[RISCV] Don't use VLMAX for start value splat in reduction lowering.

[craig.topper]

Summary

The reduction instructions only reads the first element. The
execution time for a splat may take longer with a larger VL.
We should use the smallest VL we can.

[HsiangKai]

LGTM.

[frasercrmck]

LGTM too but clang-format needs appeasing. Thanks, Craig!

llvm/lib/Target/RISCV/RISCVISelLowering.cpp

SDValue RISCVTargetLowering::lowerVECREDUCE(SDValue Op,
 
   MVT ContainerVT = VecVT;
   if (VecVT.isFixedLengthVector()) {
     ContainerVT = getContainerForFixedLengthVector(VecVT);
     Vec = convertToScalableVector(ContainerVT, Vec, DAG, Subtarget);
   }
 
   MVT M1VT = getLMUL1VT(ContainerVT);
+  MVT XLenVT = Subtarget.getXLenVT();
 
   SDValue Mask, VL;
   std::tie(Mask, VL) = getDefaultVLOps(VecVT, ContainerVT, DL, DAG, Subtarget);
 
-  // FIXME: This is a VLMAX splat which might be too large and can prevent
-  // vsetvli removal.
   SDValue NeutralElem =
       DAG.getNeutralElement(BaseOpc, DL, VecEltVT, SDNodeFlags());
-  SDValue IdentitySplat = DAG.getSplatVector(M1VT, DL, NeutralElem);
+  SDValue IdentitySplat = lowerScalarSplat(
+      NeutralElem, DAG.getConstant(1, DL, XLenVT), M1VT, DL, DAG, Subtarget);
   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, Subtarget.getXLenVT()));
+                             DAG.getConstant(0, DL, XLenVT));
   return DAG.getSExtOrTrunc(Elt0, DL, Op.getValueType());
 }
 
 // Given a reduction op, this function returns the matching reduction opcode,
 // the vector SDValue and the scalar SDValue required to lower this to a
 // RISCVISD node.
 static std::tuple<unsigned, SDValue, SDValue>
 getRVVFPReductionOpAndOperands(SDValue Op, SelectionDAG &DAG, EVT EltVT) {
SDValue RISCVTargetLowering::lowerFPVECREDUCE(SDValue Op,
 
   MVT ContainerVT = VecVT;
   if (VecVT.isFixedLengthVector()) {
     ContainerVT = getContainerForFixedLengthVector(VecVT);
     VectorVal = convertToScalableVector(ContainerVT, VectorVal, DAG, Subtarget);
   }
 
   MVT M1VT = getLMUL1VT(VectorVal.getSimpleValueType());
+  MVT XLenVT = Subtarget.getXLenVT();
 
   SDValue Mask, VL;
   std::tie(Mask, VL) = getDefaultVLOps(VecVT, ContainerVT, DL, DAG, Subtarget);
 
-  // FIXME: This is a VLMAX splat which might be too large and can prevent
-  // vsetvli removal.
-  SDValue ScalarSplat = DAG.getSplatVector(M1VT, DL, ScalarVal);
+  SDValue ScalarSplat = lowerScalarSplat(
+      ScalarVal, DAG.getConstant(1, DL, XLenVT), M1VT, DL, DAG, Subtarget);
   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, Subtarget.getXLenVT()));
+                     DAG.getConstant(0, DL, XLenVT));
 }
 
 static unsigned getRVVVPReductionOp(unsigned ISDOpcode) {
   switch (ISDOpcode) {
   default:
     llvm_unreachable("Unhandled reduction");
   case ISD::VP_REDUCE_ADD:
     return RISCVISD::VECREDUCE_ADD_VL;
SDValue RISCVTargetLowering::lowerVPREDUCE(SDValue Op,
 
   SDValue VL = Op.getOperand(3);
   SDValue Mask = Op.getOperand(2);
 
   MVT M1VT = getLMUL1VT(ContainerVT);
   MVT XLenVT = Subtarget.getXLenVT();
   MVT ResVT = !VecVT.isInteger() || VecEltVT.bitsGE(XLenVT) ? VecEltVT : XLenVT;
 
-  // FIXME: This is a VLMAX splat which might be too large and can prevent
-  // vsetvli removal.
-  SDValue StartSplat = DAG.getSplatVector(M1VT, DL, Op.getOperand(0));
+  SDValue StartSplat =
+      lowerScalarSplat(Op.getOperand(0), DAG.getConstant(1, DL, XLenVT), M1VT,
+                       DL, DAG, Subtarget);
   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, Subtarget.getXLenVT()));
+                             DAG.getConstant(0, DL, XLenVT));
   if (!VecVT.isInteger())
     return Elt0;
   return DAG.getSExtOrTrunc(Elt0, DL, Op.getValueType());
 }
 
 SDValue RISCVTargetLowering::lowerINSERT_SUBVECTOR(SDValue Op,
                                                    SelectionDAG &DAG) const {
   SDValue Vec = Op.getOperand(0);

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,+experimental-zfh,+experimental-v -target-abi=ilp32d -riscv-v-vector-bits-min=128 \
 ; RUN:   -verify-machineinstrs < %s | FileCheck %s
 ; RUN: llc -mtriple=riscv64 -mattr=+d,+experimental-zfh,+experimental-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:    vsetvli a1, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v9, fa0
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, tu, mu
 ; 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:    vsetvli a1, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v9, fa0
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, tu, mu
 ; 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:    vsetvli a1, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v9, fa0
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, tu, mu
 ; 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:    vsetvli a1, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v9, fa0
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, tu, mu
 ; 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:    vsetvli a1, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v9, fa0
 ; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, tu, mu
 ; 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:    vsetvli a1, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v9, fa0
 ; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, tu, mu
 ; 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:    vsetvli a1, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v9, fa0
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m1, tu, mu
 ; 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:    vsetvli a1, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v9, fa0
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m1, tu, mu
 ; 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 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:    vsetvli a1, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v9, fa0
 ; CHECK-NEXT:    vsetvli zero, a0, e64, m1, tu, mu
 ; 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:    vsetvli a1, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v9, fa0
 ; CHECK-NEXT:    vsetvli zero, a0, e64, m1, tu, mu
 ; 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.v4f64(double, <4 x double>, <4 x i1>, i32)
 
 define double @vpreduce_fadd_v4f64(double %s, <4 x double> %v, <4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_fadd_v4f64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a1, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v10, fa0
 ; CHECK-NEXT:    vsetvli zero, a0, e64, m2, tu, mu
 ; CHECK-NEXT:    vfredusum.vs v10, v8, v10, v0.t
 ; CHECK-NEXT:    vfmv.f.s fa0, v10
 ; CHECK-NEXT:    ret
   %r = call reassoc double @llvm.vp.reduce.fadd.v4f64(double %s, <4 x double> %v, <4 x i1> %m, i32 %evl)
   ret double %r
 }
 
 define double @vpreduce_ord_fadd_v4f64(double %s, <4 x double> %v, <4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_ord_fadd_v4f64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a1, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v10, fa0
 ; CHECK-NEXT:    vsetvli zero, a0, e64, m2, tu, mu
 ; CHECK-NEXT:    vfredosum.vs v10, v8, v10, v0.t
 ; CHECK-NEXT:    vfmv.f.s fa0, v10
 ; CHECK-NEXT:    ret
   %r = call double @llvm.vp.reduce.fadd.v4f64(double %s, <4 x double> %v, <4 x i1> %m, i32 %evl)
   ret double %r
 }

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

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc -mtriple=riscv32 -target-abi=ilp32d -mattr=+experimental-v,+experimental-zfh,+f,+d -riscv-v-vector-bits-min=128 -verify-machineinstrs < %s | FileCheck %s --check-prefixes=CHECK,RV32
-; RUN: llc -mtriple=riscv64 -target-abi=lp64d -mattr=+experimental-v,+experimental-zfh,+f,+d -riscv-v-vector-bits-min=128  -verify-machineinstrs < %s | FileCheck %s --check-prefixes=CHECK,RV64
+; RUN: llc -mtriple=riscv32 -target-abi=ilp32d -mattr=+experimental-v,+experimental-zfh,+f,+d -riscv-v-vector-bits-min=128 -verify-machineinstrs < %s | FileCheck %s --check-prefixes=CHECK
+; RUN: llc -mtriple=riscv64 -target-abi=lp64d -mattr=+experimental-v,+experimental-zfh,+f,+d -riscv-v-vector-bits-min=128  -verify-machineinstrs < %s | FileCheck %s --check-prefixes=CHECK
 
 declare half @llvm.vector.reduce.fadd.v1f16(half, <1 x half>)
 
 define half @vreduce_fadd_v1f16(<1 x half>* %x, half %s) {
 ; CHECK-LABEL: vreduce_fadd_v1f16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 1, e16, mf4, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
 ; CHECK-NEXT:    vfmv.f.s ft0, v8
 ; CHECK-NEXT:    fadd.h fa0, fa0, ft0
 ; CHECK-NEXT:    ret
   %v = load <1 x half>, <1 x half>* %x
   %red = call reassoc half @llvm.vector.reduce.fadd.v1f16(half %s, <1 x half> %v)
   ret half %red
 }
 
 define half @vreduce_ord_fadd_v1f16(<1 x half>* %x, half %s) {
 ; CHECK-LABEL: vreduce_ord_fadd_v1f16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 1, e16, mf4, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v9, fa0
 ; CHECK-NEXT:    vsetivli zero, 1, e16, mf4, ta, mu
 ; CHECK-NEXT:    vfredosum.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <1 x half>, <1 x half>* %x
   %red = call half @llvm.vector.reduce.fadd.v1f16(half %s, <1 x half> %v)
   ret half %red
 }
 
 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:    lui a1, %hi(.LCPI2_0)
-; CHECK-NEXT:    flh ft0, %lo(.LCPI2_0)(a1)
 ; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI2_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI2_0)
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
+; CHECK-NEXT:    vlse16.v v9, (a0), zero
 ; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, mu
 ; CHECK-NEXT:    vfredusum.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s ft0, v8
 ; CHECK-NEXT:    fadd.h fa0, fa0, ft0
 ; 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
 }
 
 define half @vreduce_ord_fadd_v2f16(<2 x half>* %x, half %s) {
 ; CHECK-LABEL: vreduce_ord_fadd_v2f16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v9, fa0
 ; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, mu
 ; CHECK-NEXT:    vfredosum.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 half @llvm.vector.reduce.fadd.v2f16(half %s, <2 x half> %v)
   ret half %red
 }
 
 declare half @llvm.vector.reduce.fadd.v4f16(half, <4 x half>)
 
 define half @vreduce_fadd_v4f16(<4 x half>* %x, half %s) {
 ; CHECK-LABEL: vreduce_fadd_v4f16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a1, %hi(.LCPI4_0)
-; CHECK-NEXT:    flh ft0, %lo(.LCPI4_0)(a1)
 ; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI4_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI4_0)
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
+; CHECK-NEXT:    vlse16.v v9, (a0), zero
 ; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
 ; CHECK-NEXT:    vfredusum.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s ft0, v8
 ; CHECK-NEXT:    fadd.h fa0, fa0, ft0
 ; CHECK-NEXT:    ret
   %v = load <4 x half>, <4 x half>* %x
   %red = call reassoc half @llvm.vector.reduce.fadd.v4f16(half %s, <4 x half> %v)
   ret half %red
 }
 
 define half @vreduce_ord_fadd_v4f16(<4 x half>* %x, half %s) {
 ; CHECK-LABEL: vreduce_ord_fadd_v4f16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v9, fa0
 ; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
 ; CHECK-NEXT:    vfredosum.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x half>, <4 x half>* %x
   %red = call half @llvm.vector.reduce.fadd.v4f16(half %s, <4 x half> %v)
   ret half %red
 }
 
 declare half @llvm.vector.reduce.fadd.v8f16(half, <8 x half>)
 
 define half @vreduce_fadd_v8f16(<8 x half>* %x, half %s) {
 ; CHECK-LABEL: vreduce_fadd_v8f16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a1, %hi(.LCPI6_0)
-; CHECK-NEXT:    flh ft0, %lo(.LCPI6_0)(a1)
 ; CHECK-NEXT:    vsetivli zero, 8, e16, m1, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI6_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI6_0)
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
+; CHECK-NEXT:    vlse16.v v9, (a0), zero
 ; CHECK-NEXT:    vsetivli zero, 8, e16, m1, ta, mu
 ; CHECK-NEXT:    vfredusum.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s ft0, v8
 ; CHECK-NEXT:    fadd.h fa0, fa0, ft0
 ; CHECK-NEXT:    ret
   %v = load <8 x half>, <8 x half>* %x
   %red = call reassoc half @llvm.vector.reduce.fadd.v8f16(half %s, <8 x half> %v)
   ret half %red
 }
 
 define half @vreduce_ord_fadd_v8f16(<8 x half>* %x, half %s) {
 ; CHECK-LABEL: vreduce_ord_fadd_v8f16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 8, e16, m1, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v9, fa0
 ; CHECK-NEXT:    vsetivli zero, 8, e16, m1, ta, mu
 ; CHECK-NEXT:    vfredosum.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <8 x half>, <8 x half>* %x
   %red = call half @llvm.vector.reduce.fadd.v8f16(half %s, <8 x half> %v)
   ret half %red
 }
 
 declare half @llvm.vector.reduce.fadd.v16f16(half, <16 x half>)
 
 define half @vreduce_fadd_v16f16(<16 x half>* %x, half %s) {
 ; CHECK-LABEL: vreduce_fadd_v16f16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a1, %hi(.LCPI8_0)
-; CHECK-NEXT:    flh ft0, %lo(.LCPI8_0)(a1)
 ; CHECK-NEXT:    vsetivli zero, 16, e16, m2, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v10, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI8_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI8_0)
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
+; CHECK-NEXT:    vlse16.v v10, (a0), zero
 ; CHECK-NEXT:    vsetivli zero, 16, e16, m2, ta, mu
 ; CHECK-NEXT:    vfredusum.vs v8, v8, v10
 ; CHECK-NEXT:    vfmv.f.s ft0, v8
 ; CHECK-NEXT:    fadd.h fa0, fa0, ft0
 ; CHECK-NEXT:    ret
   %v = load <16 x half>, <16 x half>* %x
   %red = call reassoc half @llvm.vector.reduce.fadd.v16f16(half %s, <16 x half> %v)
   ret half %red
 }
 
 define half @vreduce_ord_fadd_v16f16(<16 x half>* %x, half %s) {
 ; CHECK-LABEL: vreduce_ord_fadd_v16f16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 16, e16, m2, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v10, fa0
 ; CHECK-NEXT:    vsetivli zero, 16, e16, m2, ta, mu
 ; CHECK-NEXT:    vfredosum.vs v8, v8, v10
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <16 x half>, <16 x half>* %x
   %red = call half @llvm.vector.reduce.fadd.v16f16(half %s, <16 x half> %v)
   ret half %red
 }
 
 declare half @llvm.vector.reduce.fadd.v32f16(half, <32 x half>)
 
 define half @vreduce_fadd_v32f16(<32 x half>* %x, half %s) {
-; RV32-LABEL: vreduce_fadd_v32f16:
-; RV32:       # %bb.0:
-; RV32-NEXT:    li a1, 32
-; RV32-NEXT:    lui a2, %hi(.LCPI10_0)
-; RV32-NEXT:    flh ft0, %lo(.LCPI10_0)(a2)
-; RV32-NEXT:    vsetvli zero, a1, e16, m4, ta, mu
-; RV32-NEXT:    vle16.v v8, (a0)
-; RV32-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
-; RV32-NEXT:    vfmv.v.f v12, ft0
-; RV32-NEXT:    vsetvli zero, a1, e16, m4, ta, mu
-; RV32-NEXT:    vfredusum.vs v8, v8, v12
-; RV32-NEXT:    vfmv.f.s ft0, v8
-; RV32-NEXT:    fadd.h fa0, fa0, ft0
-; RV32-NEXT:    ret
-;
-; RV64-LABEL: vreduce_fadd_v32f16:
-; RV64:       # %bb.0:
-; RV64-NEXT:    lui a1, %hi(.LCPI10_0)
-; RV64-NEXT:    flh ft0, %lo(.LCPI10_0)(a1)
-; RV64-NEXT:    li a1, 32
-; RV64-NEXT:    vsetvli zero, a1, e16, m4, ta, mu
-; RV64-NEXT:    vle16.v v8, (a0)
-; RV64-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
-; RV64-NEXT:    vfmv.v.f v12, ft0
-; RV64-NEXT:    vsetvli zero, a1, e16, m4, ta, mu
-; RV64-NEXT:    vfredusum.vs v8, v8, v12
-; RV64-NEXT:    vfmv.f.s ft0, v8
-; RV64-NEXT:    fadd.h fa0, fa0, ft0
-; RV64-NEXT:    ret
+; CHECK-LABEL: vreduce_fadd_v32f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    li a1, 32
+; CHECK-NEXT:    vsetvli zero, a1, e16, m4, ta, mu
+; CHECK-NEXT:    vle16.v v8, (a0)
+; CHECK-NEXT:    lui a0, %hi(.LCPI10_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI10_0)
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
+; CHECK-NEXT:    vlse16.v v12, (a0), zero
+; CHECK-NEXT:    vsetvli zero, a1, e16, m4, ta, mu
+; CHECK-NEXT:    vfredusum.vs v8, v8, v12
+; CHECK-NEXT:    vfmv.f.s ft0, v8
+; CHECK-NEXT:    fadd.h fa0, fa0, ft0
+; CHECK-NEXT:    ret
   %v = load <32 x half>, <32 x half>* %x
   %red = call reassoc half @llvm.vector.reduce.fadd.v32f16(half %s, <32 x half> %v)
   ret half %red
 }
 
 define half @vreduce_ord_fadd_v32f16(<32 x half>* %x, half %s) {
 ; CHECK-LABEL: vreduce_ord_fadd_v32f16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m4, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v12, fa0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m4, ta, mu
 ; CHECK-NEXT:    vfredosum.vs v8, v8, v12
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <32 x half>, <32 x half>* %x
   %red = call half @llvm.vector.reduce.fadd.v32f16(half %s, <32 x half> %v)
   ret half %red
 }
 
 declare half @llvm.vector.reduce.fadd.v64f16(half, <64 x half>)
 
 define half @vreduce_fadd_v64f16(<64 x half>* %x, half %s) {
-; RV32-LABEL: vreduce_fadd_v64f16:
-; RV32:       # %bb.0:
-; RV32-NEXT:    li a1, 64
-; RV32-NEXT:    lui a2, %hi(.LCPI12_0)
-; RV32-NEXT:    flh ft0, %lo(.LCPI12_0)(a2)
-; RV32-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
-; RV32-NEXT:    vle16.v v8, (a0)
-; RV32-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
-; RV32-NEXT:    vfmv.v.f v16, ft0
-; RV32-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
-; RV32-NEXT:    vfredusum.vs v8, v8, v16
-; RV32-NEXT:    vfmv.f.s ft0, v8
-; RV32-NEXT:    fadd.h fa0, fa0, ft0
-; RV32-NEXT:    ret
-;
-; RV64-LABEL: vreduce_fadd_v64f16:
-; RV64:       # %bb.0:
-; RV64-NEXT:    lui a1, %hi(.LCPI12_0)
-; RV64-NEXT:    flh ft0, %lo(.LCPI12_0)(a1)
-; RV64-NEXT:    li a1, 64
-; RV64-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
-; RV64-NEXT:    vle16.v v8, (a0)
-; RV64-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
-; RV64-NEXT:    vfmv.v.f v16, ft0
-; RV64-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
-; RV64-NEXT:    vfredusum.vs v8, v8, v16
-; RV64-NEXT:    vfmv.f.s ft0, v8
-; RV64-NEXT:    fadd.h fa0, fa0, ft0
-; RV64-NEXT:    ret
+; CHECK-LABEL: vreduce_fadd_v64f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    li a1, 64
+; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
+; CHECK-NEXT:    vle16.v v8, (a0)
+; CHECK-NEXT:    lui a0, %hi(.LCPI12_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI12_0)
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
+; CHECK-NEXT:    vlse16.v v16, (a0), zero
+; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
+; CHECK-NEXT:    vfredusum.vs v8, v8, v16
+; CHECK-NEXT:    vfmv.f.s ft0, v8
+; CHECK-NEXT:    fadd.h fa0, fa0, ft0
+; CHECK-NEXT:    ret
   %v = load <64 x half>, <64 x half>* %x
   %red = call reassoc half @llvm.vector.reduce.fadd.v64f16(half %s, <64 x half> %v)
   ret half %red
 }
 
 define half @vreduce_ord_fadd_v64f16(<64 x half>* %x, half %s) {
 ; CHECK-LABEL: vreduce_ord_fadd_v64f16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 64
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v16, fa0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; CHECK-NEXT:    vfredosum.vs v8, v8, v16
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <64 x half>, <64 x half>* %x
   %red = call half @llvm.vector.reduce.fadd.v64f16(half %s, <64 x half> %v)
   ret half %red
 }
 
 declare half @llvm.vector.reduce.fadd.v128f16(half, <128 x half>)
 
 define half @vreduce_fadd_v128f16(<128 x half>* %x, half %s) {
 ; CHECK-LABEL: vreduce_fadd_v128f16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 64
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
 ; CHECK-NEXT:    addi a0, a0, 128
 ; CHECK-NEXT:    vle16.v v16, (a0)
-; CHECK-NEXT:    lui a0, %hi(.LCPI14_0)
-; CHECK-NEXT:    flh ft0, %lo(.LCPI14_0)(a0)
 ; CHECK-NEXT:    vfadd.vv v8, v8, v16
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v16, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI14_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI14_0)
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
+; CHECK-NEXT:    vlse16.v v16, (a0), zero
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; CHECK-NEXT:    vfredusum.vs v8, v8, v16
 ; CHECK-NEXT:    vfmv.f.s ft0, v8
 ; CHECK-NEXT:    fadd.h fa0, fa0, ft0
 ; CHECK-NEXT:    ret
   %v = load <128 x half>, <128 x half>* %x
   %red = call reassoc half @llvm.vector.reduce.fadd.v128f16(half %s, <128 x half> %v)
   ret half %red
 }
 
 define half @vreduce_ord_fadd_v128f16(<128 x half>* %x, half %s) {
 ; CHECK-LABEL: vreduce_ord_fadd_v128f16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    addi a1, a0, 128
 ; CHECK-NEXT:    li a2, 64
 ; CHECK-NEXT:    vsetvli zero, a2, e16, m8, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a1)
 ; CHECK-NEXT:    vle16.v v16, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v24, fa0
 ; CHECK-NEXT:    vsetvli zero, a2, e16, m8, ta, mu
 ; CHECK-NEXT:    vfredosum.vs v16, v16, v24
 ; CHECK-NEXT:    vfmv.f.s ft0, v16
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v16, ft0
 ; CHECK-NEXT:    vsetvli zero, a2, e16, m8, ta, mu
 ; CHECK-NEXT:    vfredosum.vs v8, v8, v16
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <128 x half>, <128 x half>* %x
   %red = call half @llvm.vector.reduce.fadd.v128f16(half %s, <128 x half> %v)
   ret half %red
; CHECK-NEXT:    ret
   ret float %red
 }
 
 define float @vreduce_ord_fadd_v1f32(<1 x float>* %x, float %s) {
 ; CHECK-LABEL: vreduce_ord_fadd_v1f32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 1, e32, mf2, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v9, fa0
 ; CHECK-NEXT:    vsetivli zero, 1, e32, mf2, ta, mu
 ; CHECK-NEXT:    vfredosum.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <1 x float>, <1 x float>* %x
   %red = call float @llvm.vector.reduce.fadd.v1f32(float %s, <1 x float> %v)
   ret float %red
 }
 
 declare float @llvm.vector.reduce.fadd.v2f32(float, <2 x float>)
 
