmaxu.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i8 @llvm.vector.reduce.umax.nxv2i8(<vscale x 2 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.smax.nxv2i8(<vscale x 2 x i8>)
 
 define signext i8 @vreduce_smax_nxv2i8(<vscale x 2 x i8> %v) {
 ; CHECK-LABEL: vreduce_smax_nxv2i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a0, -128
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli a0, zero, e8, mf4, ta, ma
+; CHECK-NEXT:    vsetvli a1, zero, e8, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vredmax.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i8 @llvm.vector.reduce.smax.nxv2i8(<vscale x 2 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.umin.nxv2i8(<vscale x 2 x i8>)
 
 define signext i8 @vreduce_umin_nxv2i8(<vscale x 2 x i8> %v) {
 ; CHECK-LABEL: vreduce_umin_nxv2i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vredminu.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i8 @llvm.vector.reduce.umin.nxv2i8(<vscale x 2 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.smin.nxv2i8(<vscale x 2 x i8>)
 
 define signext i8 @vreduce_smin_nxv2i8(<vscale x 2 x i8> %v) {
 ; CHECK-LABEL: vreduce_smin_nxv2i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a0, 127
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli a0, zero, e8, mf4, ta, ma
+; CHECK-NEXT:    vsetvli a1, zero, e8, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vredmin.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i8 @llvm.vector.reduce.smin.nxv2i8(<vscale x 2 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.and.nxv2i8(<vscale x 2 x i8>)
 
 define signext i8 @vreduce_and_nxv2i8(<vscale x 2 x i8> %v) {
 ; CHECK-LABEL: vreduce_and_nxv2i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vredand.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i8 @llvm.vector.reduce.and.nxv2i8(<vscale x 2 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.or.nxv2i8(<vscale x 2 x i8>)
 
 define signext i8 @vreduce_or_nxv2i8(<vscale x 2 x i8> %v) {
 ; CHECK-LABEL: vreduce_or_nxv2i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredor.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i8 @llvm.vector.reduce.or.nxv2i8(<vscale x 2 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.xor.nxv2i8(<vscale x 2 x i8>)
 
 define signext i8 @vreduce_xor_nxv2i8(<vscale x 2 x i8> %v) {
 ; CHECK-LABEL: vreduce_xor_nxv2i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredxor.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i8 @llvm.vector.reduce.xor.nxv2i8(<vscale x 2 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.add.nxv4i8(<vscale x 4 x i8>)
 
 define signext i8 @vreduce_add_nxv4i8(<vscale x 4 x i8> %v) {
 ; CHECK-LABEL: vreduce_add_nxv4i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredsum.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i8 @llvm.vector.reduce.add.nxv4i8(<vscale x 4 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.umax.nxv4i8(<vscale x 4 x i8>)
 
 define signext i8 @vreduce_umax_nxv4i8(<vscale x 4 x i8> %v) {
 ; CHECK-LABEL: vreduce_umax_nxv4i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredmaxu.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i8 @llvm.vector.reduce.umax.nxv4i8(<vscale x 4 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.smax.nxv4i8(<vscale x 4 x i8>)
 
 define signext i8 @vreduce_smax_nxv4i8(<vscale x 4 x i8> %v) {
 ; CHECK-LABEL: vreduce_smax_nxv4i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a0, -128
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli a0, zero, e8, mf2, ta, ma
+; CHECK-NEXT:    vsetvli a1, zero, e8, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vredmax.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i8 @llvm.vector.reduce.smax.nxv4i8(<vscale x 4 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.umin.nxv4i8(<vscale x 4 x i8>)
 
 define signext i8 @vreduce_umin_nxv4i8(<vscale x 4 x i8> %v) {
 ; CHECK-LABEL: vreduce_umin_nxv4i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vredminu.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i8 @llvm.vector.reduce.umin.nxv4i8(<vscale x 4 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.smin.nxv4i8(<vscale x 4 x i8>)
 
 define signext i8 @vreduce_smin_nxv4i8(<vscale x 4 x i8> %v) {
 ; CHECK-LABEL: vreduce_smin_nxv4i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a0, 127
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli a0, zero, e8, mf2, ta, ma
+; CHECK-NEXT:    vsetvli a1, zero, e8, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vredmin.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i8 @llvm.vector.reduce.smin.nxv4i8(<vscale x 4 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.and.nxv4i8(<vscale x 4 x i8>)
 
 define signext i8 @vreduce_and_nxv4i8(<vscale x 4 x i8> %v) {
 ; CHECK-LABEL: vreduce_and_nxv4i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vredand.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i8 @llvm.vector.reduce.and.nxv4i8(<vscale x 4 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.or.nxv4i8(<vscale x 4 x i8>)
 
 define signext i8 @vreduce_or_nxv4i8(<vscale x 4 x i8> %v) {
 ; CHECK-LABEL: vreduce_or_nxv4i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredor.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i8 @llvm.vector.reduce.or.nxv4i8(<vscale x 4 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.xor.nxv4i8(<vscale x 4 x i8>)
 
 define signext i8 @vreduce_xor_nxv4i8(<vscale x 4 x i8> %v) {
 ; CHECK-LABEL: vreduce_xor_nxv4i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredxor.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i8 @llvm.vector.reduce.xor.nxv4i8(<vscale x 4 x i8> %v)
   ret i8 %red
 }
 
 declare i16 @llvm.vector.reduce.add.nxv1i16(<vscale x 1 x i16>)
 
 define signext i16 @vreduce_add_nxv1i16(<vscale x 1 x i16> %v) {
 ; CHECK-LABEL: vreduce_add_nxv1i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredsum.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i16 @llvm.vector.reduce.add.nxv1i16(<vscale x 1 x i16> %v)
   ret i16 %red
 }
 
 define signext i16 @vwreduce_add_nxv1i8(<vscale x 1 x i8> %v) {
 ; CHECK-LABEL: vwreduce_add_nxv1i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
-; CHECK-NEXT:    vsetvli a0, zero, e8, mf8, ta, ma
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf8, ta, ma
 ; CHECK-NEXT:    vwredsum.vs v8, v8, v9
 ; CHECK-NEXT:    vsetivli zero, 0, e16, m1, ta, ma
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %e = sext <vscale x 1 x i8> %v to <vscale x 1 x i16>
   %red = call i16 @llvm.vector.reduce.add.nxv1i16(<vscale x 1 x i16> %e)
   ret i16 %red
 }
 
 define signext i16 @vwreduce_uadd_nxv1i8(<vscale x 1 x i8> %v) {
 ; CHECK-LABEL: vwreduce_uadd_nxv1i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
-; CHECK-NEXT:    vsetvli a0, zero, e8, mf8, ta, ma
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf8, ta, ma
 ; CHECK-NEXT:    vwredsum.vs v8, v8, v9
 ; CHECK-NEXT:    vsetivli zero, 0, e16, m1, ta, ma
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %e = sext <vscale x 1 x i8> %v to <vscale x 1 x i16>
   %red = call i16 @llvm.vector.reduce.add.nxv1i16(<vscale x 1 x i16> %e)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.umax.nxv1i16(<vscale x 1 x i16>)
 
 define signext i16 @vreduce_umax_nxv1i16(<vscale x 1 x i16> %v) {
 ; CHECK-LABEL: vreduce_umax_nxv1i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredmaxu.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i16 @llvm.vector.reduce.umax.nxv1i16(<vscale x 1 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.smax.nxv1i16(<vscale x 1 x i16>)
 
 define signext i16 @vreduce_smax_nxv1i16(<vscale x 1 x i16> %v) {
 ; CHECK-LABEL: vreduce_smax_nxv1i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, 1048568
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vsetvli a1, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vredmax.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i16 @llvm.vector.reduce.smax.nxv1i16(<vscale x 1 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.umin.nxv1i16(<vscale x 1 x i16>)
 
 define signext i16 @vreduce_umin_nxv1i16(<vscale x 1 x i16> %v) {
 ; CHECK-LABEL: vreduce_umin_nxv1i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vredminu.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i16 @llvm.vector.reduce.umin.nxv1i16(<vscale x 1 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.smin.nxv1i16(<vscale x 1 x i16>)
 
 define signext i16 @vreduce_smin_nxv1i16(<vscale x 1 x i16> %v) {
 ; RV32-LABEL: vreduce_smin_nxv1i16:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    lui a0, 8
 ; RV32-NEXT:    addi a0, a0, -1
-; RV32-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; RV32-NEXT:    vmv.s.x v9, a0
-; RV32-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; RV32-NEXT:    vsetvli a1, zero, e16, mf4, ta, ma
+; RV32-NEXT:    vmv.v.x v9, a0
 ; RV32-NEXT:    vredmin.vs v8, v8, v9
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_smin_nxv1i16:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    lui a0, 8
 ; RV64-NEXT:    addiw a0, a0, -1
-; RV64-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; RV64-NEXT:    vsetvli a1, zero, e16, mf4, ta, ma
+; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vredmin.vs v8, v8, v9
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %red = call i16 @llvm.vector.reduce.smin.nxv1i16(<vscale x 1 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.and.nxv1i16(<vscale x 1 x i16>)
 
 define signext i16 @vreduce_and_nxv1i16(<vscale x 1 x i16> %v) {
 ; CHECK-LABEL: vreduce_and_nxv1i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vredand.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i16 @llvm.vector.reduce.and.nxv1i16(<vscale x 1 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.or.nxv1i16(<vscale x 1 x i16>)
 
 define signext i16 @vreduce_or_nxv1i16(<vscale x 1 x i16> %v) {
 ; CHECK-LABEL: vreduce_or_nxv1i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredor.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i16 @llvm.vector.reduce.or.nxv1i16(<vscale x 1 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.xor.nxv1i16(<vscale x 1 x i16>)
 