 define float @vreduce_fadd_v2f32(<2 x float>* %x, float %s) {
 ; CHECK-LABEL: vreduce_fadd_v2f32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a1, %hi(.LCPI18_0)
-; CHECK-NEXT:    flw ft0, %lo(.LCPI18_0)(a1)
 ; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI18_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI18_0)
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
+; CHECK-NEXT:    vlse32.v v9, (a0), zero
 ; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
 ; CHECK-NEXT:    vfredusum.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s ft0, v8
 ; CHECK-NEXT:    fadd.s fa0, fa0, ft0
 ; CHECK-NEXT:    ret
   %v = load <2 x float>, <2 x float>* %x
   %red = call reassoc float @llvm.vector.reduce.fadd.v2f32(float %s, <2 x float> %v)
   ret float %red
 }
 
 define float @vreduce_ord_fadd_v2f32(<2 x float>* %x, float %s) {
 ; CHECK-LABEL: vreduce_ord_fadd_v2f32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v9, fa0
 ; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
 ; CHECK-NEXT:    vfredosum.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <2 x float>, <2 x float>* %x
   %red = call float @llvm.vector.reduce.fadd.v2f32(float %s, <2 x float> %v)
   ret float %red
 }
 
 declare float @llvm.vector.reduce.fadd.v4f32(float, <4 x float>)
 
 define float @vreduce_fadd_v4f32(<4 x float>* %x, float %s) {
 ; CHECK-LABEL: vreduce_fadd_v4f32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a1, %hi(.LCPI20_0)
-; CHECK-NEXT:    flw ft0, %lo(.LCPI20_0)(a1)
 ; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI20_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI20_0)
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
+; CHECK-NEXT:    vlse32.v v9, (a0), zero
 ; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
 ; CHECK-NEXT:    vfredusum.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s ft0, v8
 ; CHECK-NEXT:    fadd.s fa0, fa0, ft0
 ; CHECK-NEXT:    ret
   %v = load <4 x float>, <4 x float>* %x
   %red = call reassoc float @llvm.vector.reduce.fadd.v4f32(float %s, <4 x float> %v)
   ret float %red
 }
 
 define float @vreduce_ord_fadd_v4f32(<4 x float>* %x, float %s) {
 ; CHECK-LABEL: vreduce_ord_fadd_v4f32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v9, fa0
 ; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
 ; CHECK-NEXT:    vfredosum.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x float>, <4 x float>* %x
   %red = call float @llvm.vector.reduce.fadd.v4f32(float %s, <4 x float> %v)
   ret float %red
 }
 
 declare float @llvm.vector.reduce.fadd.v8f32(float, <8 x float>)
 
 define float @vreduce_fadd_v8f32(<8 x float>* %x, float %s) {
 ; CHECK-LABEL: vreduce_fadd_v8f32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a1, %hi(.LCPI22_0)
-; CHECK-NEXT:    flw ft0, %lo(.LCPI22_0)(a1)
 ; CHECK-NEXT:    vsetivli zero, 8, e32, m2, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v10, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI22_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI22_0)
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
+; CHECK-NEXT:    vlse32.v v10, (a0), zero
 ; CHECK-NEXT:    vsetivli zero, 8, e32, m2, ta, mu
 ; CHECK-NEXT:    vfredusum.vs v8, v8, v10
 ; CHECK-NEXT:    vfmv.f.s ft0, v8
 ; CHECK-NEXT:    fadd.s fa0, fa0, ft0
 ; CHECK-NEXT:    ret
   %v = load <8 x float>, <8 x float>* %x
   %red = call reassoc float @llvm.vector.reduce.fadd.v8f32(float %s, <8 x float> %v)
   ret float %red
 }
 
 define float @vreduce_ord_fadd_v8f32(<8 x float>* %x, float %s) {
 ; CHECK-LABEL: vreduce_ord_fadd_v8f32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 8, e32, m2, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v10, fa0
 ; CHECK-NEXT:    vsetivli zero, 8, e32, m2, ta, mu
 ; CHECK-NEXT:    vfredosum.vs v8, v8, v10
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <8 x float>, <8 x float>* %x
   %red = call float @llvm.vector.reduce.fadd.v8f32(float %s, <8 x float> %v)
   ret float %red
 }
 
 declare float @llvm.vector.reduce.fadd.v16f32(float, <16 x float>)
 
 define float @vreduce_fadd_v16f32(<16 x float>* %x, float %s) {
 ; CHECK-LABEL: vreduce_fadd_v16f32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a1, %hi(.LCPI24_0)
-; CHECK-NEXT:    flw ft0, %lo(.LCPI24_0)(a1)
 ; CHECK-NEXT:    vsetivli zero, 16, e32, m4, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v12, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI24_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI24_0)
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
+; CHECK-NEXT:    vlse32.v v12, (a0), zero
 ; CHECK-NEXT:    vsetivli zero, 16, e32, m4, ta, mu
 ; CHECK-NEXT:    vfredusum.vs v8, v8, v12
 ; CHECK-NEXT:    vfmv.f.s ft0, v8
 ; CHECK-NEXT:    fadd.s fa0, fa0, ft0
 ; CHECK-NEXT:    ret
   %v = load <16 x float>, <16 x float>* %x
   %red = call reassoc float @llvm.vector.reduce.fadd.v16f32(float %s, <16 x float> %v)
   ret float %red
 }
 
 define float @vreduce_ord_fadd_v16f32(<16 x float>* %x, float %s) {
 ; CHECK-LABEL: vreduce_ord_fadd_v16f32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 16, e32, m4, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v12, fa0
 ; CHECK-NEXT:    vsetivli zero, 16, e32, m4, ta, mu
 ; CHECK-NEXT:    vfredosum.vs v8, v8, v12
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <16 x float>, <16 x float>* %x
   %red = call float @llvm.vector.reduce.fadd.v16f32(float %s, <16 x float> %v)
   ret float %red
 }
 
 declare float @llvm.vector.reduce.fadd.v32f32(float, <32 x float>)
 
 define float @vreduce_fadd_v32f32(<32 x float>* %x, float %s) {
-; RV32-LABEL: vreduce_fadd_v32f32:
-; RV32:       # %bb.0:
-; RV32-NEXT:    li a1, 32
-; RV32-NEXT:    lui a2, %hi(.LCPI26_0)
-; RV32-NEXT:    flw ft0, %lo(.LCPI26_0)(a2)
-; RV32-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
-; RV32-NEXT:    vle32.v v8, (a0)
-; RV32-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
-; RV32-NEXT:    vfmv.v.f v16, ft0
-; RV32-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
-; RV32-NEXT:    vfredusum.vs v8, v8, v16
-; RV32-NEXT:    vfmv.f.s ft0, v8
-; RV32-NEXT:    fadd.s fa0, fa0, ft0
-; RV32-NEXT:    ret
-;
-; RV64-LABEL: vreduce_fadd_v32f32:
-; RV64:       # %bb.0:
-; RV64-NEXT:    lui a1, %hi(.LCPI26_0)
-; RV64-NEXT:    flw ft0, %lo(.LCPI26_0)(a1)
-; RV64-NEXT:    li a1, 32
-; RV64-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
-; RV64-NEXT:    vle32.v v8, (a0)
-; RV64-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
-; RV64-NEXT:    vfmv.v.f v16, ft0
-; RV64-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
-; RV64-NEXT:    vfredusum.vs v8, v8, v16
-; RV64-NEXT:    vfmv.f.s ft0, v8
-; RV64-NEXT:    fadd.s fa0, fa0, ft0
-; RV64-NEXT:    ret
+; CHECK-LABEL: vreduce_fadd_v32f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    li a1, 32
+; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
+; CHECK-NEXT:    vle32.v v8, (a0)
+; CHECK-NEXT:    lui a0, %hi(.LCPI26_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI26_0)
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
+; CHECK-NEXT:    vlse32.v v16, (a0), zero
+; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
+; CHECK-NEXT:    vfredusum.vs v8, v8, v16
+; CHECK-NEXT:    vfmv.f.s ft0, v8
+; CHECK-NEXT:    fadd.s fa0, fa0, ft0
+; CHECK-NEXT:    ret
   %v = load <32 x float>, <32 x float>* %x
   %red = call reassoc float @llvm.vector.reduce.fadd.v32f32(float %s, <32 x float> %v)
   ret float %red
 }
 
 define float @vreduce_ord_fadd_v32f32(<32 x float>* %x, float %s) {
 ; CHECK-LABEL: vreduce_ord_fadd_v32f32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v16, fa0
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
 ; CHECK-NEXT:    vfredosum.vs v8, v8, v16
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <32 x float>, <32 x float>* %x
   %red = call float @llvm.vector.reduce.fadd.v32f32(float %s, <32 x float> %v)
   ret float %red
 }
 
 declare float @llvm.vector.reduce.fadd.v64f32(float, <64 x float>)
 
 define float @vreduce_fadd_v64f32(<64 x float>* %x, float %s) {
 ; CHECK-LABEL: vreduce_fadd_v64f32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
 ; CHECK-NEXT:    addi a0, a0, 128
 ; CHECK-NEXT:    vle32.v v16, (a0)
-; CHECK-NEXT:    lui a0, %hi(.LCPI28_0)
-; CHECK-NEXT:    flw ft0, %lo(.LCPI28_0)(a0)
 ; CHECK-NEXT:    vfadd.vv v8, v8, v16
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v16, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI28_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI28_0)
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
+; CHECK-NEXT:    vlse32.v v16, (a0), zero
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
 ; CHECK-NEXT:    vfredusum.vs v8, v8, v16
 ; CHECK-NEXT:    vfmv.f.s ft0, v8
 ; CHECK-NEXT:    fadd.s fa0, fa0, ft0
 ; CHECK-NEXT:    ret
   %v = load <64 x float>, <64 x float>* %x
   %red = call reassoc float @llvm.vector.reduce.fadd.v64f32(float %s, <64 x float> %v)
   ret float %red
 }
 
 define float @vreduce_ord_fadd_v64f32(<64 x float>* %x, float %s) {
 ; CHECK-LABEL: vreduce_ord_fadd_v64f32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    addi a1, a0, 128
 ; CHECK-NEXT:    li a2, 32
 ; CHECK-NEXT:    vsetvli zero, a2, e32, m8, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a1)
 ; CHECK-NEXT:    vle32.v v16, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v24, fa0
 ; CHECK-NEXT:    vsetvli zero, a2, e32, m8, ta, mu
 ; CHECK-NEXT:    vfredosum.vs v16, v16, v24
 ; CHECK-NEXT:    vfmv.f.s ft0, v16
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v16, ft0
 ; CHECK-NEXT:    vsetvli zero, a2, e32, m8, ta, mu
 ; CHECK-NEXT:    vfredosum.vs v8, v8, v16
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <64 x float>, <64 x float>* %x
   %red = call float @llvm.vector.reduce.fadd.v64f32(float %s, <64 x float> %v)
   ret float %red
; CHECK-NEXT:    ret
   ret double %red
 }
 
 define double @vreduce_ord_fadd_v1f64(<1 x double>* %x, double %s) {
 ; CHECK-LABEL: vreduce_ord_fadd_v1f64:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vle64.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v9, fa0
-; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vfredosum.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <1 x double>, <1 x double>* %x
   %red = call double @llvm.vector.reduce.fadd.v1f64(double %s, <1 x double> %v)
   ret double %red
 }
 
 declare double @llvm.vector.reduce.fadd.v2f64(double, <2 x double>)
 
 define double @vreduce_fadd_v2f64(<2 x double>* %x, double %s) {
 ; CHECK-LABEL: vreduce_fadd_v2f64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a1, %hi(.LCPI32_0)
-; CHECK-NEXT:    fld ft0, %lo(.LCPI32_0)(a1)
 ; CHECK-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
 ; CHECK-NEXT:    vle64.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI32_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI32_0)
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
+; CHECK-NEXT:    vlse64.v v9, (a0), zero
 ; CHECK-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
 ; CHECK-NEXT:    vfredusum.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s ft0, v8
 ; CHECK-NEXT:    fadd.d fa0, fa0, ft0
 ; CHECK-NEXT:    ret
   %v = load <2 x double>, <2 x double>* %x
   %red = call reassoc double @llvm.vector.reduce.fadd.v2f64(double %s, <2 x double> %v)
   ret double %red
 }
 
 define double @vreduce_ord_fadd_v2f64(<2 x double>* %x, double %s) {
 ; CHECK-LABEL: vreduce_ord_fadd_v2f64:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
 ; CHECK-NEXT:    vle64.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v9, fa0
 ; CHECK-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
 ; CHECK-NEXT:    vfredosum.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <2 x double>, <2 x double>* %x
   %red = call double @llvm.vector.reduce.fadd.v2f64(double %s, <2 x double> %v)
   ret double %red
 }
 
 declare double @llvm.vector.reduce.fadd.v4f64(double, <4 x double>)
 
 define double @vreduce_fadd_v4f64(<4 x double>* %x, double %s) {
 ; CHECK-LABEL: vreduce_fadd_v4f64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a1, %hi(.LCPI34_0)
-; CHECK-NEXT:    fld ft0, %lo(.LCPI34_0)(a1)
 ; CHECK-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; CHECK-NEXT:    vle64.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v10, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI34_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI34_0)
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
+; CHECK-NEXT:    vlse64.v v10, (a0), zero
 ; CHECK-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; CHECK-NEXT:    vfredusum.vs v8, v8, v10
 ; CHECK-NEXT:    vfmv.f.s ft0, v8
 ; CHECK-NEXT:    fadd.d fa0, fa0, ft0
 ; CHECK-NEXT:    ret
   %v = load <4 x double>, <4 x double>* %x
   %red = call reassoc double @llvm.vector.reduce.fadd.v4f64(double %s, <4 x double> %v)
   ret double %red
 }
 
 define double @vreduce_ord_fadd_v4f64(<4 x double>* %x, double %s) {
 ; CHECK-LABEL: vreduce_ord_fadd_v4f64:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; CHECK-NEXT:    vle64.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v10, fa0
 ; CHECK-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; CHECK-NEXT:    vfredosum.vs v8, v8, v10
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x double>, <4 x double>* %x
   %red = call double @llvm.vector.reduce.fadd.v4f64(double %s, <4 x double> %v)
   ret double %red
 }
 
 declare double @llvm.vector.reduce.fadd.v8f64(double, <8 x double>)
 
 define double @vreduce_fadd_v8f64(<8 x double>* %x, double %s) {
 ; CHECK-LABEL: vreduce_fadd_v8f64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a1, %hi(.LCPI36_0)
-; CHECK-NEXT:    fld ft0, %lo(.LCPI36_0)(a1)
 ; CHECK-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
 ; CHECK-NEXT:    vle64.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v12, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI36_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI36_0)
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
+; CHECK-NEXT:    vlse64.v v12, (a0), zero
 ; CHECK-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
 ; CHECK-NEXT:    vfredusum.vs v8, v8, v12
 ; CHECK-NEXT:    vfmv.f.s ft0, v8
 ; CHECK-NEXT:    fadd.d fa0, fa0, ft0
 ; CHECK-NEXT:    ret
   %v = load <8 x double>, <8 x double>* %x
   %red = call reassoc double @llvm.vector.reduce.fadd.v8f64(double %s, <8 x double> %v)
   ret double %red
 }
 
 define double @vreduce_ord_fadd_v8f64(<8 x double>* %x, double %s) {
 ; CHECK-LABEL: vreduce_ord_fadd_v8f64:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
 ; CHECK-NEXT:    vle64.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v12, fa0
 ; CHECK-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
 ; CHECK-NEXT:    vfredosum.vs v8, v8, v12
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <8 x double>, <8 x double>* %x
   %red = call double @llvm.vector.reduce.fadd.v8f64(double %s, <8 x double> %v)
   ret double %red
 }
 
 declare double @llvm.vector.reduce.fadd.v16f64(double, <16 x double>)
 
 define double @vreduce_fadd_v16f64(<16 x double>* %x, double %s) {
 ; CHECK-LABEL: vreduce_fadd_v16f64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a1, %hi(.LCPI38_0)
-; CHECK-NEXT:    fld ft0, %lo(.LCPI38_0)(a1)
 ; CHECK-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; CHECK-NEXT:    vle64.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v16, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI38_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI38_0)
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
+; CHECK-NEXT:    vlse64.v v16, (a0), zero
 ; CHECK-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; CHECK-NEXT:    vfredusum.vs v8, v8, v16
 ; CHECK-NEXT:    vfmv.f.s ft0, v8
 ; CHECK-NEXT:    fadd.d fa0, fa0, ft0
 ; CHECK-NEXT:    ret
   %v = load <16 x double>, <16 x double>* %x
   %red = call reassoc double @llvm.vector.reduce.fadd.v16f64(double %s, <16 x double> %v)
   ret double %red
 }
 
 define double @vreduce_ord_fadd_v16f64(<16 x double>* %x, double %s) {
 ; CHECK-LABEL: vreduce_ord_fadd_v16f64:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; CHECK-NEXT:    vle64.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v16, fa0
 ; CHECK-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; CHECK-NEXT:    vfredosum.vs v8, v8, v16
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <16 x double>, <16 x double>* %x
   %red = call double @llvm.vector.reduce.fadd.v16f64(double %s, <16 x double> %v)
   ret double %red
 }
 
 declare double @llvm.vector.reduce.fadd.v32f64(double, <32 x double>)
 
 define double @vreduce_fadd_v32f64(<32 x double>* %x, double %s) {
 ; CHECK-LABEL: vreduce_fadd_v32f64:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; CHECK-NEXT:    vle64.v v8, (a0)
 ; CHECK-NEXT:    addi a0, a0, 128
 ; CHECK-NEXT:    vle64.v v16, (a0)
-; CHECK-NEXT:    lui a0, %hi(.LCPI40_0)
-; CHECK-NEXT:    fld ft0, %lo(.LCPI40_0)(a0)
 ; CHECK-NEXT:    vfadd.vv v8, v8, v16
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v16, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI40_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI40_0)
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
+; CHECK-NEXT:    vlse64.v v16, (a0), zero
 ; CHECK-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; CHECK-NEXT:    vfredusum.vs v8, v8, v16
 ; CHECK-NEXT:    vfmv.f.s ft0, v8
 ; CHECK-NEXT:    fadd.d fa0, fa0, ft0
 ; CHECK-NEXT:    ret
   %v = load <32 x double>, <32 x double>* %x
   %red = call reassoc double @llvm.vector.reduce.fadd.v32f64(double %s, <32 x double> %v)
   ret double %red
 }
 
 define double @vreduce_ord_fadd_v32f64(<32 x double>* %x, double %s) {
 ; CHECK-LABEL: vreduce_ord_fadd_v32f64:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    addi a1, a0, 128
 ; CHECK-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; CHECK-NEXT:    vle64.v v8, (a1)
 ; CHECK-NEXT:    vle64.v v16, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v24, fa0
 ; CHECK-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; CHECK-NEXT:    vfredosum.vs v16, v16, v24
 ; CHECK-NEXT:    vfmv.f.s ft0, v16
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v16, ft0
 ; CHECK-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; CHECK-NEXT:    vfredosum.vs v8, v8, v16
 ; 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
 }
 
 declare half @llvm.vector.reduce.fmin.v2f16(<2 x half>)
 
 define half @vreduce_fmin_v2f16(<2 x half>* %x) {
 ; CHECK-LABEL: vreduce_fmin_v2f16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a1, %hi(.LCPI42_0)
-; CHECK-NEXT:    flh ft0, %lo(.LCPI42_0)(a1)
 ; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI42_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI42_0)
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
+; CHECK-NEXT:    vlse16.v v9, (a0), zero
 ; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, mu
 ; CHECK-NEXT:    vfredmin.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 half @llvm.vector.reduce.fmin.v2f16(<2 x half> %v)
   ret half %red
 }
 
 declare half @llvm.vector.reduce.fmin.v4f16(<4 x half>)
 
 define half @vreduce_fmin_v4f16(<4 x half>* %x) {
 ; CHECK-LABEL: vreduce_fmin_v4f16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a1, %hi(.LCPI43_0)
-; CHECK-NEXT:    flh ft0, %lo(.LCPI43_0)(a1)
 ; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI43_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI43_0)
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
+; CHECK-NEXT:    vlse16.v v9, (a0), zero
 ; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
 ; CHECK-NEXT:    vfredmin.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x half>, <4 x half>* %x
   %red = call half @llvm.vector.reduce.fmin.v4f16(<4 x half> %v)
   ret half %red
 }
 
 define half @vreduce_fmin_v4f16_nonans(<4 x half>* %x) {
 ; CHECK-LABEL: vreduce_fmin_v4f16_nonans:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a1, %hi(.LCPI44_0)
-; CHECK-NEXT:    flh ft0, %lo(.LCPI44_0)(a1)
 ; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI44_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI44_0)
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
+; CHECK-NEXT:    vlse16.v v9, (a0), zero
 ; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
 ; CHECK-NEXT:    vfredmin.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x half>, <4 x half>* %x
   %red = call nnan half @llvm.vector.reduce.fmin.v4f16(<4 x half> %v)
   ret half %red
 }
 
 define half @vreduce_fmin_v4f16_nonans_noinfs(<4 x half>* %x) {
 ; CHECK-LABEL: vreduce_fmin_v4f16_nonans_noinfs:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a1, %hi(.LCPI45_0)
-; CHECK-NEXT:    flh ft0, %lo(.LCPI45_0)(a1)
 ; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI45_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI45_0)
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
+; CHECK-NEXT:    vlse16.v v9, (a0), zero
 ; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
 ; CHECK-NEXT:    vfredmin.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x half>, <4 x half>* %x
   %red = call nnan ninf half @llvm.vector.reduce.fmin.v4f16(<4 x half> %v)
   ret half %red
 }
 
 declare half @llvm.vector.reduce.fmin.v128f16(<128 x half>)
 
 define half @vreduce_fmin_v128f16(<128 x half>* %x) {
 ; CHECK-LABEL: vreduce_fmin_v128f16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 64
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
 ; CHECK-NEXT:    addi a0, a0, 128
 ; CHECK-NEXT:    vle16.v v16, (a0)
-; CHECK-NEXT:    lui a0, %hi(.LCPI46_0)
-; CHECK-NEXT:    flh ft0, %lo(.LCPI46_0)(a0)
 ; CHECK-NEXT:    vfmin.vv v8, v8, v16
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v16, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI46_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI46_0)
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
+; CHECK-NEXT:    vlse16.v v16, (a0), zero
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; CHECK-NEXT:    vfredmin.vs v8, v8, v16
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <128 x half>, <128 x half>* %x
   %red = call half @llvm.vector.reduce.fmin.v128f16(<128 x half> %v)
   ret half %red
 }
 
 declare float @llvm.vector.reduce.fmin.v2f32(<2 x float>)
 
 define float @vreduce_fmin_v2f32(<2 x float>* %x) {
 ; CHECK-LABEL: vreduce_fmin_v2f32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a1, %hi(.LCPI47_0)
-; CHECK-NEXT:    flw ft0, %lo(.LCPI47_0)(a1)
 ; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI47_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI47_0)
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
+; CHECK-NEXT:    vlse32.v v9, (a0), zero
 ; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
 ; CHECK-NEXT:    vfredmin.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <2 x float>, <2 x float>* %x
   %red = call float @llvm.vector.reduce.fmin.v2f32(<2 x float> %v)
   ret float %red
 }
 
 declare float @llvm.vector.reduce.fmin.v4f32(<4 x float>)
 
 define float @vreduce_fmin_v4f32(<4 x float>* %x) {
 ; CHECK-LABEL: vreduce_fmin_v4f32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a1, %hi(.LCPI48_0)
-; CHECK-NEXT:    flw ft0, %lo(.LCPI48_0)(a1)
 ; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI48_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI48_0)
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
+; CHECK-NEXT:    vlse32.v v9, (a0), zero
 ; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
 ; CHECK-NEXT:    vfredmin.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x float>, <4 x float>* %x
   %red = call float @llvm.vector.reduce.fmin.v4f32(<4 x float> %v)
   ret float %red
 }
 
 define float @vreduce_fmin_v4f32_nonans(<4 x float>* %x) {
 ; CHECK-LABEL: vreduce_fmin_v4f32_nonans:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a1, %hi(.LCPI49_0)
-; CHECK-NEXT:    flw ft0, %lo(.LCPI49_0)(a1)
 ; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI49_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI49_0)
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
+; CHECK-NEXT:    vlse32.v v9, (a0), zero
 ; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
 ; CHECK-NEXT:    vfredmin.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x float>, <4 x float>* %x
   %red = call nnan float @llvm.vector.reduce.fmin.v4f32(<4 x float> %v)
   ret float %red
 }
 
 define float @vreduce_fmin_v4f32_nonans_noinfs(<4 x float>* %x) {
 ; CHECK-LABEL: vreduce_fmin_v4f32_nonans_noinfs:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a1, %hi(.LCPI50_0)
-; CHECK-NEXT:    flw ft0, %lo(.LCPI50_0)(a1)
 ; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI50_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI50_0)
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
+; CHECK-NEXT:    vlse32.v v9, (a0), zero
 ; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
 ; CHECK-NEXT:    vfredmin.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x float>, <4 x float>* %x
   %red = call nnan ninf float @llvm.vector.reduce.fmin.v4f32(<4 x float> %v)
   ret float %red
 }
 
 declare float @llvm.vector.reduce.fmin.v128f32(<128 x float>)
 
 define float @vreduce_fmin_v128f32(<128 x float>* %x) {
 ; CHECK-LABEL: vreduce_fmin_v128f32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
+; CHECK-NEXT:    vle32.v v8, (a0)
 ; CHECK-NEXT:    addi a2, a0, 384
-; CHECK-NEXT:    vle32.v v8, (a2)
-; CHECK-NEXT:    addi a2, a0, 128
 ; 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:    addi a0, a0, 256
-; CHECK-NEXT:    vle32.v v0, (a0)
-; CHECK-NEXT:    vfmin.vv v8, v16, v8
+; 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(.LCPI51_0)
-; CHECK-NEXT:    flw ft0, %lo(.LCPI51_0)(a0)
-; CHECK-NEXT:    vfmin.vv v16, v24, v0
-; CHECK-NEXT:    vfmin.vv v8, v16, v8
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v16, ft0
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI51_0)
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
+; CHECK-NEXT:    vlse32.v v16, (a0), zero
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
 ; CHECK-NEXT:    vfredmin.vs v8, v8, v16
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <128 x float>, <128 x float>* %x
   %red = call float @llvm.vector.reduce.fmin.v128f32(<128 x float> %v)
   ret float %red
 }
 
 declare double @llvm.vector.reduce.fmin.v2f64(<2 x double>)
 
 define double @vreduce_fmin_v2f64(<2 x double>* %x) {
 ; CHECK-LABEL: vreduce_fmin_v2f64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a1, %hi(.LCPI52_0)
-; CHECK-NEXT:    fld ft0, %lo(.LCPI52_0)(a1)
 ; CHECK-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
 ; CHECK-NEXT:    vle64.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI52_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI52_0)
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
+; CHECK-NEXT:    vlse64.v v9, (a0), zero
 ; CHECK-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
 ; CHECK-NEXT:    vfredmin.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <2 x double>, <2 x double>* %x
   %red = call double @llvm.vector.reduce.fmin.v2f64(<2 x double> %v)
   ret double %red
 }
 
 declare double @llvm.vector.reduce.fmin.v4f64(<4 x double>)
 
 define double @vreduce_fmin_v4f64(<4 x double>* %x) {
 ; CHECK-LABEL: vreduce_fmin_v4f64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a1, %hi(.LCPI53_0)
-; CHECK-NEXT:    fld ft0, %lo(.LCPI53_0)(a1)
 ; CHECK-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; CHECK-NEXT:    vle64.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v10, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI53_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI53_0)
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
+; CHECK-NEXT:    vlse64.v v10, (a0), zero
 ; CHECK-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; CHECK-NEXT:    vfredmin.vs v8, v8, v10
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x double>, <4 x double>* %x
   %red = call double @llvm.vector.reduce.fmin.v4f64(<4 x double> %v)
   ret double %red
 }
 
 define double @vreduce_fmin_v4f64_nonans(<4 x double>* %x) {
 ; CHECK-LABEL: vreduce_fmin_v4f64_nonans:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a1, %hi(.LCPI54_0)
-; CHECK-NEXT:    fld ft0, %lo(.LCPI54_0)(a1)
 ; CHECK-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; CHECK-NEXT:    vle64.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v10, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI54_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI54_0)
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
+; CHECK-NEXT:    vlse64.v v10, (a0), zero
 ; CHECK-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; CHECK-NEXT:    vfredmin.vs v8, v8, v10
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x double>, <4 x double>* %x
   %red = call nnan double @llvm.vector.reduce.fmin.v4f64(<4 x double> %v)
   ret double %red
 }
 
 define double @vreduce_fmin_v4f64_nonans_noinfs(<4 x double>* %x) {
 ; CHECK-LABEL: vreduce_fmin_v4f64_nonans_noinfs:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a1, %hi(.LCPI55_0)
-; CHECK-NEXT:    fld ft0, %lo(.LCPI55_0)(a1)
 ; CHECK-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; CHECK-NEXT:    vle64.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v10, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI55_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI55_0)
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
+; CHECK-NEXT:    vlse64.v v10, (a0), zero
 ; CHECK-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; CHECK-NEXT:    vfredmin.vs v8, v8, v10
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x double>, <4 x double>* %x
   %red = call nnan ninf double @llvm.vector.reduce.fmin.v4f64(<4 x double> %v)
   ret double %red
 }
 
 declare double @llvm.vector.reduce.fmin.v32f64(<32 x double>)
 
 define double @vreduce_fmin_v32f64(<32 x double>* %x) {
 ; CHECK-LABEL: vreduce_fmin_v32f64:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; CHECK-NEXT:    vle64.v v8, (a0)
 ; CHECK-NEXT:    addi a0, a0, 128
 ; CHECK-NEXT:    vle64.v v16, (a0)
-; CHECK-NEXT:    lui a0, %hi(.LCPI56_0)
-; CHECK-NEXT:    fld ft0, %lo(.LCPI56_0)(a0)
 ; CHECK-NEXT:    vfmin.vv v8, v8, v16
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v16, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI56_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI56_0)
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
+; CHECK-NEXT:    vlse64.v v16, (a0), zero
 ; CHECK-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; CHECK-NEXT:    vfredmin.vs v8, v8, v16
 ; 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.fmin.v32f64(<32 x double> %v)
   ret double %red
 }
 
 declare half @llvm.vector.reduce.fmax.v2f16(<2 x half>)
 
 define half @vreduce_fmax_v2f16(<2 x half>* %x) {
 ; CHECK-LABEL: vreduce_fmax_v2f16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a1, %hi(.LCPI57_0)
-; CHECK-NEXT:    flh ft0, %lo(.LCPI57_0)(a1)
 ; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI57_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI57_0)
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
+; CHECK-NEXT:    vlse16.v v9, (a0), zero
 ; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, mu
 ; CHECK-NEXT:    vfredmax.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 half @llvm.vector.reduce.fmax.v2f16(<2 x half> %v)
   ret half %red
 }
 
 declare half @llvm.vector.reduce.fmax.v4f16(<4 x half>)
 
 define half @vreduce_fmax_v4f16(<4 x half>* %x) {
 ; CHECK-LABEL: vreduce_fmax_v4f16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a1, %hi(.LCPI58_0)
-; CHECK-NEXT:    flh ft0, %lo(.LCPI58_0)(a1)
 ; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI58_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI58_0)
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
+; CHECK-NEXT:    vlse16.v v9, (a0), zero
 ; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
 ; CHECK-NEXT:    vfredmax.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x half>, <4 x half>* %x
   %red = call half @llvm.vector.reduce.fmax.v4f16(<4 x half> %v)
   ret half %red
 }
 
 define half @vreduce_fmax_v4f16_nonans(<4 x half>* %x) {
 ; CHECK-LABEL: vreduce_fmax_v4f16_nonans:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a1, %hi(.LCPI59_0)
-; CHECK-NEXT:    flh ft0, %lo(.LCPI59_0)(a1)
 ; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI59_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI59_0)
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
+; CHECK-NEXT:    vlse16.v v9, (a0), zero
 ; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
 ; CHECK-NEXT:    vfredmax.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x half>, <4 x half>* %x
   %red = call nnan half @llvm.vector.reduce.fmax.v4f16(<4 x half> %v)
   ret half %red
 }
 
 define half @vreduce_fmax_v4f16_nonans_noinfs(<4 x half>* %x) {
 ; CHECK-LABEL: vreduce_fmax_v4f16_nonans_noinfs:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a1, %hi(.LCPI60_0)
-; CHECK-NEXT:    flh ft0, %lo(.LCPI60_0)(a1)
 ; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI60_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI60_0)
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
+; CHECK-NEXT:    vlse16.v v9, (a0), zero
 ; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
 ; CHECK-NEXT:    vfredmax.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x half>, <4 x half>* %x
   %red = call nnan ninf half @llvm.vector.reduce.fmax.v4f16(<4 x half> %v)
   ret half %red
 }
 
 declare half @llvm.vector.reduce.fmax.v128f16(<128 x half>)
 
 define half @vreduce_fmax_v128f16(<128 x half>* %x) {
 ; CHECK-LABEL: vreduce_fmax_v128f16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 64
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
 ; CHECK-NEXT:    addi a0, a0, 128
 ; CHECK-NEXT:    vle16.v v16, (a0)
-; CHECK-NEXT:    lui a0, %hi(.LCPI61_0)
-; CHECK-NEXT:    flh ft0, %lo(.LCPI61_0)(a0)
 ; CHECK-NEXT:    vfmax.vv v8, v8, v16
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v16, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI61_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI61_0)
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
+; CHECK-NEXT:    vlse16.v v16, (a0), zero
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; CHECK-NEXT:    vfredmax.vs v8, v8, v16
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <128 x half>, <128 x half>* %x
   %red = call half @llvm.vector.reduce.fmax.v128f16(<128 x half> %v)
   ret half %red
 }
 
 declare float @llvm.vector.reduce.fmax.v2f32(<2 x float>)
 
 define float @vreduce_fmax_v2f32(<2 x float>* %x) {
 ; CHECK-LABEL: vreduce_fmax_v2f32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a1, %hi(.LCPI62_0)
-; CHECK-NEXT:    flw ft0, %lo(.LCPI62_0)(a1)
 ; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI62_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI62_0)
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
+; CHECK-NEXT:    vlse32.v v9, (a0), zero
 ; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
 ; CHECK-NEXT:    vfredmax.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <2 x float>, <2 x float>* %x
   %red = call float @llvm.vector.reduce.fmax.v2f32(<2 x float> %v)
   ret float %red
 }
 
 declare float @llvm.vector.reduce.fmax.v4f32(<4 x float>)
 
 define float @vreduce_fmax_v4f32(<4 x float>* %x) {
 ; CHECK-LABEL: vreduce_fmax_v4f32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a1, %hi(.LCPI63_0)
-; CHECK-NEXT:    flw ft0, %lo(.LCPI63_0)(a1)
 ; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI63_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI63_0)
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
+; CHECK-NEXT:    vlse32.v v9, (a0), zero
 ; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
 ; CHECK-NEXT:    vfredmax.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x float>, <4 x float>* %x
   %red = call float @llvm.vector.reduce.fmax.v4f32(<4 x float> %v)
   ret float %red
 }
 
 define float @vreduce_fmax_v4f32_nonans(<4 x float>* %x) {
 ; CHECK-LABEL: vreduce_fmax_v4f32_nonans:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a1, %hi(.LCPI64_0)
-; CHECK-NEXT:    flw ft0, %lo(.LCPI64_0)(a1)
 ; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI64_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI64_0)
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
+; CHECK-NEXT:    vlse32.v v9, (a0), zero
 ; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
 ; CHECK-NEXT:    vfredmax.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x float>, <4 x float>* %x
   %red = call nnan float @llvm.vector.reduce.fmax.v4f32(<4 x float> %v)
   ret float %red
 }
 
 define float @vreduce_fmax_v4f32_nonans_noinfs(<4 x float>* %x) {
 ; CHECK-LABEL: vreduce_fmax_v4f32_nonans_noinfs:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a1, %hi(.LCPI65_0)
-; CHECK-NEXT:    flw ft0, %lo(.LCPI65_0)(a1)
 ; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI65_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI65_0)
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
+; CHECK-NEXT:    vlse32.v v9, (a0), zero
 ; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
 ; CHECK-NEXT:    vfredmax.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x float>, <4 x float>* %x
   %red = call nnan ninf float @llvm.vector.reduce.fmax.v4f32(<4 x float> %v)
   ret float %red
 }
 
 declare float @llvm.vector.reduce.fmax.v128f32(<128 x float>)
 
 define float @vreduce_fmax_v128f32(<128 x float>* %x) {
 ; CHECK-LABEL: vreduce_fmax_v128f32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
+; CHECK-NEXT:    vle32.v v8, (a0)
 ; CHECK-NEXT:    addi a2, a0, 384
-; CHECK-NEXT:    vle32.v v8, (a2)
-; CHECK-NEXT:    addi a2, a0, 128
 ; 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:    addi a0, a0, 256
-; CHECK-NEXT:    vle32.v v0, (a0)
-; CHECK-NEXT:    vfmax.vv v8, v16, v8
+; CHECK-NEXT:    vle32.v v0, (a2)
+; CHECK-NEXT:    vfmax.vv v16, v24, v16
+; CHECK-NEXT:    vfmax.vv v8, v8, v0
+; CHECK-NEXT:    vfmax.vv v8, v8, v16
 ; CHECK-NEXT:    lui a0, %hi(.LCPI66_0)
-; CHECK-NEXT:    flw ft0, %lo(.LCPI66_0)(a0)
-; CHECK-NEXT:    vfmax.vv v16, v24, v0
-; CHECK-NEXT:    vfmax.vv v8, v16, v8
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v16, ft0
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI66_0)
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
+; CHECK-NEXT:    vlse32.v v16, (a0), zero
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
 ; CHECK-NEXT:    vfredmax.vs v8, v8, v16
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <128 x float>, <128 x float>* %x
   %red = call float @llvm.vector.reduce.fmax.v128f32(<128 x float> %v)
   ret float %red
 }
 
 declare double @llvm.vector.reduce.fmax.v2f64(<2 x double>)
 
 define double @vreduce_fmax_v2f64(<2 x double>* %x) {
 ; CHECK-LABEL: vreduce_fmax_v2f64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a1, %hi(.LCPI67_0)
-; CHECK-NEXT:    fld ft0, %lo(.LCPI67_0)(a1)
 ; CHECK-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
 ; CHECK-NEXT:    vle64.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI67_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI67_0)
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
+; CHECK-NEXT:    vlse64.v v9, (a0), zero
 ; CHECK-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
 ; CHECK-NEXT:    vfredmax.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <2 x double>, <2 x double>* %x
   %red = call double @llvm.vector.reduce.fmax.v2f64(<2 x double> %v)
   ret double %red
 }
 
 declare double @llvm.vector.reduce.fmax.v4f64(<4 x double>)
 
 define double @vreduce_fmax_v4f64(<4 x double>* %x) {
 ; CHECK-LABEL: vreduce_fmax_v4f64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a1, %hi(.LCPI68_0)
-; CHECK-NEXT:    fld ft0, %lo(.LCPI68_0)(a1)
 ; CHECK-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; CHECK-NEXT:    vle64.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v10, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI68_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI68_0)
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
+; CHECK-NEXT:    vlse64.v v10, (a0), zero
 ; CHECK-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; CHECK-NEXT:    vfredmax.vs v8, v8, v10
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x double>, <4 x double>* %x
   %red = call double @llvm.vector.reduce.fmax.v4f64(<4 x double> %v)
   ret double %red
 }
 
 define double @vreduce_fmax_v4f64_nonans(<4 x double>* %x) {
 ; CHECK-LABEL: vreduce_fmax_v4f64_nonans:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a1, %hi(.LCPI69_0)
-; CHECK-NEXT:    fld ft0, %lo(.LCPI69_0)(a1)
 ; CHECK-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; CHECK-NEXT:    vle64.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v10, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI69_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI69_0)
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
+; CHECK-NEXT:    vlse64.v v10, (a0), zero
 ; CHECK-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; CHECK-NEXT:    vfredmax.vs v8, v8, v10
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x double>, <4 x double>* %x
   %red = call nnan double @llvm.vector.reduce.fmax.v4f64(<4 x double> %v)
   ret double %red
 }
 
 define double @vreduce_fmax_v4f64_nonans_noinfs(<4 x double>* %x) {
 ; CHECK-LABEL: vreduce_fmax_v4f64_nonans_noinfs:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a1, %hi(.LCPI70_0)
-; CHECK-NEXT:    fld ft0, %lo(.LCPI70_0)(a1)
 ; CHECK-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; CHECK-NEXT:    vle64.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v10, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI70_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI70_0)
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
+; CHECK-NEXT:    vlse64.v v10, (a0), zero
 ; CHECK-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; CHECK-NEXT:    vfredmax.vs v8, v8, v10
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %v = load <4 x double>, <4 x double>* %x
   %red = call nnan ninf double @llvm.vector.reduce.fmax.v4f64(<4 x double> %v)
   ret double %red
 }
 
 declare double @llvm.vector.reduce.fmax.v32f64(<32 x double>)
 
 define double @vreduce_fmax_v32f64(<32 x double>* %x) {
 ; CHECK-LABEL: vreduce_fmax_v32f64:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; CHECK-NEXT:    vle64.v v8, (a0)
 ; CHECK-NEXT:    addi a0, a0, 128
 ; CHECK-NEXT:    vle64.v v16, (a0)
-; CHECK-NEXT:    lui a0, %hi(.LCPI71_0)
-; CHECK-NEXT:    fld ft0, %lo(.LCPI71_0)(a0)
 ; CHECK-NEXT:    vfmax.vv v8, v8, v16
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v16, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI71_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI71_0)
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
+; CHECK-NEXT:    vlse64.v v16, (a0), zero
 ; CHECK-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; CHECK-NEXT:    vfredmax.vs v8, v8, v16
 ; 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.fmax.v32f64(<32 x double> %v)
   ret double %red
 }

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=+experimental-v -riscv-v-vector-bits-min=128 -verify-machineinstrs < %s \
 ; RUN:   | FileCheck %s --check-prefixes=CHECK,RV32
 ; RUN: llc -mtriple=riscv64 -mattr=+experimental-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:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, tu, mu
 ; 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:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, tu, mu
 ; 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:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, tu, mu
 ; 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:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, tu, mu
 ; 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:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, tu, mu
 ; 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:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, tu, mu
 ; 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:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, tu, mu
 ; 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:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, tu, mu
 ; 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.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:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, tu, mu
 ; 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:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, tu, mu
 ; 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:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, tu, mu
 ; 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:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, tu, mu
 ; 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:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, tu, mu
 ; 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:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, tu, mu
 ; 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:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, tu, mu
 ; 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:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, tu, mu
 ; 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:    vsetvli a2, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, mf4, tu, mu
 ; 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:    lui a2, 16
 ; RV32-NEXT:    addi a2, a2, -1
 ; RV32-NEXT:    and a0, a0, a2
-; RV32-NEXT:    vsetvli a2, zero, e16, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; RV32-NEXT:    vmv.v.x v9, a0
 ; RV32-NEXT:    vsetvli zero, a1, e16, mf4, tu, mu
 ; 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:    lui a2, 16
 ; RV64-NEXT:    addiw a2, a2, -1
 ; RV64-NEXT:    and a0, a0, a2
-; RV64-NEXT:    vsetvli a2, zero, e16, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vsetvli zero, a1, e16, mf4, tu, mu
 ; RV64-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v9
 ; RV64-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:    vsetvli a2, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, mf4, tu, mu
 ; 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:    lui a2, 16
 ; RV32-NEXT:    addi a2, a2, -1
 ; RV32-NEXT:    and a0, a0, a2
-; RV32-NEXT:    vsetvli a2, zero, e16, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; RV32-NEXT:    vmv.v.x v9, a0
 ; RV32-NEXT:    vsetvli zero, a1, e16, mf4, tu, mu
 ; 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:    lui a2, 16
 ; RV64-NEXT:    addiw a2, a2, -1
 ; RV64-NEXT:    and a0, a0, a2
-; RV64-NEXT:    vsetvli a2, zero, e16, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vsetvli zero, a1, e16, mf4, tu, mu
 ; RV64-NEXT:    vredminu.vs v9, v8, v9, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v9
 ; RV64-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:    vsetvli a2, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, mf4, tu, mu
 ; 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:    vsetvli a2, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, mf4, tu, mu
 ; 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:    vsetvli a2, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, mf4, tu, mu
 ; 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:    vsetvli a2, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, mf4, tu, mu
 ; 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:    vsetvli a2, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, mf2, tu, mu
 ; 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:    lui a2, 16
 ; RV32-NEXT:    addi a2, a2, -1
 ; RV32-NEXT:    and a0, a0, a2
-; RV32-NEXT:    vsetvli a2, zero, e16, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; RV32-NEXT:    vmv.v.x v9, a0
 ; RV32-NEXT:    vsetvli zero, a1, e16, mf2, tu, mu
 ; 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:    lui a2, 16
 ; RV64-NEXT:    addiw a2, a2, -1
 ; RV64-NEXT:    and a0, a0, a2
-; RV64-NEXT:    vsetvli a2, zero, e16, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vsetvli zero, a1, e16, mf2, tu, mu
 ; RV64-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v9
 ; RV64-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:    vsetvli a2, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, mf2, tu, mu
 ; 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:    lui a2, 16
 ; RV32-NEXT:    addi a2, a2, -1
 ; RV32-NEXT:    and a0, a0, a2
-; RV32-NEXT:    vsetvli a2, zero, e16, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; RV32-NEXT:    vmv.v.x v9, a0
 ; RV32-NEXT:    vsetvli zero, a1, e16, mf2, tu, mu
 ; 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:    lui a2, 16
 ; RV64-NEXT:    addiw a2, a2, -1
 ; RV64-NEXT:    and a0, a0, a2
-; RV64-NEXT:    vsetvli a2, zero, e16, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vsetvli zero, a1, e16, mf2, tu, mu
 ; RV64-NEXT:    vredminu.vs v9, v8, v9, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v9
 ; RV64-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:    vsetvli a2, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, mf2, tu, mu
 ; 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:    vsetvli a2, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, mf2, tu, mu
 ; 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:    vsetvli a2, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, mf2, tu, mu
 ; 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:    vsetvli a2, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, mf2, tu, mu
 ; 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:    vsetvli a2, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, tu, mu
 ; 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:    vsetvli a2, zero, e32, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; RV32-NEXT:    vmv.v.x v9, a0
 ; RV32-NEXT:    vsetvli zero, a1, e32, mf2, tu, mu
 ; 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:    vsetvli a2, zero, e32, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vsetvli zero, a1, e32, mf2, tu, mu
 ; RV64-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v9
 ; RV64-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:    vsetvli a2, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, tu, mu
 ; 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:    vsetvli a2, zero, e32, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; RV32-NEXT:    vmv.v.x v9, a0
 ; RV32-NEXT:    vsetvli zero, a1, e32, mf2, tu, mu
 ; 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:    vsetvli a2, zero, e32, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vsetvli zero, a1, e32, mf2, tu, mu
 ; RV64-NEXT:    vredminu.vs v9, v8, v9, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v9
 ; RV64-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:    vsetvli a2, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, tu, mu
 ; 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:    vsetvli a2, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, tu, mu
 ; 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:    vsetvli a2, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, tu, mu
 ; 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:    vsetvli a2, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, tu, mu
 ; 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:    vsetvli a2, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m1, tu, mu
 ; 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:    vsetvli a2, zero, e32, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; RV32-NEXT:    vmv.v.x v9, a0
 ; RV32-NEXT:    vsetvli zero, a1, e32, m1, tu, mu
 ; 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:    vsetvli a2, zero, e32, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vsetvli zero, a1, e32, m1, tu, mu
 ; RV64-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v9
 ; RV64-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:    vsetvli a2, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m1, tu, mu
 ; 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:    vsetvli a2, zero, e32, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; RV32-NEXT:    vmv.v.x v9, a0
 ; RV32-NEXT:    vsetvli zero, a1, e32, m1, tu, mu
 ; 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:    vsetvli a2, zero, e32, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vsetvli zero, a1, e32, m1, tu, mu
 ; RV64-NEXT:    vredminu.vs v9, v8, v9, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v9
 ; RV64-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:    vsetvli a2, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m1, tu, mu
 ; 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:    vsetvli a2, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m1, tu, mu
 ; 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:    vsetvli a2, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m1, tu, mu
 ; 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:    vsetvli a2, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m1, tu, mu
 ; 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 i64 @llvm.vp.reduce.add.v2i64(i64, <2 x i64>, <2 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v9, (a0), zero
 ; RV32-NEXT:    vsetvli zero, a2, e64, m1, tu, mu
 ; 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, mu
 ; 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:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vsetvli zero, a1, e64, m1, tu, mu
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v9, (a0), zero
 ; RV32-NEXT:    vsetvli zero, a2, e64, m1, tu, mu
 ; 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, mu
 ; 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:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vsetvli zero, a1, e64, m1, tu, mu
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v9, (a0), zero
 ; RV32-NEXT:    vsetvli zero, a2, e64, m1, tu, mu
 ; 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, mu
 ; 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:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vsetvli zero, a1, e64, m1, tu, mu
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v9, (a0), zero
 ; RV32-NEXT:    vsetvli zero, a2, e64, m1, tu, mu
 ; 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, mu
 ; 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:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vsetvli zero, a1, e64, m1, tu, mu
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v9, (a0), zero
 ; RV32-NEXT:    vsetvli zero, a2, e64, m1, tu, mu
 ; 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, mu
 ; 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:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vsetvli zero, a1, e64, m1, tu, mu
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v9, (a0), zero
 ; RV32-NEXT:    vsetvli zero, a2, e64, m1, tu, mu
 ; 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, mu
 ; 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:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vsetvli zero, a1, e64, m1, tu, mu
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v9, (a0), zero
 ; RV32-NEXT:    vsetvli zero, a2, e64, m1, tu, mu
 ; 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, mu
 ; 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:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vsetvli zero, a1, e64, m1, tu, mu
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v9, (a0), zero
 ; RV32-NEXT:    vsetvli zero, a2, e64, m1, tu, mu
 ; 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, mu
 ; 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:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vsetvli zero, a1, e64, m1, tu, mu
 ; 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)
 