 define signext i16 @vreduce_xor_nxv1i16(<vscale x 1 x i16> %v) {
 ; CHECK-LABEL: vreduce_xor_nxv1i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredxor.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i16 @llvm.vector.reduce.xor.nxv1i16(<vscale x 1 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.add.nxv2i16(<vscale x 2 x i16>)
 
 define signext i16 @vreduce_add_nxv2i16(<vscale x 2 x i16> %v) {
 ; CHECK-LABEL: vreduce_add_nxv2i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredsum.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i16 @llvm.vector.reduce.add.nxv2i16(<vscale x 2 x i16> %v)
   ret i16 %red
 }
 
 define signext i16 @vwreduce_add_nxv2i8(<vscale x 2 x i8> %v) {
 ; CHECK-LABEL: vwreduce_add_nxv2i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
-; CHECK-NEXT:    vsetvli a0, zero, e8, mf4, ta, ma
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf4, ta, ma
 ; CHECK-NEXT:    vwredsum.vs v8, v8, v9
 ; CHECK-NEXT:    vsetivli zero, 0, e16, m1, ta, ma
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %e = sext <vscale x 2 x i8> %v to <vscale x 2 x i16>
   %red = call i16 @llvm.vector.reduce.add.nxv2i16(<vscale x 2 x i16> %e)
   ret i16 %red
 }
 
 define signext i16 @vwreduce_uadd_nxv2i8(<vscale x 2 x i8> %v) {
 ; CHECK-LABEL: vwreduce_uadd_nxv2i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
-; CHECK-NEXT:    vsetvli a0, zero, e8, mf4, ta, ma
+; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf4, ta, ma
 ; CHECK-NEXT:    vwredsum.vs v8, v8, v9
 ; CHECK-NEXT:    vsetivli zero, 0, e16, m1, ta, ma
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %e = sext <vscale x 2 x i8> %v to <vscale x 2 x i16>
   %red = call i16 @llvm.vector.reduce.add.nxv2i16(<vscale x 2 x i16> %e)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.umax.nxv2i16(<vscale x 2 x i16>)
 
 define signext i16 @vreduce_umax_nxv2i16(<vscale x 2 x i16> %v) {
 ; CHECK-LABEL: vreduce_umax_nxv2i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredmaxu.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i16 @llvm.vector.reduce.umax.nxv2i16(<vscale x 2 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.smax.nxv2i16(<vscale x 2 x i16>)
 
 define signext i16 @vreduce_smax_nxv2i16(<vscale x 2 x i16> %v) {
 ; CHECK-LABEL: vreduce_smax_nxv2i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, 1048568
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vsetvli a1, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vredmax.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i16 @llvm.vector.reduce.smax.nxv2i16(<vscale x 2 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.umin.nxv2i16(<vscale x 2 x i16>)
 
 define signext i16 @vreduce_umin_nxv2i16(<vscale x 2 x i16> %v) {
 ; CHECK-LABEL: vreduce_umin_nxv2i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vredminu.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i16 @llvm.vector.reduce.umin.nxv2i16(<vscale x 2 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.smin.nxv2i16(<vscale x 2 x i16>)
 
 define signext i16 @vreduce_smin_nxv2i16(<vscale x 2 x i16> %v) {
 ; RV32-LABEL: vreduce_smin_nxv2i16:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    lui a0, 8
 ; RV32-NEXT:    addi a0, a0, -1
-; RV32-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; RV32-NEXT:    vmv.s.x v9, a0
-; RV32-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; RV32-NEXT:    vsetvli a1, zero, e16, mf2, ta, ma
+; RV32-NEXT:    vmv.v.x v9, a0
 ; RV32-NEXT:    vredmin.vs v8, v8, v9
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_smin_nxv2i16:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    lui a0, 8
 ; RV64-NEXT:    addiw a0, a0, -1
-; RV64-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; RV64-NEXT:    vsetvli a1, zero, e16, mf2, ta, ma
+; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vredmin.vs v8, v8, v9
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %red = call i16 @llvm.vector.reduce.smin.nxv2i16(<vscale x 2 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.and.nxv2i16(<vscale x 2 x i16>)
 
 define signext i16 @vreduce_and_nxv2i16(<vscale x 2 x i16> %v) {
 ; CHECK-LABEL: vreduce_and_nxv2i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vredand.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i16 @llvm.vector.reduce.and.nxv2i16(<vscale x 2 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.or.nxv2i16(<vscale x 2 x i16>)
 
 define signext i16 @vreduce_or_nxv2i16(<vscale x 2 x i16> %v) {
 ; CHECK-LABEL: vreduce_or_nxv2i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredor.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i16 @llvm.vector.reduce.or.nxv2i16(<vscale x 2 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.xor.nxv2i16(<vscale x 2 x i16>)
 