 define signext i64 @vpreduce_add_v4i64(i64 signext %s, <4 x i64> %v, <4 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_add_v4i64:
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v10, (a0), zero
 ; RV32-NEXT:    vsetvli zero, a2, e64, m2, tu, mu
 ; RV32-NEXT:    vredsum.vs v10, v8, v10, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v10
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v10, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_add_v4i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v10, a0
 ; RV64-NEXT:    vsetvli zero, a1, e64, m2, tu, mu
 ; RV64-NEXT:    vredsum.vs v10, v8, v10, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v10
 ; RV64-NEXT:    ret
   %r = call i64 @llvm.vp.reduce.add.v4i64(i64 %s, <4 x i64> %v, <4 x i1> %m, i32 %evl)
   ret i64 %r
 }
 
 declare i64 @llvm.vp.reduce.umax.v4i64(i64, <4 x i64>, <4 x i1>, i32)
 
 define signext i64 @vpreduce_umax_v4i64(i64 signext %s, <4 x i64> %v, <4 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_umax_v4i64:
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v10, (a0), zero
 ; RV32-NEXT:    vsetvli zero, a2, e64, m2, tu, mu
 ; RV32-NEXT:    vredmaxu.vs v10, v8, v10, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v10
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v10, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_umax_v4i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v10, a0
 ; RV64-NEXT:    vsetvli zero, a1, e64, m2, tu, mu
 ; RV64-NEXT:    vredmaxu.vs v10, v8, v10, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v10
 ; RV64-NEXT:    ret
   %r = call i64 @llvm.vp.reduce.umax.v4i64(i64 %s, <4 x i64> %v, <4 x i1> %m, i32 %evl)
   ret i64 %r
 }
 
 declare i64 @llvm.vp.reduce.smax.v4i64(i64, <4 x i64>, <4 x i1>, i32)
 
 define signext i64 @vpreduce_smax_v4i64(i64 signext %s, <4 x i64> %v, <4 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_smax_v4i64:
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v10, (a0), zero
 ; RV32-NEXT:    vsetvli zero, a2, e64, m2, tu, mu
 ; RV32-NEXT:    vredmax.vs v10, v8, v10, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v10
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v10, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_smax_v4i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v10, a0
 ; RV64-NEXT:    vsetvli zero, a1, e64, m2, tu, mu
 ; RV64-NEXT:    vredmax.vs v10, v8, v10, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v10
 ; RV64-NEXT:    ret
   %r = call i64 @llvm.vp.reduce.smax.v4i64(i64 %s, <4 x i64> %v, <4 x i1> %m, i32 %evl)
   ret i64 %r
 }
 
 declare i64 @llvm.vp.reduce.umin.v4i64(i64, <4 x i64>, <4 x i1>, i32)
 
 define signext i64 @vpreduce_umin_v4i64(i64 signext %s, <4 x i64> %v, <4 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_umin_v4i64:
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v10, (a0), zero
 ; RV32-NEXT:    vsetvli zero, a2, e64, m2, tu, mu
 ; RV32-NEXT:    vredminu.vs v10, v8, v10, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v10
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v10, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_umin_v4i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v10, a0
 ; RV64-NEXT:    vsetvli zero, a1, e64, m2, tu, mu
 ; RV64-NEXT:    vredminu.vs v10, v8, v10, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v10
 ; RV64-NEXT:    ret
   %r = call i64 @llvm.vp.reduce.umin.v4i64(i64 %s, <4 x i64> %v, <4 x i1> %m, i32 %evl)
   ret i64 %r
 }
 
 declare i64 @llvm.vp.reduce.smin.v4i64(i64, <4 x i64>, <4 x i1>, i32)
 
 define signext i64 @vpreduce_smin_v4i64(i64 signext %s, <4 x i64> %v, <4 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_smin_v4i64:
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v10, (a0), zero
 ; RV32-NEXT:    vsetvli zero, a2, e64, m2, tu, mu
 ; RV32-NEXT:    vredmin.vs v10, v8, v10, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v10
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v10, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_smin_v4i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v10, a0
 ; RV64-NEXT:    vsetvli zero, a1, e64, m2, tu, mu
 ; RV64-NEXT:    vredmin.vs v10, v8, v10, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v10
 ; RV64-NEXT:    ret
   %r = call i64 @llvm.vp.reduce.smin.v4i64(i64 %s, <4 x i64> %v, <4 x i1> %m, i32 %evl)
   ret i64 %r
 }
 
 declare i64 @llvm.vp.reduce.and.v4i64(i64, <4 x i64>, <4 x i1>, i32)
 
 define signext i64 @vpreduce_and_v4i64(i64 signext %s, <4 x i64> %v, <4 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_and_v4i64:
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v10, (a0), zero
 ; RV32-NEXT:    vsetvli zero, a2, e64, m2, tu, mu
 ; RV32-NEXT:    vredand.vs v10, v8, v10, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v10
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v10, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_and_v4i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v10, a0
 ; RV64-NEXT:    vsetvli zero, a1, e64, m2, tu, mu
 ; RV64-NEXT:    vredand.vs v10, v8, v10, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v10
 ; RV64-NEXT:    ret
   %r = call i64 @llvm.vp.reduce.and.v4i64(i64 %s, <4 x i64> %v, <4 x i1> %m, i32 %evl)
   ret i64 %r
 }
 
 declare i64 @llvm.vp.reduce.or.v4i64(i64, <4 x i64>, <4 x i1>, i32)
 
 define signext i64 @vpreduce_or_v4i64(i64 signext %s, <4 x i64> %v, <4 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_or_v4i64:
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v10, (a0), zero
 ; RV32-NEXT:    vsetvli zero, a2, e64, m2, tu, mu
 ; RV32-NEXT:    vredor.vs v10, v8, v10, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v10
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v10, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_or_v4i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v10, a0
 ; RV64-NEXT:    vsetvli zero, a1, e64, m2, tu, mu
 ; RV64-NEXT:    vredor.vs v10, v8, v10, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v10
 ; RV64-NEXT:    ret
   %r = call i64 @llvm.vp.reduce.or.v4i64(i64 %s, <4 x i64> %v, <4 x i1> %m, i32 %evl)
   ret i64 %r
 }
 
 declare i64 @llvm.vp.reduce.xor.v4i64(i64, <4 x i64>, <4 x i1>, i32)
 
 define signext i64 @vpreduce_xor_v4i64(i64 signext %s, <4 x i64> %v, <4 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_xor_v4i64:
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v10, (a0), zero
 ; RV32-NEXT:    vsetvli zero, a2, e64, m2, tu, mu
 ; RV32-NEXT:    vredxor.vs v10, v8, v10, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v10
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v10, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_xor_v4i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v10, a0
 ; RV64-NEXT:    vsetvli zero, a1, e64, m2, tu, mu
 ; RV64-NEXT:    vredxor.vs v10, v8, v10, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v10
 ; RV64-NEXT:    ret
   %r = call i64 @llvm.vp.reduce.xor.v4i64(i64 %s, <4 x i64> %v, <4 x i1> %m, i32 %evl)
   ret i64 %r
 }

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, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetivli zero, 2, e8, mf8, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetivli zero, 4, e8, mf4, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetivli zero, 8, e8, mf2, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetivli zero, 16, e8, m1, ta, mu
 ; 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 i8 @llvm.vector.reduce.add.v32i8(<32 x i8>)
 
 define i8 @vreduce_add_v32i8(<32 x i8>* %x) {
 ; CHECK-LABEL: vreduce_add_v32i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m2, ta, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, 0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m2, ta, mu
 ; CHECK-NEXT:    vredsum.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <32 x i8>, <32 x i8>* %x
   %red = call i8 @llvm.vector.reduce.add.v32i8(<32 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.add.v64i8(<64 x i8>)
 
 define i8 @vreduce_add_v64i8(<64 x i8>* %x) {
 ; CHECK-LABEL: vreduce_add_v64i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 64
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m4, ta, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v12, 0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m4, ta, mu
 ; CHECK-NEXT:    vredsum.vs v8, v8, v12
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <64 x i8>, <64 x i8>* %x
   %red = call i8 @llvm.vector.reduce.add.v64i8(<64 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.add.v128i8(<128 x i8>)
 
 define i8 @vreduce_add_v128i8(<128 x i8>* %x) {
 ; CHECK-LABEL: vreduce_add_v128i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 128
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m8, ta, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v16, 0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m8, ta, mu
 ; CHECK-NEXT:    vredsum.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <128 x i8>, <128 x i8>* %x
   %red = call i8 @llvm.vector.reduce.add.v128i8(<128 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.add.v256i8(<256 x i8>)
 
 define i8 @vreduce_add_v256i8(<256 x i8>* %x) {
 ; CHECK-LABEL: vreduce_add_v256i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 128
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m8, ta, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
 ; CHECK-NEXT:    addi a0, a0, 128
 ; CHECK-NEXT:    vle8.v v16, (a0)
 ; CHECK-NEXT:    vadd.vv v8, v8, v16
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v16, 0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m8, ta, mu
 ; CHECK-NEXT:    vredsum.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <256 x i8>, <256 x i8>* %x
   %red = call i8 @llvm.vector.reduce.add.v256i8(<256 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, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, mu
 ; 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
 }
 
 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, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
 ; 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
 }
 
 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, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetivli zero, 8, e16, m1, ta, mu
 ; 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
 }
 
 declare i16 @llvm.vector.reduce.add.v16i16(<16 x i16>)
 
 define i16 @vreduce_add_v16i16(<16 x i16>* %x) {
 ; CHECK-LABEL: vreduce_add_v16i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 16, e16, m2, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, 0
 ; CHECK-NEXT:    vsetivli zero, 16, e16, m2, ta, mu
 ; CHECK-NEXT:    vredsum.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <16 x i16>, <16 x i16>* %x
   %red = call i16 @llvm.vector.reduce.add.v16i16(<16 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.add.v32i16(<32 x i16>)
 
 define i16 @vreduce_add_v32i16(<32 x i16>* %x) {
 ; CHECK-LABEL: vreduce_add_v32i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m4, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v12, 0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m4, ta, mu
 ; CHECK-NEXT:    vredsum.vs v8, v8, v12
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <32 x i16>, <32 x i16>* %x
   %red = call i16 @llvm.vector.reduce.add.v32i16(<32 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.add.v64i16(<64 x i16>)
 
 define i16 @vreduce_add_v64i16(<64 x i16>* %x) {
 ; CHECK-LABEL: vreduce_add_v64i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 64
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v16, 0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; CHECK-NEXT:    vredsum.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <64 x i16>, <64 x i16>* %x
   %red = call i16 @llvm.vector.reduce.add.v64i16(<64 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.add.v128i16(<128 x i16>)
 
 define i16 @vreduce_add_v128i16(<128 x i16>* %x) {
 ; CHECK-LABEL: vreduce_add_v128i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 64
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
 ; CHECK-NEXT:    addi a0, a0, 128
 ; CHECK-NEXT:    vle16.v v16, (a0)
 ; CHECK-NEXT:    vadd.vv v8, v8, v16
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v16, 0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; CHECK-NEXT:    vredsum.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <128 x i16>, <128 x i16>* %x
   %red = call i16 @llvm.vector.reduce.add.v128i16(<128 x i16> %v)
   ret i16 %red

 
 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, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
 ; 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
 }
 
 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, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
 ; 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
 }
 
 declare i32 @llvm.vector.reduce.add.v8i32(<8 x i32>)
 
 define i32 @vreduce_add_v8i32(<8 x i32>* %x) {
 ; CHECK-LABEL: vreduce_add_v8i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 8, e32, m2, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, 0
 ; CHECK-NEXT:    vsetivli zero, 8, e32, m2, ta, mu
 ; CHECK-NEXT:    vredsum.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <8 x i32>, <8 x i32>* %x
   %red = call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.add.v16i32(<16 x i32>)
 
 define i32 @vreduce_add_v16i32(<16 x i32>* %x) {
 ; CHECK-LABEL: vreduce_add_v16i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 16, e32, m4, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v12, 0
 ; CHECK-NEXT:    vsetivli zero, 16, e32, m4, ta, mu
 ; CHECK-NEXT:    vredsum.vs v8, v8, v12
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <16 x i32>, <16 x i32>* %x
   %red = call i32 @llvm.vector.reduce.add.v16i32(<16 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.add.v32i32(<32 x i32>)
 
 define i32 @vreduce_add_v32i32(<32 x i32>* %x) {
 ; CHECK-LABEL: vreduce_add_v32i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v16, 0
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
 ; CHECK-NEXT:    vredsum.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <32 x i32>, <32 x i32>* %x
   %red = call i32 @llvm.vector.reduce.add.v32i32(<32 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.add.v64i32(<64 x i32>)
 
 define i32 @vreduce_add_v64i32(<64 x i32>* %x) {
 ; CHECK-LABEL: vreduce_add_v64i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
 ; CHECK-NEXT:    addi a0, a0, 128
 ; CHECK-NEXT:    vle32.v v16, (a0)
 ; CHECK-NEXT:    vadd.vv v8, v8, v16
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v16, 0
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
 ; CHECK-NEXT:    vredsum.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <64 x i32>, <64 x i32>* %x
   %red = call i32 @llvm.vector.reduce.add.v64i32(<64 x i32> %v)
   ret i32 %red

 
 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, mu
 ; RV32-NEXT:    vle64.v v8, (a0)
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vmv.v.i v9, 0
 ; RV32-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
 ; 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, mu
 ; 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, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
-; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.i v9, 0
 ; RV64-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
 ; 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
 }
 
 declare i64 @llvm.vector.reduce.add.v4i64(<4 x i64>)
 
 define i64 @vreduce_add_v4i64(<4 x i64>* %x) {
 ; RV32-LABEL: vreduce_add_v4i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; RV32-NEXT:    vle64.v v8, (a0)
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vmv.v.i v10, 0
 ; RV32-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; RV32-NEXT:    vredsum.vs v8, v8, v10
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_add_v4i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
-; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.i v10, 0
 ; RV64-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; RV64-NEXT:    vredsum.vs v8, v8, v10
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <4 x i64>, <4 x i64>* %x
   %red = call i64 @llvm.vector.reduce.add.v4i64(<4 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.add.v8i64(<8 x i64>)
 
 define i64 @vreduce_add_v8i64(<8 x i64>* %x) {
 ; RV32-LABEL: vreduce_add_v8i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
 ; RV32-NEXT:    vle64.v v8, (a0)
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vmv.v.i v12, 0
 ; RV32-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
 ; RV32-NEXT:    vredsum.vs v8, v8, v12
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_add_v8i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
-; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.i v12, 0
 ; RV64-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
 ; RV64-NEXT:    vredsum.vs v8, v8, v12
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <8 x i64>, <8 x i64>* %x
   %red = call i64 @llvm.vector.reduce.add.v8i64(<8 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.add.v16i64(<16 x i64>)
 
 define i64 @vreduce_add_v16i64(<16 x i64>* %x) {
 ; RV32-LABEL: vreduce_add_v16i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV32-NEXT:    vle64.v v8, (a0)
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vmv.v.i v16, 0
 ; RV32-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV32-NEXT:    vredsum.vs v8, v8, v16
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_add_v16i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
-; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.i v16, 0
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vredsum.vs v8, v8, v16
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <16 x i64>, <16 x i64>* %x
   %red = call i64 @llvm.vector.reduce.add.v16i64(<16 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.add.v32i64(<32 x i64>)
 
 define i64 @vreduce_add_v32i64(<32 x i64>* %x) {
 ; RV32-LABEL: vreduce_add_v32i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV32-NEXT:    vle64.v v8, (a0)
 ; RV32-NEXT:    addi a0, a0, 128
 ; RV32-NEXT:    vle64.v v16, (a0)
 ; RV32-NEXT:    vadd.vv v8, v8, v16
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vmv.v.i v16, 0
 ; RV32-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV32-NEXT:    vredsum.vs v8, v8, v16
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_add_v32i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
 ; RV64-NEXT:    addi a0, a0, 128
 ; RV64-NEXT:    vle64.v v16, (a0)
 ; RV64-NEXT:    vadd.vv v8, v8, v16
-; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.i v16, 0
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vredsum.vs v8, v8, v16
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <32 x i64>, <32 x i64>* %x
   %red = call i64 @llvm.vector.reduce.add.v32i64(<32 x i64> %v)
   ret i64 %red

 ; RV32-NEXT:    vle64.v v16, (a1)
 ; RV32-NEXT:    addi a1, a0, 256
 ; RV32-NEXT:    addi a0, a0, 128
 ; RV32-NEXT:    vle64.v v24, (a0)
 ; RV32-NEXT:    vle64.v v0, (a1)
 ; RV32-NEXT:    vadd.vv v16, v24, v16
 ; RV32-NEXT:    vadd.vv v8, v8, v0
 ; RV32-NEXT:    vadd.vv v8, v8, v16
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vmv.v.i v16, 0
 ; RV32-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV32-NEXT:    vredsum.vs v8, v8, v16
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_add_v64i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
 ; RV64-NEXT:    addi a1, a0, 384
 ; RV64-NEXT:    vle64.v v16, (a1)
 ; RV64-NEXT:    addi a1, a0, 256
 ; RV64-NEXT:    addi a0, a0, 128
 ; RV64-NEXT:    vle64.v v24, (a0)
 ; RV64-NEXT:    vle64.v v0, (a1)
 ; RV64-NEXT:    vadd.vv v16, v24, v16
 ; RV64-NEXT:    vadd.vv v8, v8, v0
 ; RV64-NEXT:    vadd.vv v8, v8, v16
-; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.i v16, 0
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vredsum.vs v8, v8, v16
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <64 x i64>, <64 x i64>* %x
   %red = call i64 @llvm.vector.reduce.add.v64i64(<64 x i64> %v)
   ret i64 %red

 
 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, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetivli zero, 2, e8, mf8, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetivli zero, 4, e8, mf4, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetivli zero, 8, e8, mf2, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetivli zero, 16, e8, m1, ta, mu
 ; 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 i8 @llvm.vector.reduce.and.v32i8(<32 x i8>)
 
 define i8 @vreduce_and_v32i8(<32 x i8>* %x) {
 ; CHECK-LABEL: vreduce_and_v32i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m2, ta, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, -1
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m2, ta, mu
 ; CHECK-NEXT:    vredand.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <32 x i8>, <32 x i8>* %x
   %red = call i8 @llvm.vector.reduce.and.v32i8(<32 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.and.v64i8(<64 x i8>)
 
 define i8 @vreduce_and_v64i8(<64 x i8>* %x) {
 ; CHECK-LABEL: vreduce_and_v64i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 64
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m4, ta, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v12, -1
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m4, ta, mu
 ; CHECK-NEXT:    vredand.vs v8, v8, v12
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <64 x i8>, <64 x i8>* %x
   %red = call i8 @llvm.vector.reduce.and.v64i8(<64 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.and.v128i8(<128 x i8>)
 
 define i8 @vreduce_and_v128i8(<128 x i8>* %x) {
 ; CHECK-LABEL: vreduce_and_v128i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 128
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m8, ta, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v16, -1
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m8, ta, mu
 ; CHECK-NEXT:    vredand.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <128 x i8>, <128 x i8>* %x
   %red = call i8 @llvm.vector.reduce.and.v128i8(<128 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.and.v256i8(<256 x i8>)
 
 define i8 @vreduce_and_v256i8(<256 x i8>* %x) {
 ; CHECK-LABEL: vreduce_and_v256i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 128
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m8, ta, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
 ; CHECK-NEXT:    addi a0, a0, 128
 ; CHECK-NEXT:    vle8.v v16, (a0)
 ; CHECK-NEXT:    vand.vv v8, v8, v16
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v16, -1
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m8, ta, mu
 ; CHECK-NEXT:    vredand.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <256 x i8>, <256 x i8>* %x
   %red = call i8 @llvm.vector.reduce.and.v256i8(<256 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, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetivli zero, 8, e16, m1, ta, mu
 ; 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 i16 @llvm.vector.reduce.and.v16i16(<16 x i16>)
 
 define i16 @vreduce_and_v16i16(<16 x i16>* %x) {
 ; CHECK-LABEL: vreduce_and_v16i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 16, e16, m2, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, -1
 ; CHECK-NEXT:    vsetivli zero, 16, e16, m2, ta, mu
 ; CHECK-NEXT:    vredand.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <16 x i16>, <16 x i16>* %x
   %red = call i16 @llvm.vector.reduce.and.v16i16(<16 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.and.v32i16(<32 x i16>)
 
 define i16 @vreduce_and_v32i16(<32 x i16>* %x) {
 ; CHECK-LABEL: vreduce_and_v32i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m4, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v12, -1
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m4, ta, mu
 ; CHECK-NEXT:    vredand.vs v8, v8, v12
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <32 x i16>, <32 x i16>* %x
   %red = call i16 @llvm.vector.reduce.and.v32i16(<32 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.and.v64i16(<64 x i16>)
 
 define i16 @vreduce_and_v64i16(<64 x i16>* %x) {
 ; CHECK-LABEL: vreduce_and_v64i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 64
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v16, -1
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; CHECK-NEXT:    vredand.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <64 x i16>, <64 x i16>* %x
   %red = call i16 @llvm.vector.reduce.and.v64i16(<64 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.and.v128i16(<128 x i16>)
 
 define i16 @vreduce_and_v128i16(<128 x i16>* %x) {
 ; CHECK-LABEL: vreduce_and_v128i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 64
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
 ; CHECK-NEXT:    addi a0, a0, 128
 ; CHECK-NEXT:    vle16.v v16, (a0)
 ; CHECK-NEXT:    vand.vv v8, v8, v16
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v16, -1
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; CHECK-NEXT:    vredand.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <128 x i16>, <128 x i16>* %x
   %red = call i16 @llvm.vector.reduce.and.v128i16(<128 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, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
 ; 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 i32 @llvm.vector.reduce.and.v8i32(<8 x i32>)
 