 define signext i16 @vreduce_xor_nxv2i16(<vscale x 2 x i16> %v) {
 ; CHECK-LABEL: vreduce_xor_nxv2i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredxor.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i16 @llvm.vector.reduce.xor.nxv2i16(<vscale x 2 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.add.nxv4i16(<vscale x 4 x i16>)
 
 define signext i16 @vreduce_add_nxv4i16(<vscale x 4 x i16> %v) {
 ; CHECK-LABEL: vreduce_add_nxv4i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
 ; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredsum.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i16 @llvm.vector.reduce.add.nxv4i16(<vscale x 4 x i16> %v)
   ret i16 %red
 }
 
 define signext i16 @vwreduce_add_nxv4i8(<vscale x 4 x i8> %v) {
 ; CHECK-LABEL: vwreduce_add_nxv4i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
-; CHECK-NEXT:    vsetvli a0, zero, e8, mf2, ta, ma
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf2, ta, ma
 ; CHECK-NEXT:    vwredsum.vs v8, v8, v9
 ; CHECK-NEXT:    vsetvli zero, zero, e16, m1, ta, ma
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %e = sext <vscale x 4 x i8> %v to <vscale x 4 x i16>
   %red = call i16 @llvm.vector.reduce.add.nxv4i16(<vscale x 4 x i16> %e)
   ret i16 %red
 }
 
 define signext i16 @vwreduce_uadd_nxv4i8(<vscale x 4 x i8> %v) {
 ; CHECK-LABEL: vwreduce_uadd_nxv4i8:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
-; CHECK-NEXT:    vsetvli a0, zero, e8, mf2, ta, ma
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
+; CHECK-NEXT:    vsetvli zero, zero, e8, mf2, ta, ma
 ; CHECK-NEXT:    vwredsum.vs v8, v8, v9
 ; CHECK-NEXT:    vsetvli zero, zero, e16, m1, ta, ma
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %e = sext <vscale x 4 x i8> %v to <vscale x 4 x i16>
   %red = call i16 @llvm.vector.reduce.add.nxv4i16(<vscale x 4 x i16> %e)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.umax.nxv4i16(<vscale x 4 x i16>)
 
 define signext i16 @vreduce_umax_nxv4i16(<vscale x 4 x i16> %v) {
 ; CHECK-LABEL: vreduce_umax_nxv4i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
 ; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredmaxu.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i16 @llvm.vector.reduce.umax.nxv4i16(<vscale x 4 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.smax.nxv4i16(<vscale x 4 x i16>)
 
 define signext i16 @vreduce_smax_nxv4i16(<vscale x 4 x i16> %v) {
 ; CHECK-LABEL: vreduce_smax_nxv4i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, 1048568
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vsetvli a1, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vredmax.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i16 @llvm.vector.reduce.smax.nxv4i16(<vscale x 4 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.umin.nxv4i16(<vscale x 4 x i16>)
 
 define signext i16 @vreduce_umin_nxv4i16(<vscale x 4 x i16> %v) {
 ; CHECK-LABEL: vreduce_umin_nxv4i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vredminu.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i16 @llvm.vector.reduce.umin.nxv4i16(<vscale x 4 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.smin.nxv4i16(<vscale x 4 x i16>)
 
 define signext i16 @vreduce_smin_nxv4i16(<vscale x 4 x i16> %v) {
 ; RV32-LABEL: vreduce_smin_nxv4i16:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    lui a0, 8
 ; RV32-NEXT:    addi a0, a0, -1
-; RV32-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; RV32-NEXT:    vmv.s.x v9, a0
-; RV32-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; RV32-NEXT:    vsetvli a1, zero, e16, m1, ta, ma
+; RV32-NEXT:    vmv.v.x v9, a0
 ; RV32-NEXT:    vredmin.vs v8, v8, v9
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_smin_nxv4i16:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    lui a0, 8
 ; RV64-NEXT:    addiw a0, a0, -1
-; RV64-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; RV64-NEXT:    vsetvli a1, zero, e16, m1, ta, ma
+; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vredmin.vs v8, v8, v9
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %red = call i16 @llvm.vector.reduce.smin.nxv4i16(<vscale x 4 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.and.nxv4i16(<vscale x 4 x i16>)
 
 define signext i16 @vreduce_and_nxv4i16(<vscale x 4 x i16> %v) {
 ; CHECK-LABEL: vreduce_and_nxv4i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vredand.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i16 @llvm.vector.reduce.and.nxv4i16(<vscale x 4 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.or.nxv4i16(<vscale x 4 x i16>)
 
 define signext i16 @vreduce_or_nxv4i16(<vscale x 4 x i16> %v) {
 ; CHECK-LABEL: vreduce_or_nxv4i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
 ; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredor.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i16 @llvm.vector.reduce.or.nxv4i16(<vscale x 4 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.xor.nxv4i16(<vscale x 4 x i16>)
 