 define i32 @vreduce_and_v8i32(<8 x i32>* %x) {
 ; CHECK-LABEL: vreduce_and_v8i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 8, e32, m2, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, -1
 ; CHECK-NEXT:    vsetivli zero, 8, e32, m2, ta, mu
 ; CHECK-NEXT:    vredand.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <8 x i32>, <8 x i32>* %x
   %red = call i32 @llvm.vector.reduce.and.v8i32(<8 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.and.v16i32(<16 x i32>)
 
 define i32 @vreduce_and_v16i32(<16 x i32>* %x) {
 ; CHECK-LABEL: vreduce_and_v16i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 16, e32, m4, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v12, -1
 ; CHECK-NEXT:    vsetivli zero, 16, e32, m4, ta, mu
 ; CHECK-NEXT:    vredand.vs v8, v8, v12
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <16 x i32>, <16 x i32>* %x
   %red = call i32 @llvm.vector.reduce.and.v16i32(<16 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.and.v32i32(<32 x i32>)
 
 define i32 @vreduce_and_v32i32(<32 x i32>* %x) {
 ; CHECK-LABEL: vreduce_and_v32i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v16, -1
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
 ; CHECK-NEXT:    vredand.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <32 x i32>, <32 x i32>* %x
   %red = call i32 @llvm.vector.reduce.and.v32i32(<32 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.and.v64i32(<64 x i32>)
 
 define i32 @vreduce_and_v64i32(<64 x i32>* %x) {
 ; CHECK-LABEL: vreduce_and_v64i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
 ; CHECK-NEXT:    addi a0, a0, 128
 ; CHECK-NEXT:    vle32.v v16, (a0)
 ; CHECK-NEXT:    vand.vv v8, v8, v16
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v16, -1
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
 ; CHECK-NEXT:    vredand.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <64 x i32>, <64 x i32>* %x
   %red = call i32 @llvm.vector.reduce.and.v64i32(<64 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, mu
 ; RV32-NEXT:    vle64.v v8, (a0)
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vmv.v.i v9, -1
 ; RV32-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
 ; 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, mu
 ; 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, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
-; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.i v9, -1
 ; RV64-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
 ; 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 i64 @llvm.vector.reduce.and.v4i64(<4 x i64>)
 
 define i64 @vreduce_and_v4i64(<4 x i64>* %x) {
 ; RV32-LABEL: vreduce_and_v4i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; RV32-NEXT:    vle64.v v8, (a0)
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vmv.v.i v10, -1
 ; RV32-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; RV32-NEXT:    vredand.vs v8, v8, v10
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_and_v4i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
-; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.i v10, -1
 ; RV64-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; RV64-NEXT:    vredand.vs v8, v8, v10
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <4 x i64>, <4 x i64>* %x
   %red = call i64 @llvm.vector.reduce.and.v4i64(<4 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.and.v8i64(<8 x i64>)
 
 define i64 @vreduce_and_v8i64(<8 x i64>* %x) {
 ; RV32-LABEL: vreduce_and_v8i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
 ; RV32-NEXT:    vle64.v v8, (a0)
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vmv.v.i v12, -1
 ; RV32-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
 ; RV32-NEXT:    vredand.vs v8, v8, v12
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_and_v8i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
-; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.i v12, -1
 ; RV64-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
 ; RV64-NEXT:    vredand.vs v8, v8, v12
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <8 x i64>, <8 x i64>* %x
   %red = call i64 @llvm.vector.reduce.and.v8i64(<8 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.and.v16i64(<16 x i64>)
 
 define i64 @vreduce_and_v16i64(<16 x i64>* %x) {
 ; RV32-LABEL: vreduce_and_v16i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV32-NEXT:    vle64.v v8, (a0)
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vmv.v.i v16, -1
 ; RV32-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV32-NEXT:    vredand.vs v8, v8, v16
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_and_v16i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
-; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.i v16, -1
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vredand.vs v8, v8, v16
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <16 x i64>, <16 x i64>* %x
   %red = call i64 @llvm.vector.reduce.and.v16i64(<16 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.and.v32i64(<32 x i64>)
 
 define i64 @vreduce_and_v32i64(<32 x i64>* %x) {
 ; RV32-LABEL: vreduce_and_v32i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV32-NEXT:    vle64.v v8, (a0)
 ; RV32-NEXT:    addi a0, a0, 128
 ; RV32-NEXT:    vle64.v v16, (a0)
 ; RV32-NEXT:    vand.vv v8, v8, v16
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vmv.v.i v16, -1
 ; RV32-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV32-NEXT:    vredand.vs v8, v8, v16
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_and_v32i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
 ; RV64-NEXT:    addi a0, a0, 128
 ; RV64-NEXT:    vle64.v v16, (a0)
 ; RV64-NEXT:    vand.vv v8, v8, v16
-; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.i v16, -1
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vredand.vs v8, v8, v16
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <32 x i64>, <32 x i64>* %x
   %red = call i64 @llvm.vector.reduce.and.v32i64(<32 x i64> %v)
   ret i64 %red

 ; RV32-NEXT:    vle64.v v16, (a1)
 ; RV32-NEXT:    addi a1, a0, 256
 ; RV32-NEXT:    addi a0, a0, 128
 ; RV32-NEXT:    vle64.v v24, (a0)
 ; RV32-NEXT:    vle64.v v0, (a1)
 ; RV32-NEXT:    vand.vv v16, v24, v16
 ; RV32-NEXT:    vand.vv v8, v8, v0
 ; RV32-NEXT:    vand.vv v8, v8, v16
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vmv.v.i v16, -1
 ; RV32-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV32-NEXT:    vredand.vs v8, v8, v16
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_and_v64i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
 ; RV64-NEXT:    addi a1, a0, 384
 ; RV64-NEXT:    vle64.v v16, (a1)
 ; RV64-NEXT:    addi a1, a0, 256
 ; RV64-NEXT:    addi a0, a0, 128
 ; RV64-NEXT:    vle64.v v24, (a0)
 ; RV64-NEXT:    vle64.v v0, (a1)
 ; RV64-NEXT:    vand.vv v16, v24, v16
 ; RV64-NEXT:    vand.vv v8, v8, v0
 ; RV64-NEXT:    vand.vv v8, v8, v16
-; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.i v16, -1
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vredand.vs v8, v8, v16
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <64 x i64>, <64 x i64>* %x
   %red = call i64 @llvm.vector.reduce.and.v64i64(<64 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, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetivli zero, 2, e8, mf8, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetivli zero, 4, e8, mf4, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetivli zero, 8, e8, mf2, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetivli zero, 16, e8, m1, ta, mu
 ; 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 i8 @llvm.vector.reduce.or.v32i8(<32 x i8>)
 
 define i8 @vreduce_or_v32i8(<32 x i8>* %x) {
 ; CHECK-LABEL: vreduce_or_v32i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m2, ta, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, 0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m2, ta, mu
 ; CHECK-NEXT:    vredor.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <32 x i8>, <32 x i8>* %x
   %red = call i8 @llvm.vector.reduce.or.v32i8(<32 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.or.v64i8(<64 x i8>)
 
 define i8 @vreduce_or_v64i8(<64 x i8>* %x) {
 ; CHECK-LABEL: vreduce_or_v64i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 64
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m4, ta, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v12, 0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m4, ta, mu
 ; CHECK-NEXT:    vredor.vs v8, v8, v12
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <64 x i8>, <64 x i8>* %x
   %red = call i8 @llvm.vector.reduce.or.v64i8(<64 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.or.v128i8(<128 x i8>)
 
 define i8 @vreduce_or_v128i8(<128 x i8>* %x) {
 ; CHECK-LABEL: vreduce_or_v128i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 128
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m8, ta, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v16, 0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m8, ta, mu
 ; CHECK-NEXT:    vredor.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <128 x i8>, <128 x i8>* %x
   %red = call i8 @llvm.vector.reduce.or.v128i8(<128 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.or.v256i8(<256 x i8>)
 
 define i8 @vreduce_or_v256i8(<256 x i8>* %x) {
 ; CHECK-LABEL: vreduce_or_v256i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 128
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m8, ta, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
 ; CHECK-NEXT:    addi a0, a0, 128
 ; CHECK-NEXT:    vle8.v v16, (a0)
 ; CHECK-NEXT:    vor.vv v8, v8, v16
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v16, 0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m8, ta, mu
 ; CHECK-NEXT:    vredor.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <256 x i8>, <256 x i8>* %x
   %red = call i8 @llvm.vector.reduce.or.v256i8(<256 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, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetivli zero, 8, e16, m1, ta, mu
 ; 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 i16 @llvm.vector.reduce.or.v16i16(<16 x i16>)
 
 define i16 @vreduce_or_v16i16(<16 x i16>* %x) {
 ; CHECK-LABEL: vreduce_or_v16i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 16, e16, m2, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, 0
 ; CHECK-NEXT:    vsetivli zero, 16, e16, m2, ta, mu
 ; CHECK-NEXT:    vredor.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <16 x i16>, <16 x i16>* %x
   %red = call i16 @llvm.vector.reduce.or.v16i16(<16 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.or.v32i16(<32 x i16>)
 
 define i16 @vreduce_or_v32i16(<32 x i16>* %x) {
 ; CHECK-LABEL: vreduce_or_v32i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m4, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v12, 0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m4, ta, mu
 ; CHECK-NEXT:    vredor.vs v8, v8, v12
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <32 x i16>, <32 x i16>* %x
   %red = call i16 @llvm.vector.reduce.or.v32i16(<32 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.or.v64i16(<64 x i16>)
 
 define i16 @vreduce_or_v64i16(<64 x i16>* %x) {
 ; CHECK-LABEL: vreduce_or_v64i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 64
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v16, 0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; CHECK-NEXT:    vredor.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <64 x i16>, <64 x i16>* %x
   %red = call i16 @llvm.vector.reduce.or.v64i16(<64 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.or.v128i16(<128 x i16>)
 
 define i16 @vreduce_or_v128i16(<128 x i16>* %x) {
 ; CHECK-LABEL: vreduce_or_v128i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 64
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
 ; CHECK-NEXT:    addi a0, a0, 128
 ; CHECK-NEXT:    vle16.v v16, (a0)
 ; CHECK-NEXT:    vor.vv v8, v8, v16
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v16, 0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; CHECK-NEXT:    vredor.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <128 x i16>, <128 x i16>* %x
   %red = call i16 @llvm.vector.reduce.or.v128i16(<128 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, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
 ; 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 i32 @llvm.vector.reduce.or.v8i32(<8 x i32>)
 
 define i32 @vreduce_or_v8i32(<8 x i32>* %x) {
 ; CHECK-LABEL: vreduce_or_v8i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 8, e32, m2, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, 0
 ; CHECK-NEXT:    vsetivli zero, 8, e32, m2, ta, mu
 ; CHECK-NEXT:    vredor.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <8 x i32>, <8 x i32>* %x
   %red = call i32 @llvm.vector.reduce.or.v8i32(<8 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.or.v16i32(<16 x i32>)
 
 define i32 @vreduce_or_v16i32(<16 x i32>* %x) {
 ; CHECK-LABEL: vreduce_or_v16i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 16, e32, m4, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v12, 0
 ; CHECK-NEXT:    vsetivli zero, 16, e32, m4, ta, mu
 ; CHECK-NEXT:    vredor.vs v8, v8, v12
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <16 x i32>, <16 x i32>* %x
   %red = call i32 @llvm.vector.reduce.or.v16i32(<16 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.or.v32i32(<32 x i32>)
 
 define i32 @vreduce_or_v32i32(<32 x i32>* %x) {
 ; CHECK-LABEL: vreduce_or_v32i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v16, 0
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
 ; CHECK-NEXT:    vredor.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <32 x i32>, <32 x i32>* %x
   %red = call i32 @llvm.vector.reduce.or.v32i32(<32 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.or.v64i32(<64 x i32>)
 
 define i32 @vreduce_or_v64i32(<64 x i32>* %x) {
 ; CHECK-LABEL: vreduce_or_v64i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
 ; CHECK-NEXT:    addi a0, a0, 128
 ; CHECK-NEXT:    vle32.v v16, (a0)
 ; CHECK-NEXT:    vor.vv v8, v8, v16
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v16, 0
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
 ; CHECK-NEXT:    vredor.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <64 x i32>, <64 x i32>* %x
   %red = call i32 @llvm.vector.reduce.or.v64i32(<64 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, mu
 ; RV32-NEXT:    vle64.v v8, (a0)
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vmv.v.i v9, 0
 ; RV32-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
 ; 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, mu
 ; 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, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
-; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.i v9, 0
 ; RV64-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
 ; 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 i64 @llvm.vector.reduce.or.v4i64(<4 x i64>)
 
 define i64 @vreduce_or_v4i64(<4 x i64>* %x) {
 ; RV32-LABEL: vreduce_or_v4i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; RV32-NEXT:    vle64.v v8, (a0)
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vmv.v.i v10, 0
 ; RV32-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; RV32-NEXT:    vredor.vs v8, v8, v10
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_or_v4i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
-; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.i v10, 0
 ; RV64-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; RV64-NEXT:    vredor.vs v8, v8, v10
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <4 x i64>, <4 x i64>* %x
   %red = call i64 @llvm.vector.reduce.or.v4i64(<4 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.or.v8i64(<8 x i64>)
 
 define i64 @vreduce_or_v8i64(<8 x i64>* %x) {
 ; RV32-LABEL: vreduce_or_v8i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
 ; RV32-NEXT:    vle64.v v8, (a0)
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vmv.v.i v12, 0
 ; RV32-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
 ; RV32-NEXT:    vredor.vs v8, v8, v12
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_or_v8i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
-; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.i v12, 0
 ; RV64-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
 ; RV64-NEXT:    vredor.vs v8, v8, v12
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <8 x i64>, <8 x i64>* %x
   %red = call i64 @llvm.vector.reduce.or.v8i64(<8 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.or.v16i64(<16 x i64>)
 
 define i64 @vreduce_or_v16i64(<16 x i64>* %x) {
 ; RV32-LABEL: vreduce_or_v16i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV32-NEXT:    vle64.v v8, (a0)
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vmv.v.i v16, 0
 ; RV32-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV32-NEXT:    vredor.vs v8, v8, v16
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_or_v16i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
-; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.i v16, 0
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vredor.vs v8, v8, v16
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <16 x i64>, <16 x i64>* %x
   %red = call i64 @llvm.vector.reduce.or.v16i64(<16 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.or.v32i64(<32 x i64>)
 
 define i64 @vreduce_or_v32i64(<32 x i64>* %x) {
 ; RV32-LABEL: vreduce_or_v32i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV32-NEXT:    vle64.v v8, (a0)
 ; RV32-NEXT:    addi a0, a0, 128
 ; RV32-NEXT:    vle64.v v16, (a0)
 ; RV32-NEXT:    vor.vv v8, v8, v16
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vmv.v.i v16, 0
 ; RV32-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV32-NEXT:    vredor.vs v8, v8, v16
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_or_v32i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
 ; RV64-NEXT:    addi a0, a0, 128
 ; RV64-NEXT:    vle64.v v16, (a0)
 ; RV64-NEXT:    vor.vv v8, v8, v16
-; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.i v16, 0
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vredor.vs v8, v8, v16
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <32 x i64>, <32 x i64>* %x
   %red = call i64 @llvm.vector.reduce.or.v32i64(<32 x i64> %v)
   ret i64 %red

 ; RV32-NEXT:    vle64.v v16, (a1)
 ; RV32-NEXT:    addi a1, a0, 256
 ; RV32-NEXT:    addi a0, a0, 128
 ; RV32-NEXT:    vle64.v v24, (a0)
 ; RV32-NEXT:    vle64.v v0, (a1)
 ; RV32-NEXT:    vor.vv v16, v24, v16
 ; RV32-NEXT:    vor.vv v8, v8, v0
 ; RV32-NEXT:    vor.vv v8, v8, v16
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vmv.v.i v16, 0
 ; RV32-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV32-NEXT:    vredor.vs v8, v8, v16
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_or_v64i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
 ; RV64-NEXT:    addi a1, a0, 384
 ; RV64-NEXT:    vle64.v v16, (a1)
 ; RV64-NEXT:    addi a1, a0, 256
 ; RV64-NEXT:    addi a0, a0, 128
 ; RV64-NEXT:    vle64.v v24, (a0)
 ; RV64-NEXT:    vle64.v v0, (a1)
 ; RV64-NEXT:    vor.vv v16, v24, v16
 ; RV64-NEXT:    vor.vv v8, v8, v0
 ; RV64-NEXT:    vor.vv v8, v8, v16
-; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.i v16, 0
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vredor.vs v8, v8, v16
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <64 x i64>, <64 x i64>* %x
   %red = call i64 @llvm.vector.reduce.or.v64i64(<64 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, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetivli zero, 2, e8, mf8, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetivli zero, 4, e8, mf4, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetivli zero, 8, e8, mf2, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetivli zero, 16, e8, m1, ta, mu
 ; 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 i8 @llvm.vector.reduce.xor.v32i8(<32 x i8>)
 
 define i8 @vreduce_xor_v32i8(<32 x i8>* %x) {
 ; CHECK-LABEL: vreduce_xor_v32i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m2, ta, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, 0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m2, ta, mu
 ; CHECK-NEXT:    vredxor.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <32 x i8>, <32 x i8>* %x
   %red = call i8 @llvm.vector.reduce.xor.v32i8(<32 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.xor.v64i8(<64 x i8>)
 
 define i8 @vreduce_xor_v64i8(<64 x i8>* %x) {
 ; CHECK-LABEL: vreduce_xor_v64i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 64
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m4, ta, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v12, 0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m4, ta, mu
 ; CHECK-NEXT:    vredxor.vs v8, v8, v12
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <64 x i8>, <64 x i8>* %x
   %red = call i8 @llvm.vector.reduce.xor.v64i8(<64 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.xor.v128i8(<128 x i8>)
 
 define i8 @vreduce_xor_v128i8(<128 x i8>* %x) {
 ; CHECK-LABEL: vreduce_xor_v128i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 128
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m8, ta, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v16, 0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m8, ta, mu
 ; CHECK-NEXT:    vredxor.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <128 x i8>, <128 x i8>* %x
   %red = call i8 @llvm.vector.reduce.xor.v128i8(<128 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.xor.v256i8(<256 x i8>)
 
 define i8 @vreduce_xor_v256i8(<256 x i8>* %x) {
 ; CHECK-LABEL: vreduce_xor_v256i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 128
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m8, ta, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
 ; CHECK-NEXT:    addi a0, a0, 128
 ; CHECK-NEXT:    vle8.v v16, (a0)
 ; CHECK-NEXT:    vxor.vv v8, v8, v16
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v16, 0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m8, ta, mu
 ; CHECK-NEXT:    vredxor.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <256 x i8>, <256 x i8>* %x
   %red = call i8 @llvm.vector.reduce.xor.v256i8(<256 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, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetivli zero, 8, e16, m1, ta, mu
 ; 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 i16 @llvm.vector.reduce.xor.v16i16(<16 x i16>)
 
 define i16 @vreduce_xor_v16i16(<16 x i16>* %x) {
 ; CHECK-LABEL: vreduce_xor_v16i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 16, e16, m2, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, 0
 ; CHECK-NEXT:    vsetivli zero, 16, e16, m2, ta, mu
 ; CHECK-NEXT:    vredxor.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <16 x i16>, <16 x i16>* %x
   %red = call i16 @llvm.vector.reduce.xor.v16i16(<16 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.xor.v32i16(<32 x i16>)
 
 define i16 @vreduce_xor_v32i16(<32 x i16>* %x) {
 ; CHECK-LABEL: vreduce_xor_v32i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m4, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v12, 0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m4, ta, mu
 ; CHECK-NEXT:    vredxor.vs v8, v8, v12
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <32 x i16>, <32 x i16>* %x
   %red = call i16 @llvm.vector.reduce.xor.v32i16(<32 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.xor.v64i16(<64 x i16>)
 
 define i16 @vreduce_xor_v64i16(<64 x i16>* %x) {
 ; CHECK-LABEL: vreduce_xor_v64i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 64
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v16, 0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; CHECK-NEXT:    vredxor.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <64 x i16>, <64 x i16>* %x
   %red = call i16 @llvm.vector.reduce.xor.v64i16(<64 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.xor.v128i16(<128 x i16>)
 
 define i16 @vreduce_xor_v128i16(<128 x i16>* %x) {
 ; CHECK-LABEL: vreduce_xor_v128i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 64
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
 ; CHECK-NEXT:    addi a0, a0, 128
 ; CHECK-NEXT:    vle16.v v16, (a0)
 ; CHECK-NEXT:    vxor.vv v8, v8, v16
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v16, 0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; CHECK-NEXT:    vredxor.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <128 x i16>, <128 x i16>* %x
   %red = call i16 @llvm.vector.reduce.xor.v128i16(<128 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, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
 ; 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 i32 @llvm.vector.reduce.xor.v8i32(<8 x i32>)
 
 define i32 @vreduce_xor_v8i32(<8 x i32>* %x) {
 ; CHECK-LABEL: vreduce_xor_v8i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 8, e32, m2, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, 0
 ; CHECK-NEXT:    vsetivli zero, 8, e32, m2, ta, mu
 ; CHECK-NEXT:    vredxor.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <8 x i32>, <8 x i32>* %x
   %red = call i32 @llvm.vector.reduce.xor.v8i32(<8 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.xor.v16i32(<16 x i32>)
 
 define i32 @vreduce_xor_v16i32(<16 x i32>* %x) {
 ; CHECK-LABEL: vreduce_xor_v16i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 16, e32, m4, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v12, 0
 ; CHECK-NEXT:    vsetivli zero, 16, e32, m4, ta, mu
 ; CHECK-NEXT:    vredxor.vs v8, v8, v12
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <16 x i32>, <16 x i32>* %x
   %red = call i32 @llvm.vector.reduce.xor.v16i32(<16 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.xor.v32i32(<32 x i32>)
 
 define i32 @vreduce_xor_v32i32(<32 x i32>* %x) {
 ; CHECK-LABEL: vreduce_xor_v32i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v16, 0
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
 ; CHECK-NEXT:    vredxor.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <32 x i32>, <32 x i32>* %x
   %red = call i32 @llvm.vector.reduce.xor.v32i32(<32 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.xor.v64i32(<64 x i32>)
 
 define i32 @vreduce_xor_v64i32(<64 x i32>* %x) {
 ; CHECK-LABEL: vreduce_xor_v64i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
 ; CHECK-NEXT:    addi a0, a0, 128
 ; CHECK-NEXT:    vle32.v v16, (a0)
 ; CHECK-NEXT:    vxor.vv v8, v8, v16
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v16, 0
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
 ; CHECK-NEXT:    vredxor.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <64 x i32>, <64 x i32>* %x
   %red = call i32 @llvm.vector.reduce.xor.v64i32(<64 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, mu
 ; RV32-NEXT:    vle64.v v8, (a0)
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vmv.v.i v9, 0
 ; RV32-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
 ; 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, mu
 ; 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, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
-; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.i v9, 0
 ; RV64-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
 ; 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 i64 @llvm.vector.reduce.xor.v4i64(<4 x i64>)
 
 define i64 @vreduce_xor_v4i64(<4 x i64>* %x) {
 ; RV32-LABEL: vreduce_xor_v4i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; RV32-NEXT:    vle64.v v8, (a0)
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vmv.v.i v10, 0
 ; RV32-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; RV32-NEXT:    vredxor.vs v8, v8, v10
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_xor_v4i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
-; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.i v10, 0
 ; RV64-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; RV64-NEXT:    vredxor.vs v8, v8, v10
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <4 x i64>, <4 x i64>* %x
   %red = call i64 @llvm.vector.reduce.xor.v4i64(<4 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.xor.v8i64(<8 x i64>)
 
 define i64 @vreduce_xor_v8i64(<8 x i64>* %x) {
 ; RV32-LABEL: vreduce_xor_v8i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
 ; RV32-NEXT:    vle64.v v8, (a0)
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vmv.v.i v12, 0
 ; RV32-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
 ; RV32-NEXT:    vredxor.vs v8, v8, v12
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_xor_v8i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
-; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.i v12, 0
 ; RV64-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
 ; RV64-NEXT:    vredxor.vs v8, v8, v12
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <8 x i64>, <8 x i64>* %x
   %red = call i64 @llvm.vector.reduce.xor.v8i64(<8 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.xor.v16i64(<16 x i64>)
 
 define i64 @vreduce_xor_v16i64(<16 x i64>* %x) {
 ; RV32-LABEL: vreduce_xor_v16i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV32-NEXT:    vle64.v v8, (a0)
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vmv.v.i v16, 0
 ; RV32-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV32-NEXT:    vredxor.vs v8, v8, v16
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_xor_v16i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
-; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.i v16, 0
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vredxor.vs v8, v8, v16
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <16 x i64>, <16 x i64>* %x
   %red = call i64 @llvm.vector.reduce.xor.v16i64(<16 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.xor.v32i64(<32 x i64>)
 
 define i64 @vreduce_xor_v32i64(<32 x i64>* %x) {
 ; RV32-LABEL: vreduce_xor_v32i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV32-NEXT:    vle64.v v8, (a0)
 ; RV32-NEXT:    addi a0, a0, 128
 ; RV32-NEXT:    vle64.v v16, (a0)
 ; RV32-NEXT:    vxor.vv v8, v8, v16
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vmv.v.i v16, 0
 ; RV32-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV32-NEXT:    vredxor.vs v8, v8, v16
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_xor_v32i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
 ; RV64-NEXT:    addi a0, a0, 128
 ; RV64-NEXT:    vle64.v v16, (a0)
 ; RV64-NEXT:    vxor.vv v8, v8, v16
-; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.i v16, 0
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vredxor.vs v8, v8, v16
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <32 x i64>, <32 x i64>* %x
   %red = call i64 @llvm.vector.reduce.xor.v32i64(<32 x i64> %v)
   ret i64 %red