 define signext i16 @vreduce_xor_nxv4i16(<vscale x 4 x i16> %v) {
 ; CHECK-LABEL: vreduce_xor_nxv4i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
 ; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredxor.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i16 @llvm.vector.reduce.xor.nxv4i16(<vscale x 4 x i16> %v)
   ret i16 %red
 }
 
 declare i32 @llvm.vector.reduce.add.nxv1i32(<vscale x 1 x i32>)
 
 define signext i32 @vreduce_add_nxv1i32(<vscale x 1 x i32> %v) {
 ; CHECK-LABEL: vreduce_add_nxv1i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
 ; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredsum.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i32 @llvm.vector.reduce.add.nxv1i32(<vscale x 1 x i32> %v)
   ret i32 %red
 }
 
 define signext i32 @vwreduce_add_nxv1i16(<vscale x 1 x i16> %v) {
 ; CHECK-LABEL: vwreduce_add_nxv1i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
-; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf4, ta, ma
 ; CHECK-NEXT:    vwredsum.vs v8, v8, v9
 ; CHECK-NEXT:    vsetivli zero, 0, e32, m1, ta, ma
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %e = sext <vscale x 1 x i16> %v to <vscale x 1 x i32>
   %red = call i32 @llvm.vector.reduce.add.nxv1i32(<vscale x 1 x i32> %e)
   ret i32 %red
 }
 
 define signext i32 @vwreduce_uadd_nxv1i16(<vscale x 1 x i16> %v) {
 ; CHECK-LABEL: vwreduce_uadd_nxv1i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
-; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, ma
+; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf4, ta, ma
 ; CHECK-NEXT:    vwredsumu.vs v8, v8, v9
 ; CHECK-NEXT:    vsetivli zero, 0, e32, m1, ta, ma
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %e = zext <vscale x 1 x i16> %v to <vscale x 1 x i32>
   %red = call i32 @llvm.vector.reduce.add.nxv1i32(<vscale x 1 x i32> %e)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.umax.nxv1i32(<vscale x 1 x i32>)
 
 define signext i32 @vreduce_umax_nxv1i32(<vscale x 1 x i32> %v) {
 ; CHECK-LABEL: vreduce_umax_nxv1i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
 ; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredmaxu.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i32 @llvm.vector.reduce.umax.nxv1i32(<vscale x 1 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.smax.nxv1i32(<vscale x 1 x i32>)
 
 define signext i32 @vreduce_smax_nxv1i32(<vscale x 1 x i32> %v) {
 ; CHECK-LABEL: vreduce_smax_nxv1i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, 524288
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vsetvli a1, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vredmax.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i32 @llvm.vector.reduce.smax.nxv1i32(<vscale x 1 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.umin.nxv1i32(<vscale x 1 x i32>)
 
 define signext i32 @vreduce_umin_nxv1i32(<vscale x 1 x i32> %v) {
 ; CHECK-LABEL: vreduce_umin_nxv1i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vredminu.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i32 @llvm.vector.reduce.umin.nxv1i32(<vscale x 1 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.smin.nxv1i32(<vscale x 1 x i32>)
 
 define signext i32 @vreduce_smin_nxv1i32(<vscale x 1 x i32> %v) {
 ; RV32-LABEL: vreduce_smin_nxv1i32:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    lui a0, 524288
 ; RV32-NEXT:    addi a0, a0, -1
-; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; RV32-NEXT:    vmv.s.x v9, a0
-; RV32-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; RV32-NEXT:    vsetvli a1, zero, e32, mf2, ta, ma
+; RV32-NEXT:    vmv.v.x v9, a0
 ; RV32-NEXT:    vredmin.vs v8, v8, v9
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_smin_nxv1i32:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    lui a0, 524288
 ; RV64-NEXT:    addiw a0, a0, -1
-; RV64-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; RV64-NEXT:    vsetvli a1, zero, e32, mf2, ta, ma
+; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vredmin.vs v8, v8, v9
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %red = call i32 @llvm.vector.reduce.smin.nxv1i32(<vscale x 1 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.and.nxv1i32(<vscale x 1 x i32>)
 
 define signext i32 @vreduce_and_nxv1i32(<vscale x 1 x i32> %v) {
 ; CHECK-LABEL: vreduce_and_nxv1i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vredand.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i32 @llvm.vector.reduce.and.nxv1i32(<vscale x 1 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.or.nxv1i32(<vscale x 1 x i32>)
 
 define signext i32 @vreduce_or_nxv1i32(<vscale x 1 x i32> %v) {
 ; CHECK-LABEL: vreduce_or_nxv1i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
 ; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredor.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i32 @llvm.vector.reduce.or.nxv1i32(<vscale x 1 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.xor.nxv1i32(<vscale x 1 x i32>)
 
 define signext i32 @vreduce_xor_nxv1i32(<vscale x 1 x i32> %v) {
 ; CHECK-LABEL: vreduce_xor_nxv1i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
 ; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredxor.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i32 @llvm.vector.reduce.xor.nxv1i32(<vscale x 1 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.add.nxv2i32(<vscale x 2 x i32>)
 
 define signext i32 @vreduce_add_nxv2i32(<vscale x 2 x i32> %v) {
 ; CHECK-LABEL: vreduce_add_nxv2i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
 ; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredsum.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i32 @llvm.vector.reduce.add.nxv2i32(<vscale x 2 x i32> %v)
   ret i32 %red
 }
 
 define signext i32 @vwreduce_add_nxv2i16(<vscale x 2 x i16> %v) {
 ; CHECK-LABEL: vwreduce_add_nxv2i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
-; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf2, ta, ma
 ; CHECK-NEXT:    vwredsum.vs v8, v8, v9
 ; CHECK-NEXT:    vsetvli zero, zero, e32, m1, ta, ma
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %e = sext <vscale x 2 x i16> %v to <vscale x 2 x i32>
   %red = call i32 @llvm.vector.reduce.add.nxv2i32(<vscale x 2 x i32> %e)
   ret i32 %red
 }
 
 define signext i32 @vwreduce_uadd_nxv2i16(<vscale x 2 x i16> %v) {
 ; CHECK-LABEL: vwreduce_uadd_nxv2i16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
-; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, ma
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
+; CHECK-NEXT:    vsetvli zero, zero, e16, mf2, ta, ma
 ; CHECK-NEXT:    vwredsumu.vs v8, v8, v9
 ; CHECK-NEXT:    vsetvli zero, zero, e32, m1, ta, ma
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %e = zext <vscale x 2 x i16> %v to <vscale x 2 x i32>
   %red = call i32 @llvm.vector.reduce.add.nxv2i32(<vscale x 2 x i32> %e)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.umax.nxv2i32(<vscale x 2 x i32>)
 
 define signext i32 @vreduce_umax_nxv2i32(<vscale x 2 x i32> %v) {
 ; CHECK-LABEL: vreduce_umax_nxv2i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
 ; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredmaxu.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i32 @llvm.vector.reduce.umax.nxv2i32(<vscale x 2 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.smax.nxv2i32(<vscale x 2 x i32>)
 
 define signext i32 @vreduce_smax_nxv2i32(<vscale x 2 x i32> %v) {
 ; CHECK-LABEL: vreduce_smax_nxv2i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, 524288
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, a0
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vsetvli a1, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vredmax.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i32 @llvm.vector.reduce.smax.nxv2i32(<vscale x 2 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.umin.nxv2i32(<vscale x 2 x i32>)
 
 define signext i32 @vreduce_umin_nxv2i32(<vscale x 2 x i32> %v) {
 ; CHECK-LABEL: vreduce_umin_nxv2i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vredminu.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i32 @llvm.vector.reduce.umin.nxv2i32(<vscale x 2 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.smin.nxv2i32(<vscale x 2 x i32>)
 
 define signext i32 @vreduce_smin_nxv2i32(<vscale x 2 x i32> %v) {
 ; RV32-LABEL: vreduce_smin_nxv2i32:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    lui a0, 524288
 ; RV32-NEXT:    addi a0, a0, -1
-; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; RV32-NEXT:    vmv.s.x v9, a0
-; RV32-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
+; RV32-NEXT:    vsetvli a1, zero, e32, m1, ta, ma
+; RV32-NEXT:    vmv.v.x v9, a0
 ; RV32-NEXT:    vredmin.vs v8, v8, v9
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_smin_nxv2i32:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    lui a0, 524288
 ; RV64-NEXT:    addiw a0, a0, -1
-; RV64-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
+; RV64-NEXT:    vsetvli a1, zero, e32, m1, ta, ma
+; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vredmin.vs v8, v8, v9
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %red = call i32 @llvm.vector.reduce.smin.nxv2i32(<vscale x 2 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.and.nxv2i32(<vscale x 2 x i32>)
 
 define signext i32 @vreduce_and_nxv2i32(<vscale x 2 x i32> %v) {
 ; CHECK-LABEL: vreduce_and_nxv2i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vredand.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i32 @llvm.vector.reduce.and.nxv2i32(<vscale x 2 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.or.nxv2i32(<vscale x 2 x i32>)
 
 define signext i32 @vreduce_or_nxv2i32(<vscale x 2 x i32> %v) {
 ; CHECK-LABEL: vreduce_or_nxv2i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
 ; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredor.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i32 @llvm.vector.reduce.or.nxv2i32(<vscale x 2 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.xor.nxv2i32(<vscale x 2 x i32>)
 
 define signext i32 @vreduce_xor_nxv2i32(<vscale x 2 x i32> %v) {
 ; CHECK-LABEL: vreduce_xor_nxv2i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, ma
-; CHECK-NEXT:    vmv.s.x v9, zero
 ; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, ma
+; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vredxor.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i32 @llvm.vector.reduce.xor.nxv2i32(<vscale x 2 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.add.nxv4i32(<vscale x 4 x i32>)
; CHECK-NEXT:    ret
   ret i32 %red
 }
 