 ; RV32-NEXT:    vle64.v v16, (a1)
 ; RV32-NEXT:    addi a1, a0, 256
 ; RV32-NEXT:    addi a0, a0, 128
 ; RV32-NEXT:    vle64.v v24, (a0)
 ; RV32-NEXT:    vle64.v v0, (a1)
 ; RV32-NEXT:    vxor.vv v16, v24, v16
 ; RV32-NEXT:    vxor.vv v8, v8, v0
 ; RV32-NEXT:    vxor.vv v8, v8, v16
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vmv.v.i v16, 0
 ; RV32-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV32-NEXT:    vredxor.vs v8, v8, v16
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_xor_v64i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
 ; RV64-NEXT:    addi a1, a0, 384
 ; RV64-NEXT:    vle64.v v16, (a1)
 ; RV64-NEXT:    addi a1, a0, 256
 ; RV64-NEXT:    addi a0, a0, 128
 ; RV64-NEXT:    vle64.v v24, (a0)
 ; RV64-NEXT:    vle64.v v0, (a1)
 ; RV64-NEXT:    vxor.vv v16, v24, v16
 ; RV64-NEXT:    vxor.vv v8, v8, v0
 ; RV64-NEXT:    vxor.vv v8, v8, v16
-; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.i v16, 0
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vredxor.vs v8, v8, v16
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <64 x i64>, <64 x i64>* %x
   %red = call i64 @llvm.vector.reduce.xor.v64i64(<64 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, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
 ; CHECK-NEXT:    li a0, 127
-; CHECK-NEXT:    vsetvli a1, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetivli zero, 2, e8, mf8, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
 ; CHECK-NEXT:    li a0, 127
-; CHECK-NEXT:    vsetvli a1, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetivli zero, 4, e8, mf4, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
 ; CHECK-NEXT:    li a0, 127
-; CHECK-NEXT:    vsetvli a1, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetivli zero, 8, e8, mf2, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
 ; CHECK-NEXT:    li a0, 127
-; CHECK-NEXT:    vsetvli a1, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetivli zero, 16, e8, m1, ta, mu
 ; 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
 }
 
 declare i8 @llvm.vector.reduce.smin.v32i8(<32 x i8>)
 
 define i8 @vreduce_smin_v32i8(<32 x i8>* %x) {
 ; CHECK-LABEL: vreduce_smin_v32i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m2, ta, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
 ; CHECK-NEXT:    li a0, 127
-; CHECK-NEXT:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v10, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m2, ta, mu
 ; CHECK-NEXT:    vredmin.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <32 x i8>, <32 x i8>* %x
   %red = call i8 @llvm.vector.reduce.smin.v32i8(<32 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.smin.v64i8(<64 x i8>)
 
 define i8 @vreduce_smin_v64i8(<64 x i8>* %x) {
 ; CHECK-LABEL: vreduce_smin_v64i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 64
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m4, ta, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
 ; CHECK-NEXT:    li a0, 127
-; CHECK-NEXT:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v12, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m4, ta, mu
 ; CHECK-NEXT:    vredmin.vs v8, v8, v12
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <64 x i8>, <64 x i8>* %x
   %red = call i8 @llvm.vector.reduce.smin.v64i8(<64 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.smin.v128i8(<128 x i8>)
 
 define i8 @vreduce_smin_v128i8(<128 x i8>* %x) {
 ; CHECK-LABEL: vreduce_smin_v128i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 128
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m8, ta, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
 ; CHECK-NEXT:    li a0, 127
-; CHECK-NEXT:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v16, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m8, ta, mu
 ; CHECK-NEXT:    vredmin.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <128 x i8>, <128 x i8>* %x
   %red = call i8 @llvm.vector.reduce.smin.v128i8(<128 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.smin.v256i8(<256 x i8>)
 
 define i8 @vreduce_smin_v256i8(<256 x i8>* %x) {
 ; CHECK-LABEL: vreduce_smin_v256i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 128
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m8, ta, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
 ; CHECK-NEXT:    addi a0, a0, 128
 ; CHECK-NEXT:    vle8.v v16, (a0)
 ; CHECK-NEXT:    vmin.vv v8, v8, v16
 ; CHECK-NEXT:    li a0, 127
-; CHECK-NEXT:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v16, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m8, ta, mu
 ; CHECK-NEXT:    vredmin.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <256 x i8>, <256 x i8>* %x
   %red = call i8 @llvm.vector.reduce.smin.v256i8(<256 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, mu
 ; RV32-NEXT:    vle16.v v8, (a0)
 ; RV32-NEXT:    lui a0, 8
 ; RV32-NEXT:    addi a0, a0, -1
-; RV32-NEXT:    vsetvli a1, zero, e16, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; RV32-NEXT:    vmv.v.x v9, a0
 ; RV32-NEXT:    vsetivli zero, 2, e16, mf4, ta, mu
 ; 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, mu
 ; RV64-NEXT:    vle16.v v8, (a0)
 ; RV64-NEXT:    lui a0, 8
 ; RV64-NEXT:    addiw a0, a0, -1
-; RV64-NEXT:    vsetvli a1, zero, e16, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vsetivli zero, 2, e16, mf4, ta, mu
 ; 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, mu
 ; RV32-NEXT:    vle16.v v8, (a0)
 ; RV32-NEXT:    lui a0, 8
 ; RV32-NEXT:    addi a0, a0, -1
-; RV32-NEXT:    vsetvli a1, zero, e16, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; RV32-NEXT:    vmv.v.x v9, a0
 ; RV32-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
 ; 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, mu
 ; RV64-NEXT:    vle16.v v8, (a0)
 ; RV64-NEXT:    lui a0, 8
 ; RV64-NEXT:    addiw a0, a0, -1
-; RV64-NEXT:    vsetvli a1, zero, e16, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
 ; 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, mu
 ; RV32-NEXT:    vle16.v v8, (a0)
 ; RV32-NEXT:    lui a0, 8
 ; RV32-NEXT:    addi a0, a0, -1
-; RV32-NEXT:    vsetvli a1, zero, e16, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; RV32-NEXT:    vmv.v.x v9, a0
 ; RV32-NEXT:    vsetivli zero, 8, e16, m1, ta, mu
 ; 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, mu
 ; RV64-NEXT:    vle16.v v8, (a0)
 ; RV64-NEXT:    lui a0, 8
 ; RV64-NEXT:    addiw a0, a0, -1
-; RV64-NEXT:    vsetvli a1, zero, e16, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vsetivli zero, 8, e16, m1, ta, mu
 ; 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
 }
 
 declare i16 @llvm.vector.reduce.smin.v16i16(<16 x i16>)
 
 define i16 @vreduce_smin_v16i16(<16 x i16>* %x) {
 ; RV32-LABEL: vreduce_smin_v16i16:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    vsetivli zero, 16, e16, m2, ta, mu
 ; RV32-NEXT:    vle16.v v8, (a0)
 ; RV32-NEXT:    lui a0, 8
 ; RV32-NEXT:    addi a0, a0, -1
-; RV32-NEXT:    vsetvli a1, zero, e16, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; RV32-NEXT:    vmv.v.x v10, a0
 ; RV32-NEXT:    vsetivli zero, 16, e16, m2, ta, mu
 ; RV32-NEXT:    vredmin.vs v8, v8, v10
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_smin_v16i16:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 16, e16, m2, ta, mu
 ; RV64-NEXT:    vle16.v v8, (a0)
 ; RV64-NEXT:    lui a0, 8
 ; RV64-NEXT:    addiw a0, a0, -1
-; RV64-NEXT:    vsetvli a1, zero, e16, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v10, a0
 ; RV64-NEXT:    vsetivli zero, 16, e16, m2, ta, mu
 ; RV64-NEXT:    vredmin.vs v8, v8, v10
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <16 x i16>, <16 x i16>* %x
   %red = call i16 @llvm.vector.reduce.smin.v16i16(<16 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.smin.v32i16(<32 x i16>)
 
 define i16 @vreduce_smin_v32i16(<32 x i16>* %x) {
 ; RV32-LABEL: vreduce_smin_v32i16:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetvli zero, a1, e16, m4, ta, mu
 ; RV32-NEXT:    vle16.v v8, (a0)
 ; RV32-NEXT:    lui a0, 8
 ; RV32-NEXT:    addi a0, a0, -1
-; RV32-NEXT:    vsetvli a2, zero, e16, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; RV32-NEXT:    vmv.v.x v12, a0
 ; RV32-NEXT:    vsetvli zero, a1, e16, m4, ta, mu
 ; RV32-NEXT:    vredmin.vs v8, v8, v12
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_smin_v32i16:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 32
 ; RV64-NEXT:    vsetvli zero, a1, e16, m4, ta, mu
 ; RV64-NEXT:    vle16.v v8, (a0)
 ; RV64-NEXT:    lui a0, 8
 ; RV64-NEXT:    addiw a0, a0, -1
-; RV64-NEXT:    vsetvli a2, zero, e16, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v12, a0
 ; RV64-NEXT:    vsetvli zero, a1, e16, m4, ta, mu
 ; RV64-NEXT:    vredmin.vs v8, v8, v12
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <32 x i16>, <32 x i16>* %x
   %red = call i16 @llvm.vector.reduce.smin.v32i16(<32 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.smin.v64i16(<64 x i16>)
 
 define i16 @vreduce_smin_v64i16(<64 x i16>* %x) {
 ; RV32-LABEL: vreduce_smin_v64i16:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    li a1, 64
 ; RV32-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; RV32-NEXT:    vle16.v v8, (a0)
 ; RV32-NEXT:    lui a0, 8
 ; RV32-NEXT:    addi a0, a0, -1
-; RV32-NEXT:    vsetvli a2, zero, e16, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; RV32-NEXT:    vmv.v.x v16, a0
 ; RV32-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; RV32-NEXT:    vredmin.vs v8, v8, v16
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_smin_v64i16:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 64
 ; RV64-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; RV64-NEXT:    vle16.v v8, (a0)
 ; RV64-NEXT:    lui a0, 8
 ; RV64-NEXT:    addiw a0, a0, -1
-; RV64-NEXT:    vsetvli a2, zero, e16, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v16, a0
 ; RV64-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; RV64-NEXT:    vredmin.vs v8, v8, v16
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <64 x i16>, <64 x i16>* %x
   %red = call i16 @llvm.vector.reduce.smin.v64i16(<64 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.smin.v128i16(<128 x i16>)
 
 define i16 @vreduce_smin_v128i16(<128 x i16>* %x) {
 ; RV32-LABEL: vreduce_smin_v128i16:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    li a1, 64
 ; RV32-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; RV32-NEXT:    vle16.v v8, (a0)
 ; RV32-NEXT:    addi a0, a0, 128
 ; RV32-NEXT:    vle16.v v16, (a0)
 ; RV32-NEXT:    vmin.vv v8, v8, v16
 ; RV32-NEXT:    lui a0, 8
 ; RV32-NEXT:    addi a0, a0, -1
-; RV32-NEXT:    vsetvli a2, zero, e16, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; RV32-NEXT:    vmv.v.x v16, a0
 ; RV32-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; RV32-NEXT:    vredmin.vs v8, v8, v16
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_smin_v128i16:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    li a1, 64
 ; RV64-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; RV64-NEXT:    vle16.v v8, (a0)
 ; RV64-NEXT:    addi a0, a0, 128
 ; RV64-NEXT:    vle16.v v16, (a0)
 ; RV64-NEXT:    vmin.vv v8, v8, v16
 ; RV64-NEXT:    lui a0, 8
 ; RV64-NEXT:    addiw a0, a0, -1
-; RV64-NEXT:    vsetvli a2, zero, e16, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v16, a0
 ; RV64-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; RV64-NEXT:    vredmin.vs v8, v8, v16
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <128 x i16>, <128 x i16>* %x
   %red = call i16 @llvm.vector.reduce.smin.v128i16(<128 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, mu
 ; RV32-NEXT:    vle32.v v8, (a0)
 ; RV32-NEXT:    lui a0, 524288
 ; RV32-NEXT:    addi a0, a0, -1
-; RV32-NEXT:    vsetvli a1, zero, e32, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; RV32-NEXT:    vmv.v.x v9, a0
 ; RV32-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
 ; 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, mu
 ; RV64-NEXT:    vle32.v v8, (a0)
 ; RV64-NEXT:    lui a0, 524288
 ; RV64-NEXT:    addiw a0, a0, -1
-; RV64-NEXT:    vsetvli a1, zero, e32, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
 ; 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, mu
 ; RV32-NEXT:    vle32.v v8, (a0)
 ; RV32-NEXT:    lui a0, 524288
 ; RV32-NEXT:    addi a0, a0, -1
-; RV32-NEXT:    vsetvli a1, zero, e32, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; RV32-NEXT:    vmv.v.x v9, a0
 ; RV32-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
 ; 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, mu
 ; RV64-NEXT:    vle32.v v8, (a0)
 ; RV64-NEXT:    lui a0, 524288
 ; RV64-NEXT:    addiw a0, a0, -1
-; RV64-NEXT:    vsetvli a1, zero, e32, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
 ; 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
 }
 
 declare i32 @llvm.vector.reduce.smin.v8i32(<8 x i32>)
 
 define i32 @vreduce_smin_v8i32(<8 x i32>* %x) {
 ; RV32-LABEL: vreduce_smin_v8i32:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    vsetivli zero, 8, e32, m2, ta, mu
 ; RV32-NEXT:    vle32.v v8, (a0)
 ; RV32-NEXT:    lui a0, 524288
 ; RV32-NEXT:    addi a0, a0, -1
-; RV32-NEXT:    vsetvli a1, zero, e32, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; RV32-NEXT:    vmv.v.x v10, a0
 ; RV32-NEXT:    vsetivli zero, 8, e32, m2, ta, mu
 ; RV32-NEXT:    vredmin.vs v8, v8, v10
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_smin_v8i32:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 8, e32, m2, ta, mu
 ; RV64-NEXT:    vle32.v v8, (a0)
 ; RV64-NEXT:    lui a0, 524288
 ; RV64-NEXT:    addiw a0, a0, -1
-; RV64-NEXT:    vsetvli a1, zero, e32, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v10, a0
 ; RV64-NEXT:    vsetivli zero, 8, e32, m2, ta, mu
 ; RV64-NEXT:    vredmin.vs v8, v8, v10
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <8 x i32>, <8 x i32>* %x
   %red = call i32 @llvm.vector.reduce.smin.v8i32(<8 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.smin.v16i32(<16 x i32>)
 
 define i32 @vreduce_smin_v16i32(<16 x i32>* %x) {
 ; RV32-LABEL: vreduce_smin_v16i32:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    vsetivli zero, 16, e32, m4, ta, mu
 ; RV32-NEXT:    vle32.v v8, (a0)
 ; RV32-NEXT:    lui a0, 524288
 ; RV32-NEXT:    addi a0, a0, -1
-; RV32-NEXT:    vsetvli a1, zero, e32, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; RV32-NEXT:    vmv.v.x v12, a0
 ; RV32-NEXT:    vsetivli zero, 16, e32, m4, ta, mu
 ; RV32-NEXT:    vredmin.vs v8, v8, v12
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_smin_v16i32:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 16, e32, m4, ta, mu
 ; RV64-NEXT:    vle32.v v8, (a0)
 ; RV64-NEXT:    lui a0, 524288
 ; RV64-NEXT:    addiw a0, a0, -1
-; RV64-NEXT:    vsetvli a1, zero, e32, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v12, a0
 ; RV64-NEXT:    vsetivli zero, 16, e32, m4, ta, mu
 ; RV64-NEXT:    vredmin.vs v8, v8, v12
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <16 x i32>, <16 x i32>* %x
   %red = call i32 @llvm.vector.reduce.smin.v16i32(<16 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.smin.v32i32(<32 x i32>)
 
 define i32 @vreduce_smin_v32i32(<32 x i32>* %x) {
 ; RV32-LABEL: vreduce_smin_v32i32:
 ; 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:    lui a0, 524288
 ; RV32-NEXT:    addi a0, a0, -1
-; RV32-NEXT:    vsetvli a2, zero, e32, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; RV32-NEXT:    vmv.v.x v16, a0
 ; RV32-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
 ; RV32-NEXT:    vredmin.vs v8, v8, v16
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_smin_v32i32:
 ; 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:    lui a0, 524288
 ; RV64-NEXT:    addiw a0, a0, -1
-; RV64-NEXT:    vsetvli a2, zero, e32, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v16, a0
 ; RV64-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
 ; RV64-NEXT:    vredmin.vs v8, v8, v16
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <32 x i32>, <32 x i32>* %x
   %red = call i32 @llvm.vector.reduce.smin.v32i32(<32 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.smin.v64i32(<64 x i32>)
 
 define i32 @vreduce_smin_v64i32(<64 x i32>* %x) {
 ; RV32-LABEL: vreduce_smin_v64i32:
 ; 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:    addi a0, a0, 128
 ; RV32-NEXT:    vle32.v v16, (a0)
 ; RV32-NEXT:    vmin.vv v8, v8, v16
 ; RV32-NEXT:    lui a0, 524288
 ; RV32-NEXT:    addi a0, a0, -1
-; RV32-NEXT:    vsetvli a2, zero, e32, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; RV32-NEXT:    vmv.v.x v16, a0
 ; RV32-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
 ; RV32-NEXT:    vredmin.vs v8, v8, v16
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_smin_v64i32:
 ; 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:    addi a0, a0, 128
 ; RV64-NEXT:    vle32.v v16, (a0)
 ; RV64-NEXT:    vmin.vv v8, v8, v16
 ; RV64-NEXT:    lui a0, 524288
 ; RV64-NEXT:    addiw a0, a0, -1
-; RV64-NEXT:    vsetvli a2, zero, e32, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v16, a0
 ; RV64-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
 ; RV64-NEXT:    vredmin.vs v8, v8, v16
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <64 x i32>, <64 x i32>* %x
   %red = call i32 @llvm.vector.reduce.smin.v64i32(<64 x i32> %v)
   ret i32 %red

 ; RV32-NEXT:    .cfi_def_cfa_offset 16
 ; RV32-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v9, (a0), zero
 ; RV32-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
 ; 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, mu
 ; 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, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
 ; RV64-NEXT:    li a0, -1
 ; RV64-NEXT:    srli a0, a0, 1
-; RV64-NEXT:    vsetvli a1, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
 ; 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 i64 @llvm.vector.reduce.smin.v4i64(<4 x i64>)
 
 define i64 @vreduce_smin_v4i64(<4 x i64>* %x) {
 ; RV32-LABEL: vreduce_smin_v4i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    addi sp, sp, -16
 ; RV32-NEXT:    .cfi_def_cfa_offset 16
 ; RV32-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v10, (a0), zero
 ; RV32-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; RV32-NEXT:    vredmin.vs v8, v8, v10
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; 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_v4i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
 ; RV64-NEXT:    li a0, -1
 ; RV64-NEXT:    srli a0, a0, 1
-; RV64-NEXT:    vsetvli a1, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v10, a0
 ; RV64-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; RV64-NEXT:    vredmin.vs v8, v8, v10
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <4 x i64>, <4 x i64>* %x
   %red = call i64 @llvm.vector.reduce.smin.v4i64(<4 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.smin.v8i64(<8 x i64>)
 
 define i64 @vreduce_smin_v8i64(<8 x i64>* %x) {
 ; RV32-LABEL: vreduce_smin_v8i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    addi sp, sp, -16
 ; RV32-NEXT:    .cfi_def_cfa_offset 16
 ; RV32-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v12, (a0), zero
 ; RV32-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
 ; RV32-NEXT:    vredmin.vs v8, v8, v12
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; 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_v8i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
 ; RV64-NEXT:    li a0, -1
 ; RV64-NEXT:    srli a0, a0, 1
-; RV64-NEXT:    vsetvli a1, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v12, a0
 ; RV64-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
 ; RV64-NEXT:    vredmin.vs v8, v8, v12
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <8 x i64>, <8 x i64>* %x
   %red = call i64 @llvm.vector.reduce.smin.v8i64(<8 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.smin.v16i64(<16 x i64>)
 
 define i64 @vreduce_smin_v16i64(<16 x i64>* %x) {
 ; RV32-LABEL: vreduce_smin_v16i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    addi sp, sp, -16
 ; RV32-NEXT:    .cfi_def_cfa_offset 16
 ; RV32-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v16, (a0), zero
 ; RV32-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV32-NEXT:    vredmin.vs v8, v8, v16
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; 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_v16i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
 ; RV64-NEXT:    li a0, -1
 ; RV64-NEXT:    srli a0, a0, 1
-; RV64-NEXT:    vsetvli a1, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v16, a0
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vredmin.vs v8, v8, v16
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <16 x i64>, <16 x i64>* %x
   %red = call i64 @llvm.vector.reduce.smin.v16i64(<16 x i64> %v)
   ret i64 %red

 ; RV32-NEXT:    addi a0, a0, 128
 ; RV32-NEXT:    vle64.v v16, (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:    vmin.vv v8, v8, v16
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v16, (a0), zero
 ; RV32-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV32-NEXT:    vredmin.vs v8, v8, v16
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; 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_v32i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
 ; RV64-NEXT:    addi a0, a0, 128
 ; RV64-NEXT:    vle64.v v16, (a0)
 ; RV64-NEXT:    vmin.vv v8, v8, v16
 ; RV64-NEXT:    li a0, -1
 ; RV64-NEXT:    srli a0, a0, 1
-; RV64-NEXT:    vsetvli a1, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v16, a0
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vredmin.vs v8, v8, v16
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <32 x i64>, <32 x i64>* %x
   %red = call i64 @llvm.vector.reduce.smin.v32i64(<32 x i64> %v)
   ret i64 %red

 ; 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:    vmin.vv v24, v0, v24
 ; RV32-NEXT:    vmin.vv v8, v8, v16
 ; RV32-NEXT:    vmin.vv v8, v8, v24
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v16, (a0), zero
 ; RV32-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV32-NEXT:    vredmin.vs v8, v8, v16
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v8, a1