 declare i64 @llvm.vector.reduce.add.nxv1i64(<vscale x 1 x i64>)
 
 define i64 @vreduce_add_nxv1i64(<vscale x 1 x i64> %v) {
 ; RV32-LABEL: vreduce_add_nxv1i64:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
-; RV32-NEXT:    vmv.s.x v9, zero
 ; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
+; RV32-NEXT:    vmv.v.i v9, 0
 ; RV32-NEXT:    vredsum.vs v8, v8, v9
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_add_nxv1i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, zero
 ; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
+; RV64-NEXT:    vmv.v.i v9, 0
 ; RV64-NEXT:    vredsum.vs v8, v8, v9
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %red = call i64 @llvm.vector.reduce.add.nxv1i64(<vscale x 1 x i64> %v)
   ret i64 %red
 }
 
 define i64 @vwreduce_add_nxv1i32(<vscale x 1 x i32> %v) {
 ; RV32-LABEL: vwreduce_add_nxv1i32:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
-; RV32-NEXT:    vmv.s.x v9, zero
-; RV32-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
+; RV32-NEXT:    vmv.v.i v9, 0
+; RV32-NEXT:    vsetvli zero, zero, e32, mf2, ta, ma
 ; RV32-NEXT:    vwredsum.vs v8, v8, v9
 ; RV32-NEXT:    vsetvli zero, zero, e64, m1, ta, ma
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vwreduce_add_nxv1i32:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, zero
-; RV64-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
+; RV64-NEXT:    vmv.v.i v9, 0
+; RV64-NEXT:    vsetvli zero, zero, e32, mf2, ta, ma
 ; RV64-NEXT:    vwredsum.vs v8, v8, v9
 ; RV64-NEXT:    vsetvli zero, zero, e64, m1, ta, ma
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %e = sext <vscale x 1 x i32> %v to <vscale x 1 x i64>
   %red = call i64 @llvm.vector.reduce.add.nxv1i64(<vscale x 1 x i64> %e)
   ret i64 %red
 }
 
 define i64 @vwreduce_uadd_nxv1i32(<vscale x 1 x i32> %v) {
 ; RV32-LABEL: vwreduce_uadd_nxv1i32:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
-; RV32-NEXT:    vmv.s.x v9, zero
-; RV32-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
+; RV32-NEXT:    vmv.v.i v9, 0
+; RV32-NEXT:    vsetvli zero, zero, e32, mf2, ta, ma
 ; RV32-NEXT:    vwredsumu.vs v8, v8, v9
 ; RV32-NEXT:    vsetvli zero, zero, e64, m1, ta, ma
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vwreduce_uadd_nxv1i32:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, zero
-; RV64-NEXT:    vsetvli a0, zero, e32, mf2, ta, ma
+; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
+; RV64-NEXT:    vmv.v.i v9, 0
+; RV64-NEXT:    vsetvli zero, zero, e32, mf2, ta, ma
 ; RV64-NEXT:    vwredsumu.vs v8, v8, v9
 ; RV64-NEXT:    vsetvli zero, zero, e64, m1, ta, ma
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %e = zext <vscale x 1 x i32> %v to <vscale x 1 x i64>
   %red = call i64 @llvm.vector.reduce.add.nxv1i64(<vscale x 1 x i64> %e)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.umax.nxv1i64(<vscale x 1 x i64>)
 
 define i64 @vreduce_umax_nxv1i64(<vscale x 1 x i64> %v) {
 ; RV32-LABEL: vreduce_umax_nxv1i64:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
-; RV32-NEXT:    vmv.s.x v9, zero
 ; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
+; RV32-NEXT:    vmv.v.i v9, 0
 ; RV32-NEXT:    vredmaxu.vs v8, v8, v9
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_umax_nxv1i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, zero
 ; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
+; RV64-NEXT:    vmv.v.i v9, 0
 ; RV64-NEXT:    vredmaxu.vs v8, v8, v9
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %red = call i64 @llvm.vector.reduce.umax.nxv1i64(<vscale x 1 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.smax.nxv1i64(<vscale x 1 x i64>)
 
 define i64 @vreduce_smax_nxv1i64(<vscale x 1 x i64> %v) {
 ; RV32-LABEL: vreduce_smax_nxv1i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    addi sp, sp, -16
 ; RV32-NEXT:    .cfi_def_cfa_offset 16
 ; RV32-NEXT:    lui a0, 524288
 ; RV32-NEXT:    sw a0, 12(sp)
 ; RV32-NEXT:    sw zero, 8(sp)
 ; RV32-NEXT:    addi a0, sp, 8
-; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV32-NEXT:    vsetvli a1, zero, e64, m1, ta, ma
 ; RV32-NEXT:    vlse64.v v9, (a0), zero
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
 ; RV32-NEXT:    vredmax.vs v8, v8, v9
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_smax_nxv1i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a0, -1
 ; RV64-NEXT:    slli a0, a0, 63
-; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
+; RV64-NEXT:    vsetvli a1, zero, e64, m1, ta, ma
+; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vredmax.vs v8, v8, v9
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %red = call i64 @llvm.vector.reduce.smax.nxv1i64(<vscale x 1 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.umin.nxv1i64(<vscale x 1 x i64>)
 