 ; RV64-NEXT:    addi a0, a0, 128
 ; RV64-NEXT:    vle64.v v24, (a0)
 ; RV64-NEXT:    vle64.v v0, (a1)
 ; RV64-NEXT:    vmin.vv v16, v24, v16
 ; RV64-NEXT:    vmin.vv v8, v8, v0
 ; RV64-NEXT:    vmin.vv v8, v8, v16
 ; RV64-NEXT:    li a0, -1
 ; RV64-NEXT:    srli a0, a0, 1
-; RV64-NEXT:    vsetvli a1, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v16, a0
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vredmin.vs v8, v8, v16
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <64 x i64>, <64 x i64>* %x
   %red = call i64 @llvm.vector.reduce.smin.v64i64(<64 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, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
 ; CHECK-NEXT:    li a0, -128
-; CHECK-NEXT:    vsetvli a1, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetivli zero, 2, e8, mf8, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
 ; CHECK-NEXT:    li a0, -128
-; CHECK-NEXT:    vsetvli a1, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetivli zero, 4, e8, mf4, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
 ; CHECK-NEXT:    li a0, -128
-; CHECK-NEXT:    vsetvli a1, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetivli zero, 8, e8, mf2, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
 ; CHECK-NEXT:    li a0, -128
-; CHECK-NEXT:    vsetvli a1, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetivli zero, 16, e8, m1, ta, mu
 ; 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 i8 @llvm.vector.reduce.smax.v32i8(<32 x i8>)
 
 define i8 @vreduce_smax_v32i8(<32 x i8>* %x) {
 ; CHECK-LABEL: vreduce_smax_v32i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m2, ta, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
 ; CHECK-NEXT:    li a0, -128
-; CHECK-NEXT:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v10, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m2, ta, mu
 ; CHECK-NEXT:    vredmax.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <32 x i8>, <32 x i8>* %x
   %red = call i8 @llvm.vector.reduce.smax.v32i8(<32 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.smax.v64i8(<64 x i8>)
 
 define i8 @vreduce_smax_v64i8(<64 x i8>* %x) {
 ; CHECK-LABEL: vreduce_smax_v64i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 64
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m4, ta, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
 ; CHECK-NEXT:    li a0, -128
-; CHECK-NEXT:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v12, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m4, ta, mu
 ; CHECK-NEXT:    vredmax.vs v8, v8, v12
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <64 x i8>, <64 x i8>* %x
   %red = call i8 @llvm.vector.reduce.smax.v64i8(<64 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.smax.v128i8(<128 x i8>)
 
 define i8 @vreduce_smax_v128i8(<128 x i8>* %x) {
 ; CHECK-LABEL: vreduce_smax_v128i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 128
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m8, ta, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
 ; CHECK-NEXT:    li a0, -128
-; CHECK-NEXT:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v16, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m8, ta, mu
 ; CHECK-NEXT:    vredmax.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <128 x i8>, <128 x i8>* %x
   %red = call i8 @llvm.vector.reduce.smax.v128i8(<128 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.smax.v256i8(<256 x i8>)
 
 define i8 @vreduce_smax_v256i8(<256 x i8>* %x) {
 ; CHECK-LABEL: vreduce_smax_v256i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 128
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m8, ta, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
 ; CHECK-NEXT:    addi a0, a0, 128
 ; CHECK-NEXT:    vle8.v v16, (a0)
 ; CHECK-NEXT:    vmax.vv v8, v8, v16
 ; CHECK-NEXT:    li a0, -128
-; CHECK-NEXT:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v16, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m8, ta, mu
 ; CHECK-NEXT:    vredmax.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <256 x i8>, <256 x i8>* %x
   %red = call i8 @llvm.vector.reduce.smax.v256i8(<256 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, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
 ; CHECK-NEXT:    lui a0, 1048568
-; CHECK-NEXT:    vsetvli a1, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
 ; CHECK-NEXT:    lui a0, 1048568
-; CHECK-NEXT:    vsetvli a1, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
 ; CHECK-NEXT:    lui a0, 1048568
-; CHECK-NEXT:    vsetvli a1, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetivli zero, 8, e16, m1, ta, mu
 ; 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 i16 @llvm.vector.reduce.smax.v16i16(<16 x i16>)
 
 define i16 @vreduce_smax_v16i16(<16 x i16>* %x) {
 ; CHECK-LABEL: vreduce_smax_v16i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 16, e16, m2, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
 ; CHECK-NEXT:    lui a0, 1048568
-; CHECK-NEXT:    vsetvli a1, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v10, a0
 ; CHECK-NEXT:    vsetivli zero, 16, e16, m2, ta, mu
 ; CHECK-NEXT:    vredmax.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <16 x i16>, <16 x i16>* %x
   %red = call i16 @llvm.vector.reduce.smax.v16i16(<16 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.smax.v32i16(<32 x i16>)
 
 define i16 @vreduce_smax_v32i16(<32 x i16>* %x) {
 ; CHECK-LABEL: vreduce_smax_v32i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m4, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
 ; CHECK-NEXT:    lui a0, 1048568
-; CHECK-NEXT:    vsetvli a2, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v12, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m4, ta, mu
 ; CHECK-NEXT:    vredmax.vs v8, v8, v12
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <32 x i16>, <32 x i16>* %x
   %red = call i16 @llvm.vector.reduce.smax.v32i16(<32 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.smax.v64i16(<64 x i16>)
 
 define i16 @vreduce_smax_v64i16(<64 x i16>* %x) {
 ; CHECK-LABEL: vreduce_smax_v64i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 64
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
 ; CHECK-NEXT:    lui a0, 1048568
-; CHECK-NEXT:    vsetvli a2, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v16, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; CHECK-NEXT:    vredmax.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <64 x i16>, <64 x i16>* %x
   %red = call i16 @llvm.vector.reduce.smax.v64i16(<64 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.smax.v128i16(<128 x i16>)
 
 define i16 @vreduce_smax_v128i16(<128 x i16>* %x) {
 ; CHECK-LABEL: vreduce_smax_v128i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 64
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
 ; CHECK-NEXT:    addi a0, a0, 128
 ; CHECK-NEXT:    vle16.v v16, (a0)
 ; CHECK-NEXT:    vmax.vv v8, v8, v16
 ; CHECK-NEXT:    lui a0, 1048568
-; CHECK-NEXT:    vsetvli a2, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v16, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; CHECK-NEXT:    vredmax.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <128 x i16>, <128 x i16>* %x
   %red = call i16 @llvm.vector.reduce.smax.v128i16(<128 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, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
 ; CHECK-NEXT:    lui a0, 524288
-; CHECK-NEXT:    vsetvli a1, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
 ; CHECK-NEXT:    lui a0, 524288
-; CHECK-NEXT:    vsetvli a1, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
 ; 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
 }
 
 declare i32 @llvm.vector.reduce.smax.v8i32(<8 x i32>)
 
 define i32 @vreduce_smax_v8i32(<8 x i32>* %x) {
 ; CHECK-LABEL: vreduce_smax_v8i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 8, e32, m2, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
 ; CHECK-NEXT:    lui a0, 524288
-; CHECK-NEXT:    vsetvli a1, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v10, a0
 ; CHECK-NEXT:    vsetivli zero, 8, e32, m2, ta, mu
 ; CHECK-NEXT:    vredmax.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <8 x i32>, <8 x i32>* %x
   %red = call i32 @llvm.vector.reduce.smax.v8i32(<8 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.smax.v16i32(<16 x i32>)
 
 define i32 @vreduce_smax_v16i32(<16 x i32>* %x) {
 ; CHECK-LABEL: vreduce_smax_v16i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 16, e32, m4, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
 ; CHECK-NEXT:    lui a0, 524288
-; CHECK-NEXT:    vsetvli a1, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v12, a0
 ; CHECK-NEXT:    vsetivli zero, 16, e32, m4, ta, mu
 ; CHECK-NEXT:    vredmax.vs v8, v8, v12
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <16 x i32>, <16 x i32>* %x
   %red = call i32 @llvm.vector.reduce.smax.v16i32(<16 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.smax.v32i32(<32 x i32>)
 
 define i32 @vreduce_smax_v32i32(<32 x i32>* %x) {
 ; CHECK-LABEL: vreduce_smax_v32i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
 ; CHECK-NEXT:    lui a0, 524288
-; CHECK-NEXT:    vsetvli a2, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v16, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
 ; CHECK-NEXT:    vredmax.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <32 x i32>, <32 x i32>* %x
   %red = call i32 @llvm.vector.reduce.smax.v32i32(<32 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.smax.v64i32(<64 x i32>)
 
 define i32 @vreduce_smax_v64i32(<64 x i32>* %x) {
 ; CHECK-LABEL: vreduce_smax_v64i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
 ; CHECK-NEXT:    addi a0, a0, 128
 ; CHECK-NEXT:    vle32.v v16, (a0)
 ; CHECK-NEXT:    vmax.vv v8, v8, v16
 ; CHECK-NEXT:    lui a0, 524288
-; CHECK-NEXT:    vsetvli a2, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v16, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
 ; CHECK-NEXT:    vredmax.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <64 x i32>, <64 x i32>* %x
   %red = call i32 @llvm.vector.reduce.smax.v64i32(<64 x i32> %v)
   ret i32 %red

 ; RV32:       # %bb.0:
 ; RV32-NEXT:    addi sp, sp, -16
 ; RV32-NEXT:    .cfi_def_cfa_offset 16
 ; RV32-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v9, (a0), zero
 ; RV32-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
 ; 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, mu
 ; 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, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
 ; RV64-NEXT:    li a0, -1
 ; RV64-NEXT:    slli a0, a0, 63
-; RV64-NEXT:    vsetvli a1, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
 ; 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 i64 @llvm.vector.reduce.smax.v4i64(<4 x i64>)
 
 define i64 @vreduce_smax_v4i64(<4 x i64>* %x) {
 ; RV32-LABEL: vreduce_smax_v4i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    addi sp, sp, -16
 ; RV32-NEXT:    .cfi_def_cfa_offset 16
 ; RV32-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v10, (a0), zero
 ; RV32-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; RV32-NEXT:    vredmax.vs v8, v8, v10
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; 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_v4i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
 ; RV64-NEXT:    li a0, -1
 ; RV64-NEXT:    slli a0, a0, 63
-; RV64-NEXT:    vsetvli a1, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v10, a0
 ; RV64-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; RV64-NEXT:    vredmax.vs v8, v8, v10
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <4 x i64>, <4 x i64>* %x
   %red = call i64 @llvm.vector.reduce.smax.v4i64(<4 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.smax.v8i64(<8 x i64>)
 
 define i64 @vreduce_smax_v8i64(<8 x i64>* %x) {
 ; RV32-LABEL: vreduce_smax_v8i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    addi sp, sp, -16
 ; RV32-NEXT:    .cfi_def_cfa_offset 16
 ; RV32-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v12, (a0), zero
 ; RV32-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
 ; RV32-NEXT:    vredmax.vs v8, v8, v12
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; 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_v8i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
 ; RV64-NEXT:    li a0, -1
 ; RV64-NEXT:    slli a0, a0, 63
-; RV64-NEXT:    vsetvli a1, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v12, a0
 ; RV64-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
 ; RV64-NEXT:    vredmax.vs v8, v8, v12
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <8 x i64>, <8 x i64>* %x
   %red = call i64 @llvm.vector.reduce.smax.v8i64(<8 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.smax.v16i64(<16 x i64>)
 
 define i64 @vreduce_smax_v16i64(<16 x i64>* %x) {
 ; RV32-LABEL: vreduce_smax_v16i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    addi sp, sp, -16
 ; RV32-NEXT:    .cfi_def_cfa_offset 16
 ; RV32-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v16, (a0), zero
 ; RV32-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV32-NEXT:    vredmax.vs v8, v8, v16
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; 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_v16i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
 ; RV64-NEXT:    li a0, -1
 ; RV64-NEXT:    slli a0, a0, 63
-; RV64-NEXT:    vsetvli a1, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v16, a0
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vredmax.vs v8, v8, v16
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <16 x i64>, <16 x i64>* %x
   %red = call i64 @llvm.vector.reduce.smax.v16i64(<16 x i64> %v)
   ret i64 %red

 ; RV32-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV32-NEXT:    vle64.v v8, (a0)
 ; RV32-NEXT:    addi a0, a0, 128
 ; RV32-NEXT:    vle64.v v16, (a0)
 ; RV32-NEXT:    lui a0, 524288
 ; RV32-NEXT:    sw a0, 12(sp)
 ; RV32-NEXT:    sw zero, 8(sp)
 ; RV32-NEXT:    vmax.vv v8, v8, v16
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v16, (a0), zero
 ; RV32-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV32-NEXT:    vredmax.vs v8, v8, v16
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; 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_v32i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
 ; RV64-NEXT:    addi a0, a0, 128
 ; RV64-NEXT:    vle64.v v16, (a0)
 ; RV64-NEXT:    vmax.vv v8, v8, v16
 ; RV64-NEXT:    li a0, -1
 ; RV64-NEXT:    slli a0, a0, 63
-; RV64-NEXT:    vsetvli a1, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v16, a0
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vredmax.vs v8, v8, v16
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <32 x i64>, <32 x i64>* %x
   %red = call i64 @llvm.vector.reduce.smax.v32i64(<32 x i64> %v)
   ret i64 %red

 ; RV32-NEXT:    addi a0, a0, 128
 ; RV32-NEXT:    vle64.v v0, (a0)
 ; RV32-NEXT:    lui a0, 524288
 ; RV32-NEXT:    sw a0, 12(sp)
 ; RV32-NEXT:    sw zero, 8(sp)
 ; RV32-NEXT:    vmax.vv v24, v0, v24
 ; RV32-NEXT:    vmax.vv v8, v8, v16
 ; RV32-NEXT:    vmax.vv v8, v8, v24
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v16, (a0), zero
 ; RV32-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV32-NEXT:    vredmax.vs v8, v8, v16
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v8, a1

 ; RV64-NEXT:    addi a0, a0, 128
 ; RV64-NEXT:    vle64.v v24, (a0)
 ; RV64-NEXT:    vle64.v v0, (a1)
 ; RV64-NEXT:    vmax.vv v16, v24, v16
 ; RV64-NEXT:    vmax.vv v8, v8, v0
 ; RV64-NEXT:    vmax.vv v8, v8, v16
 ; RV64-NEXT:    li a0, -1
 ; RV64-NEXT:    slli a0, a0, 63
-; RV64-NEXT:    vsetvli a1, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v16, a0
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vredmax.vs v8, v8, v16
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <64 x i64>, <64 x i64>* %x
   %red = call i64 @llvm.vector.reduce.smax.v64i64(<64 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, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetivli zero, 2, e8, mf8, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetivli zero, 4, e8, mf4, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetivli zero, 8, e8, mf2, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetivli zero, 16, e8, m1, ta, mu
 ; 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 i8 @llvm.vector.reduce.umin.v32i8(<32 x i8>)
 
 define i8 @vreduce_umin_v32i8(<32 x i8>* %x) {
 ; CHECK-LABEL: vreduce_umin_v32i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m2, ta, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, -1
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m2, ta, mu
 ; CHECK-NEXT:    vredminu.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <32 x i8>, <32 x i8>* %x
   %red = call i8 @llvm.vector.reduce.umin.v32i8(<32 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.umin.v64i8(<64 x i8>)
 
 define i8 @vreduce_umin_v64i8(<64 x i8>* %x) {
 ; CHECK-LABEL: vreduce_umin_v64i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 64
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m4, ta, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v12, -1
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m4, ta, mu
 ; CHECK-NEXT:    vredminu.vs v8, v8, v12
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <64 x i8>, <64 x i8>* %x
   %red = call i8 @llvm.vector.reduce.umin.v64i8(<64 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.umin.v128i8(<128 x i8>)
 
 define i8 @vreduce_umin_v128i8(<128 x i8>* %x) {
 ; CHECK-LABEL: vreduce_umin_v128i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 128
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m8, ta, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v16, -1
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m8, ta, mu
 ; CHECK-NEXT:    vredminu.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <128 x i8>, <128 x i8>* %x
   %red = call i8 @llvm.vector.reduce.umin.v128i8(<128 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.umin.v256i8(<256 x i8>)
 
 define i8 @vreduce_umin_v256i8(<256 x i8>* %x) {
 ; CHECK-LABEL: vreduce_umin_v256i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 128
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m8, ta, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
 ; CHECK-NEXT:    addi a0, a0, 128
 ; CHECK-NEXT:    vle8.v v16, (a0)
 ; CHECK-NEXT:    vminu.vv v8, v8, v16
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v16, -1
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m8, ta, mu
 ; CHECK-NEXT:    vredminu.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <256 x i8>, <256 x i8>* %x
   %red = call i8 @llvm.vector.reduce.umin.v256i8(<256 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, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetivli zero, 8, e16, m1, ta, mu
 ; 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 i16 @llvm.vector.reduce.umin.v16i16(<16 x i16>)
 
 define i16 @vreduce_umin_v16i16(<16 x i16>* %x) {
 ; CHECK-LABEL: vreduce_umin_v16i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 16, e16, m2, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, -1
 ; CHECK-NEXT:    vsetivli zero, 16, e16, m2, ta, mu
 ; CHECK-NEXT:    vredminu.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <16 x i16>, <16 x i16>* %x
   %red = call i16 @llvm.vector.reduce.umin.v16i16(<16 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.umin.v32i16(<32 x i16>)
 
 define i16 @vreduce_umin_v32i16(<32 x i16>* %x) {
 ; CHECK-LABEL: vreduce_umin_v32i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m4, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v12, -1
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m4, ta, mu
 ; CHECK-NEXT:    vredminu.vs v8, v8, v12
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <32 x i16>, <32 x i16>* %x
   %red = call i16 @llvm.vector.reduce.umin.v32i16(<32 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.umin.v64i16(<64 x i16>)
 
 define i16 @vreduce_umin_v64i16(<64 x i16>* %x) {
 ; CHECK-LABEL: vreduce_umin_v64i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 64
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v16, -1
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; CHECK-NEXT:    vredminu.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <64 x i16>, <64 x i16>* %x
   %red = call i16 @llvm.vector.reduce.umin.v64i16(<64 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.umin.v128i16(<128 x i16>)
 
 define i16 @vreduce_umin_v128i16(<128 x i16>* %x) {
 ; CHECK-LABEL: vreduce_umin_v128i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 64
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
 ; CHECK-NEXT:    addi a0, a0, 128
 ; CHECK-NEXT:    vle16.v v16, (a0)
 ; CHECK-NEXT:    vminu.vv v8, v8, v16
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v16, -1
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; CHECK-NEXT:    vredminu.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <128 x i16>, <128 x i16>* %x
   %red = call i16 @llvm.vector.reduce.umin.v128i16(<128 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, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
 ; 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 i32 @llvm.vector.reduce.umin.v8i32(<8 x i32>)
 
 define i32 @vreduce_umin_v8i32(<8 x i32>* %x) {
 ; CHECK-LABEL: vreduce_umin_v8i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 8, e32, m2, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, -1
 ; CHECK-NEXT:    vsetivli zero, 8, e32, m2, ta, mu
 ; CHECK-NEXT:    vredminu.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <8 x i32>, <8 x i32>* %x
   %red = call i32 @llvm.vector.reduce.umin.v8i32(<8 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.umin.v16i32(<16 x i32>)
 
 define i32 @vreduce_umin_v16i32(<16 x i32>* %x) {
 ; CHECK-LABEL: vreduce_umin_v16i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 16, e32, m4, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v12, -1
 ; CHECK-NEXT:    vsetivli zero, 16, e32, m4, ta, mu
 ; CHECK-NEXT:    vredminu.vs v8, v8, v12
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <16 x i32>, <16 x i32>* %x
   %red = call i32 @llvm.vector.reduce.umin.v16i32(<16 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.umin.v32i32(<32 x i32>)
 
 define i32 @vreduce_umin_v32i32(<32 x i32>* %x) {
 ; CHECK-LABEL: vreduce_umin_v32i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v16, -1
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
 ; CHECK-NEXT:    vredminu.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <32 x i32>, <32 x i32>* %x
   %red = call i32 @llvm.vector.reduce.umin.v32i32(<32 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.umin.v64i32(<64 x i32>)
 
 define i32 @vreduce_umin_v64i32(<64 x i32>* %x) {
 ; CHECK-LABEL: vreduce_umin_v64i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
 ; CHECK-NEXT:    addi a0, a0, 128
 ; CHECK-NEXT:    vle32.v v16, (a0)
 ; CHECK-NEXT:    vminu.vv v8, v8, v16
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v16, -1
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
 ; CHECK-NEXT:    vredminu.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <64 x i32>, <64 x i32>* %x
   %red = call i32 @llvm.vector.reduce.umin.v64i32(<64 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, mu
 ; RV32-NEXT:    vle64.v v8, (a0)
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vmv.v.i v9, -1
 ; RV32-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
 ; 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, mu
 ; 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, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
-; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.i v9, -1
 ; RV64-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
 ; 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 i64 @llvm.vector.reduce.umin.v4i64(<4 x i64>)
 
 define i64 @vreduce_umin_v4i64(<4 x i64>* %x) {
 ; RV32-LABEL: vreduce_umin_v4i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; RV32-NEXT:    vle64.v v8, (a0)
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vmv.v.i v10, -1
 ; RV32-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; RV32-NEXT:    vredminu.vs v8, v8, v10
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_umin_v4i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
-; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.i v10, -1
 ; RV64-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; RV64-NEXT:    vredminu.vs v8, v8, v10
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <4 x i64>, <4 x i64>* %x
   %red = call i64 @llvm.vector.reduce.umin.v4i64(<4 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.umin.v8i64(<8 x i64>)
 
 define i64 @vreduce_umin_v8i64(<8 x i64>* %x) {
 ; RV32-LABEL: vreduce_umin_v8i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
 ; RV32-NEXT:    vle64.v v8, (a0)
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vmv.v.i v12, -1
 ; RV32-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
 ; RV32-NEXT:    vredminu.vs v8, v8, v12
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_umin_v8i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
-; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.i v12, -1
 ; RV64-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
 ; RV64-NEXT:    vredminu.vs v8, v8, v12
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <8 x i64>, <8 x i64>* %x
   %red = call i64 @llvm.vector.reduce.umin.v8i64(<8 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.umin.v16i64(<16 x i64>)
 
 define i64 @vreduce_umin_v16i64(<16 x i64>* %x) {
 ; RV32-LABEL: vreduce_umin_v16i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV32-NEXT:    vle64.v v8, (a0)
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vmv.v.i v16, -1
 ; RV32-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV32-NEXT:    vredminu.vs v8, v8, v16
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_umin_v16i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
-; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.i v16, -1
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vredminu.vs v8, v8, v16
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <16 x i64>, <16 x i64>* %x
   %red = call i64 @llvm.vector.reduce.umin.v16i64(<16 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.umin.v32i64(<32 x i64>)
 
 define i64 @vreduce_umin_v32i64(<32 x i64>* %x) {
 ; RV32-LABEL: vreduce_umin_v32i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV32-NEXT:    vle64.v v8, (a0)
 ; RV32-NEXT:    addi a0, a0, 128
 ; RV32-NEXT:    vle64.v v16, (a0)
 ; RV32-NEXT:    vminu.vv v8, v8, v16
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vmv.v.i v16, -1
 ; RV32-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV32-NEXT:    vredminu.vs v8, v8, v16
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_umin_v32i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
 ; RV64-NEXT:    addi a0, a0, 128
 ; RV64-NEXT:    vle64.v v16, (a0)
 ; RV64-NEXT:    vminu.vv v8, v8, v16
-; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.i v16, -1
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vredminu.vs v8, v8, v16
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <32 x i64>, <32 x i64>* %x
   %red = call i64 @llvm.vector.reduce.umin.v32i64(<32 x i64> %v)
   ret i64 %red