 define i64 @vreduce_umin_nxv1i64(<vscale x 1 x i64> %v) {
 ; RV32-LABEL: vreduce_umin_nxv1i64:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
-; RV32-NEXT:    vmv.v.i v9, -1
 ; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
+; RV32-NEXT:    vmv.v.i v9, -1
 ; RV32-NEXT:    vredminu.vs v8, v8, v9
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_umin_nxv1i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
-; RV64-NEXT:    vmv.v.i v9, -1
 ; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
+; RV64-NEXT:    vmv.v.i v9, -1
 ; RV64-NEXT:    vredminu.vs v8, v8, v9
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %red = call i64 @llvm.vector.reduce.umin.nxv1i64(<vscale x 1 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.smin.nxv1i64(<vscale x 1 x i64>)
 
 define i64 @vreduce_smin_nxv1i64(<vscale x 1 x i64> %v) {
 ; RV32-LABEL: vreduce_smin_nxv1i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    addi sp, sp, -16
 ; RV32-NEXT:    .cfi_def_cfa_offset 16
 ; RV32-NEXT:    li a0, -1
 ; RV32-NEXT:    sw a0, 8(sp)
 ; RV32-NEXT:    lui a0, 524288
 ; RV32-NEXT:    addi a0, a0, -1
 ; RV32-NEXT:    sw a0, 12(sp)
 ; RV32-NEXT:    addi a0, sp, 8
-; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
+; RV32-NEXT:    vsetvli a1, zero, e64, m1, ta, ma
 ; RV32-NEXT:    vlse64.v v9, (a0), zero
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
 ; RV32-NEXT:    vredmin.vs v8, v8, v9
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_smin_nxv1i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a0, -1
 ; RV64-NEXT:    srli a0, a0, 1
-; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, a0
-; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
+; RV64-NEXT:    vsetvli a1, zero, e64, m1, ta, ma
+; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vredmin.vs v8, v8, v9
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %red = call i64 @llvm.vector.reduce.smin.nxv1i64(<vscale x 1 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.and.nxv1i64(<vscale x 1 x i64>)
 
 define i64 @vreduce_and_nxv1i64(<vscale x 1 x i64> %v) {
 ; RV32-LABEL: vreduce_and_nxv1i64:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
-; RV32-NEXT:    vmv.v.i v9, -1
 ; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
+; RV32-NEXT:    vmv.v.i v9, -1
 ; RV32-NEXT:    vredand.vs v8, v8, v9
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_and_nxv1i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
-; RV64-NEXT:    vmv.v.i v9, -1
 ; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
+; RV64-NEXT:    vmv.v.i v9, -1
 ; RV64-NEXT:    vredand.vs v8, v8, v9
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %red = call i64 @llvm.vector.reduce.and.nxv1i64(<vscale x 1 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.or.nxv1i64(<vscale x 1 x i64>)
 
 define i64 @vreduce_or_nxv1i64(<vscale x 1 x i64> %v) {
 ; RV32-LABEL: vreduce_or_nxv1i64:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
-; RV32-NEXT:    vmv.s.x v9, zero
 ; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
+; RV32-NEXT:    vmv.v.i v9, 0
 ; RV32-NEXT:    vredor.vs v8, v8, v9
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_or_nxv1i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, zero
 ; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
+; RV64-NEXT:    vmv.v.i v9, 0
 ; RV64-NEXT:    vredor.vs v8, v8, v9
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %red = call i64 @llvm.vector.reduce.or.nxv1i64(<vscale x 1 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.xor.nxv1i64(<vscale x 1 x i64>)
 
 define i64 @vreduce_xor_nxv1i64(<vscale x 1 x i64> %v) {
 ; RV32-LABEL: vreduce_xor_nxv1i64:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
-; RV32-NEXT:    vmv.s.x v9, zero
 ; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
+; RV32-NEXT:    vmv.v.i v9, 0
 ; RV32-NEXT:    vredxor.vs v8, v8, v9
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_xor_nxv1i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, ma
-; RV64-NEXT:    vmv.s.x v9, zero
 ; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, ma
+; RV64-NEXT:    vmv.v.i v9, 0
 ; RV64-NEXT:    vredxor.vs v8, v8, v9
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %red = call i64 @llvm.vector.reduce.xor.nxv1i64(<vscale x 1 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.add.nxv2i64(<vscale x 2 x i64>)