 ; RV32-NEXT:    vle64.v v16, (a1)
 ; RV32-NEXT:    addi a1, a0, 256
 ; RV32-NEXT:    addi a0, a0, 128
 ; RV32-NEXT:    vle64.v v24, (a0)
 ; RV32-NEXT:    vle64.v v0, (a1)
 ; RV32-NEXT:    vminu.vv v16, v24, v16
 ; RV32-NEXT:    vminu.vv v8, v8, v0
 ; RV32-NEXT:    vminu.vv v8, v8, v16
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vmv.v.i v16, -1
 ; RV32-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV32-NEXT:    vredminu.vs v8, v8, v16
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_umin_v64i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
 ; RV64-NEXT:    addi a1, a0, 384
 ; RV64-NEXT:    vle64.v v16, (a1)
 ; RV64-NEXT:    addi a1, a0, 256
 ; RV64-NEXT:    addi a0, a0, 128
 ; RV64-NEXT:    vle64.v v24, (a0)
 ; RV64-NEXT:    vle64.v v0, (a1)
 ; RV64-NEXT:    vminu.vv v16, v24, v16
 ; RV64-NEXT:    vminu.vv v8, v8, v0
 ; RV64-NEXT:    vminu.vv v8, v8, v16
-; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.i v16, -1
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vredminu.vs v8, v8, v16
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <64 x i64>, <64 x i64>* %x
   %red = call i64 @llvm.vector.reduce.umin.v64i64(<64 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, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetivli zero, 2, e8, mf8, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetivli zero, 4, e8, mf4, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetivli zero, 8, e8, mf2, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetivli zero, 16, e8, m1, ta, mu
 ; 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 i8 @llvm.vector.reduce.umax.v32i8(<32 x i8>)
 
 define i8 @vreduce_umax_v32i8(<32 x i8>* %x) {
 ; CHECK-LABEL: vreduce_umax_v32i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m2, ta, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, 0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m2, ta, mu
 ; CHECK-NEXT:    vredmaxu.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <32 x i8>, <32 x i8>* %x
   %red = call i8 @llvm.vector.reduce.umax.v32i8(<32 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.umax.v64i8(<64 x i8>)
 
 define i8 @vreduce_umax_v64i8(<64 x i8>* %x) {
 ; CHECK-LABEL: vreduce_umax_v64i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 64
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m4, ta, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v12, 0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m4, ta, mu
 ; CHECK-NEXT:    vredmaxu.vs v8, v8, v12
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <64 x i8>, <64 x i8>* %x
   %red = call i8 @llvm.vector.reduce.umax.v64i8(<64 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.umax.v128i8(<128 x i8>)
 
 define i8 @vreduce_umax_v128i8(<128 x i8>* %x) {
 ; CHECK-LABEL: vreduce_umax_v128i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 128
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m8, ta, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v16, 0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m8, ta, mu
 ; CHECK-NEXT:    vredmaxu.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <128 x i8>, <128 x i8>* %x
   %red = call i8 @llvm.vector.reduce.umax.v128i8(<128 x i8> %v)
   ret i8 %red
 }
 
 declare i8 @llvm.vector.reduce.umax.v256i8(<256 x i8>)
 
 define i8 @vreduce_umax_v256i8(<256 x i8>* %x) {
 ; CHECK-LABEL: vreduce_umax_v256i8:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 128
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m8, ta, mu
 ; CHECK-NEXT:    vle8.v v8, (a0)
 ; CHECK-NEXT:    addi a0, a0, 128
 ; CHECK-NEXT:    vle8.v v16, (a0)
 ; CHECK-NEXT:    vmaxu.vv v8, v8, v16
-; CHECK-NEXT:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v16, 0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, m8, ta, mu
 ; CHECK-NEXT:    vredmaxu.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <256 x i8>, <256 x i8>* %x
   %red = call i8 @llvm.vector.reduce.umax.v256i8(<256 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, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetivli zero, 8, e16, m1, ta, mu
 ; 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 i16 @llvm.vector.reduce.umax.v16i16(<16 x i16>)
 
 define i16 @vreduce_umax_v16i16(<16 x i16>* %x) {
 ; CHECK-LABEL: vreduce_umax_v16i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 16, e16, m2, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, 0
 ; CHECK-NEXT:    vsetivli zero, 16, e16, m2, ta, mu
 ; CHECK-NEXT:    vredmaxu.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <16 x i16>, <16 x i16>* %x
   %red = call i16 @llvm.vector.reduce.umax.v16i16(<16 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.umax.v32i16(<32 x i16>)
 
 define i16 @vreduce_umax_v32i16(<32 x i16>* %x) {
 ; CHECK-LABEL: vreduce_umax_v32i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m4, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v12, 0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m4, ta, mu
 ; CHECK-NEXT:    vredmaxu.vs v8, v8, v12
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <32 x i16>, <32 x i16>* %x
   %red = call i16 @llvm.vector.reduce.umax.v32i16(<32 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.umax.v64i16(<64 x i16>)
 
 define i16 @vreduce_umax_v64i16(<64 x i16>* %x) {
 ; CHECK-LABEL: vreduce_umax_v64i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 64
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v16, 0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; CHECK-NEXT:    vredmaxu.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <64 x i16>, <64 x i16>* %x
   %red = call i16 @llvm.vector.reduce.umax.v64i16(<64 x i16> %v)
   ret i16 %red
 }
 
 declare i16 @llvm.vector.reduce.umax.v128i16(<128 x i16>)
 
 define i16 @vreduce_umax_v128i16(<128 x i16>* %x) {
 ; CHECK-LABEL: vreduce_umax_v128i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 64
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; CHECK-NEXT:    vle16.v v8, (a0)
 ; CHECK-NEXT:    addi a0, a0, 128
 ; CHECK-NEXT:    vle16.v v16, (a0)
 ; CHECK-NEXT:    vmaxu.vv v8, v8, v16
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v16, 0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m8, ta, mu
 ; CHECK-NEXT:    vredmaxu.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <128 x i16>, <128 x i16>* %x
   %red = call i16 @llvm.vector.reduce.umax.v128i16(<128 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, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
 ; 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, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
 ; 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 i32 @llvm.vector.reduce.umax.v8i32(<8 x i32>)
 
 define i32 @vreduce_umax_v8i32(<8 x i32>* %x) {
 ; CHECK-LABEL: vreduce_umax_v8i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 8, e32, m2, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, 0
 ; CHECK-NEXT:    vsetivli zero, 8, e32, m2, ta, mu
 ; CHECK-NEXT:    vredmaxu.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <8 x i32>, <8 x i32>* %x
   %red = call i32 @llvm.vector.reduce.umax.v8i32(<8 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.umax.v16i32(<16 x i32>)
 
 define i32 @vreduce_umax_v16i32(<16 x i32>* %x) {
 ; CHECK-LABEL: vreduce_umax_v16i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 16, e32, m4, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v12, 0
 ; CHECK-NEXT:    vsetivli zero, 16, e32, m4, ta, mu
 ; CHECK-NEXT:    vredmaxu.vs v8, v8, v12
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <16 x i32>, <16 x i32>* %x
   %red = call i32 @llvm.vector.reduce.umax.v16i32(<16 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.umax.v32i32(<32 x i32>)
 
 define i32 @vreduce_umax_v32i32(<32 x i32>* %x) {
 ; CHECK-LABEL: vreduce_umax_v32i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v16, 0
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
 ; CHECK-NEXT:    vredmaxu.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <32 x i32>, <32 x i32>* %x
   %red = call i32 @llvm.vector.reduce.umax.v32i32(<32 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.umax.v64i32(<64 x i32>)
 
 define i32 @vreduce_umax_v64i32(<64 x i32>* %x) {
 ; CHECK-LABEL: vreduce_umax_v64i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
 ; CHECK-NEXT:    vle32.v v8, (a0)
 ; CHECK-NEXT:    addi a0, a0, 128
 ; CHECK-NEXT:    vle32.v v16, (a0)
 ; CHECK-NEXT:    vmaxu.vv v8, v8, v16
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v16, 0
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m8, ta, mu
 ; CHECK-NEXT:    vredmaxu.vs v8, v8, v16
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %v = load <64 x i32>, <64 x i32>* %x
   %red = call i32 @llvm.vector.reduce.umax.v64i32(<64 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, mu
 ; RV32-NEXT:    vle64.v v8, (a0)
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vmv.v.i v9, 0
 ; RV32-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
 ; 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, mu
 ; 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, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
-; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.i v9, 0
 ; RV64-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
 ; 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
 }
 
 declare i64 @llvm.vector.reduce.umax.v4i64(<4 x i64>)
 
 define i64 @vreduce_umax_v4i64(<4 x i64>* %x) {
 ; RV32-LABEL: vreduce_umax_v4i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; RV32-NEXT:    vle64.v v8, (a0)
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vmv.v.i v10, 0
 ; RV32-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; RV32-NEXT:    vredmaxu.vs v8, v8, v10
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_umax_v4i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
-; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.i v10, 0
 ; RV64-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
 ; RV64-NEXT:    vredmaxu.vs v8, v8, v10
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <4 x i64>, <4 x i64>* %x
   %red = call i64 @llvm.vector.reduce.umax.v4i64(<4 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.umax.v8i64(<8 x i64>)
 
 define i64 @vreduce_umax_v8i64(<8 x i64>* %x) {
 ; RV32-LABEL: vreduce_umax_v8i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
 ; RV32-NEXT:    vle64.v v8, (a0)
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vmv.v.i v12, 0
 ; RV32-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
 ; RV32-NEXT:    vredmaxu.vs v8, v8, v12
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_umax_v8i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
-; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.i v12, 0
 ; RV64-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
 ; RV64-NEXT:    vredmaxu.vs v8, v8, v12
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <8 x i64>, <8 x i64>* %x
   %red = call i64 @llvm.vector.reduce.umax.v8i64(<8 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.umax.v16i64(<16 x i64>)
 
 define i64 @vreduce_umax_v16i64(<16 x i64>* %x) {
 ; RV32-LABEL: vreduce_umax_v16i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV32-NEXT:    vle64.v v8, (a0)
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vmv.v.i v16, 0
 ; RV32-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV32-NEXT:    vredmaxu.vs v8, v8, v16
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_umax_v16i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
-; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.i v16, 0
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vredmaxu.vs v8, v8, v16
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <16 x i64>, <16 x i64>* %x
   %red = call i64 @llvm.vector.reduce.umax.v16i64(<16 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.umax.v32i64(<32 x i64>)
 
 define i64 @vreduce_umax_v32i64(<32 x i64>* %x) {
 ; RV32-LABEL: vreduce_umax_v32i64:
 ; RV32:       # %bb.0:
 ; RV32-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV32-NEXT:    vle64.v v8, (a0)
 ; RV32-NEXT:    addi a0, a0, 128
 ; RV32-NEXT:    vle64.v v16, (a0)
 ; RV32-NEXT:    vmaxu.vv v8, v8, v16
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vmv.v.i v16, 0
 ; RV32-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV32-NEXT:    vredmaxu.vs v8, v8, v16
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_umax_v32i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
 ; RV64-NEXT:    addi a0, a0, 128
 ; RV64-NEXT:    vle64.v v16, (a0)
 ; RV64-NEXT:    vmaxu.vv v8, v8, v16
-; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.i v16, 0
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vredmaxu.vs v8, v8, v16
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <32 x i64>, <32 x i64>* %x
   %red = call i64 @llvm.vector.reduce.umax.v32i64(<32 x i64> %v)
   ret i64 %red

 ; RV32-NEXT:    vle64.v v16, (a1)
 ; RV32-NEXT:    addi a1, a0, 256
 ; RV32-NEXT:    addi a0, a0, 128
 ; RV32-NEXT:    vle64.v v24, (a0)
 ; RV32-NEXT:    vle64.v v0, (a1)
 ; RV32-NEXT:    vmaxu.vv v16, v24, v16
 ; RV32-NEXT:    vmaxu.vv v8, v8, v0
 ; RV32-NEXT:    vmaxu.vv v8, v8, v16
-; RV32-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vmv.v.i v16, 0
 ; RV32-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV32-NEXT:    vredmaxu.vs v8, v8, v16
 ; RV32-NEXT:    vmv.x.s a0, v8
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v8, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vreduce_umax_v64i64:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vle64.v v8, (a0)
 ; RV64-NEXT:    addi a1, a0, 384
 ; RV64-NEXT:    vle64.v v16, (a1)
 ; RV64-NEXT:    addi a1, a0, 256
 ; RV64-NEXT:    addi a0, a0, 128
 ; RV64-NEXT:    vle64.v v24, (a0)
 ; RV64-NEXT:    vle64.v v0, (a1)
 ; RV64-NEXT:    vmaxu.vv v16, v24, v16
 ; RV64-NEXT:    vmaxu.vv v8, v8, v0
 ; RV64-NEXT:    vmaxu.vv v8, v8, v16
-; RV64-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.i v16, 0
 ; RV64-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
 ; RV64-NEXT:    vredmaxu.vs v8, v8, v16
 ; RV64-NEXT:    vmv.x.s a0, v8
 ; RV64-NEXT:    ret
   %v = load <64 x i64>, <64 x i64>* %x
   %red = call i64 @llvm.vector.reduce.umax.v64i64(<64 x i64> %v)
   ret i64 %red
 }

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

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc -mtriple=riscv32 -mattr=+d,+experimental-zfh,+experimental-v -target-abi=ilp32d \
 ; RUN:     -verify-machineinstrs < %s | FileCheck %s
 ; RUN: llc -mtriple=riscv64 -mattr=+d,+experimental-zfh,+experimental-v -target-abi=lp64d \
 ; RUN:     -verify-machineinstrs < %s | FileCheck %s
 
 declare half @llvm.vector.reduce.fadd.nxv1f16(half, <vscale x 1 x half>)
 
 define half @vreduce_fadd_nxv1f16(<vscale x 1 x half> %v, half %s) {
 ; CHECK-LABEL: vreduce_fadd_nxv1f16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, %hi(.LCPI0_0)
-; CHECK-NEXT:    flh ft0, %lo(.LCPI0_0)(a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI0_0)
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
+; CHECK-NEXT:    vlse16.v v9, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, mu
 ; CHECK-NEXT:    vfredusum.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s ft0, v8
 ; CHECK-NEXT:    fadd.h fa0, fa0, ft0
 ; CHECK-NEXT:    ret
   %red = call reassoc half @llvm.vector.reduce.fadd.nxv1f16(half %s, <vscale x 1 x half> %v)
   ret half %red
 }
 
 define half @vreduce_ord_fadd_nxv1f16(<vscale x 1 x half> %v, half %s) {
 ; CHECK-LABEL: vreduce_ord_fadd_nxv1f16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v9, fa0
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, mu
 ; CHECK-NEXT:    vfredosum.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call half @llvm.vector.reduce.fadd.nxv1f16(half %s, <vscale x 1 x half> %v)
   ret half %red
 }
 
 declare half @llvm.vector.reduce.fadd.nxv2f16(half, <vscale x 2 x half>)
 
 define half @vreduce_fadd_nxv2f16(<vscale x 2 x half> %v, half %s) {
 ; CHECK-LABEL: vreduce_fadd_nxv2f16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, %hi(.LCPI2_0)
-; CHECK-NEXT:    flh ft0, %lo(.LCPI2_0)(a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI2_0)
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
+; CHECK-NEXT:    vlse16.v v9, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, mu
 ; CHECK-NEXT:    vfredusum.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s ft0, v8
 ; CHECK-NEXT:    fadd.h fa0, fa0, ft0
 ; CHECK-NEXT:    ret
   %red = call reassoc half @llvm.vector.reduce.fadd.nxv2f16(half %s, <vscale x 2 x half> %v)
   ret half %red
 }
 
 define half @vreduce_ord_fadd_nxv2f16(<vscale x 2 x half> %v, half %s) {
 ; CHECK-LABEL: vreduce_ord_fadd_nxv2f16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v9, fa0
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, mu
 ; CHECK-NEXT:    vfredosum.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call half @llvm.vector.reduce.fadd.nxv2f16(half %s, <vscale x 2 x half> %v)
   ret half %red
 }
 
 declare half @llvm.vector.reduce.fadd.nxv4f16(half, <vscale x 4 x half>)
 
 define half @vreduce_fadd_nxv4f16(<vscale x 4 x half> %v, half %s) {
 ; CHECK-LABEL: vreduce_fadd_nxv4f16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, %hi(.LCPI4_0)
-; CHECK-NEXT:    flh ft0, %lo(.LCPI4_0)(a0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI4_0)
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
+; CHECK-NEXT:    vlse16.v v9, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
 ; CHECK-NEXT:    vfredusum.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s ft0, v8
 ; CHECK-NEXT:    fadd.h fa0, fa0, ft0
 ; CHECK-NEXT:    ret
   %red = call reassoc half @llvm.vector.reduce.fadd.nxv4f16(half %s, <vscale x 4 x half> %v)
   ret half %red
 }
 
 define half @vreduce_ord_fadd_nxv4f16(<vscale x 4 x half> %v, half %s) {
 ; CHECK-LABEL: vreduce_ord_fadd_nxv4f16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v9, fa0
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
 ; CHECK-NEXT:    vfredosum.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call half @llvm.vector.reduce.fadd.nxv4f16(half %s, <vscale x 4 x half> %v)
   ret half %red
 }
 
 declare float @llvm.vector.reduce.fadd.nxv1f32(float, <vscale x 1 x float>)
 
 define float @vreduce_fadd_nxv1f32(<vscale x 1 x float> %v, float %s) {
 ; CHECK-LABEL: vreduce_fadd_nxv1f32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, %hi(.LCPI6_0)
-; CHECK-NEXT:    flw ft0, %lo(.LCPI6_0)(a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI6_0)
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
+; CHECK-NEXT:    vlse32.v v9, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, mu
 ; CHECK-NEXT:    vfredusum.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s ft0, v8
 ; CHECK-NEXT:    fadd.s fa0, fa0, ft0
 ; CHECK-NEXT:    ret
   %red = call reassoc float @llvm.vector.reduce.fadd.nxv1f32(float %s, <vscale x 1 x float> %v)
   ret float %red
 }
 
 define float @vreduce_ord_fadd_nxv1f32(<vscale x 1 x float> %v, float %s) {
 ; CHECK-LABEL: vreduce_ord_fadd_nxv1f32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v9, fa0
 ; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, mu
 ; CHECK-NEXT:    vfredosum.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call float @llvm.vector.reduce.fadd.nxv1f32(float %s, <vscale x 1 x float> %v)
   ret float %red
 }
 
 declare float @llvm.vector.reduce.fadd.nxv2f32(float, <vscale x 2 x float>)
 
 define float @vreduce_fadd_nxv2f32(<vscale x 2 x float> %v, float %s) {
 ; CHECK-LABEL: vreduce_fadd_nxv2f32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, %hi(.LCPI8_0)
-; CHECK-NEXT:    flw ft0, %lo(.LCPI8_0)(a0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI8_0)
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
+; CHECK-NEXT:    vlse32.v v9, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
 ; CHECK-NEXT:    vfredusum.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s ft0, v8
 ; CHECK-NEXT:    fadd.s fa0, fa0, ft0
 ; CHECK-NEXT:    ret
   %red = call reassoc float @llvm.vector.reduce.fadd.nxv2f32(float %s, <vscale x 2 x float> %v)
   ret float %red
 }
 
 define float @vreduce_ord_fadd_nxv2f32(<vscale x 2 x float> %v, float %s) {
 ; CHECK-LABEL: vreduce_ord_fadd_nxv2f32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v9, fa0
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
 ; CHECK-NEXT:    vfredosum.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call float @llvm.vector.reduce.fadd.nxv2f32(float %s, <vscale x 2 x float> %v)
   ret float %red
 }
 
 declare float @llvm.vector.reduce.fadd.nxv4f32(float, <vscale x 4 x float>)
 
 define float @vreduce_fadd_nxv4f32(<vscale x 4 x float> %v, float %s) {
 ; CHECK-LABEL: vreduce_fadd_nxv4f32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, %hi(.LCPI10_0)
-; CHECK-NEXT:    flw ft0, %lo(.LCPI10_0)(a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v10, ft0
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI10_0)
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
+; CHECK-NEXT:    vlse32.v v10, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, mu
 ; CHECK-NEXT:    vfredusum.vs v8, v8, v10
 ; CHECK-NEXT:    vfmv.f.s ft0, v8
 ; CHECK-NEXT:    fadd.s fa0, fa0, ft0
 ; CHECK-NEXT:    ret
   %red = call reassoc float @llvm.vector.reduce.fadd.nxv4f32(float %s, <vscale x 4 x float> %v)
   ret float %red
 }
 
 define float @vreduce_ord_fadd_nxv4f32(<vscale x 4 x float> %v, float %s) {
 ; CHECK-LABEL: vreduce_ord_fadd_nxv4f32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v10, fa0
 ; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, mu
 ; CHECK-NEXT:    vfredosum.vs v8, v8, v10
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call float @llvm.vector.reduce.fadd.nxv4f32(float %s, <vscale x 4 x float> %v)
   ret float %red
 }
 
 declare double @llvm.vector.reduce.fadd.nxv1f64(double, <vscale x 1 x double>)
 
 define double @vreduce_fadd_nxv1f64(<vscale x 1 x double> %v, double %s) {
 ; CHECK-LABEL: vreduce_fadd_nxv1f64:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, %hi(.LCPI12_0)
-; CHECK-NEXT:    fld ft0, %lo(.LCPI12_0)(a0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI12_0)
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
+; CHECK-NEXT:    vlse64.v v9, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
 ; CHECK-NEXT:    vfredusum.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s ft0, v8
 ; CHECK-NEXT:    fadd.d fa0, fa0, ft0
 ; CHECK-NEXT:    ret
   %red = call reassoc double @llvm.vector.reduce.fadd.nxv1f64(double %s, <vscale x 1 x double> %v)
   ret double %red
 }
 
 define double @vreduce_ord_fadd_nxv1f64(<vscale x 1 x double> %v, double %s) {
 ; CHECK-LABEL: vreduce_ord_fadd_nxv1f64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v9, fa0
+; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
 ; CHECK-NEXT:    vfredosum.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call double @llvm.vector.reduce.fadd.nxv1f64(double %s, <vscale x 1 x double> %v)
   ret double %red
 }
 
 declare double @llvm.vector.reduce.fadd.nxv2f64(double, <vscale x 2 x double>)
 
 define double @vreduce_fadd_nxv2f64(<vscale x 2 x double> %v, double %s) {
 ; CHECK-LABEL: vreduce_fadd_nxv2f64:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, %hi(.LCPI14_0)
-; CHECK-NEXT:    fld ft0, %lo(.LCPI14_0)(a0)
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v10, ft0
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI14_0)
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
+; CHECK-NEXT:    vlse64.v v10, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m2, ta, mu
 ; CHECK-NEXT:    vfredusum.vs v8, v8, v10
 ; CHECK-NEXT:    vfmv.f.s ft0, v8
 ; CHECK-NEXT:    fadd.d fa0, fa0, ft0
 ; CHECK-NEXT:    ret
   %red = call reassoc double @llvm.vector.reduce.fadd.nxv2f64(double %s, <vscale x 2 x double> %v)
   ret double %red
 }
 
 define double @vreduce_ord_fadd_nxv2f64(<vscale x 2 x double> %v, double %s) {
 ; CHECK-LABEL: vreduce_ord_fadd_nxv2f64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v10, fa0
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m2, ta, mu
 ; CHECK-NEXT:    vfredosum.vs v8, v8, v10
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call double @llvm.vector.reduce.fadd.nxv2f64(double %s, <vscale x 2 x double> %v)
   ret double %red
 }
 
 declare double @llvm.vector.reduce.fadd.nxv4f64(double, <vscale x 4 x double>)
 
 define double @vreduce_fadd_nxv4f64(<vscale x 4 x double> %v, double %s) {
 ; CHECK-LABEL: vreduce_fadd_nxv4f64:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, %hi(.LCPI16_0)
-; CHECK-NEXT:    fld ft0, %lo(.LCPI16_0)(a0)
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v12, ft0
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI16_0)
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
+; CHECK-NEXT:    vlse64.v v12, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m4, ta, mu
 ; CHECK-NEXT:    vfredusum.vs v8, v8, v12
 ; CHECK-NEXT:    vfmv.f.s ft0, v8
 ; CHECK-NEXT:    fadd.d fa0, fa0, ft0
 ; CHECK-NEXT:    ret
   %red = call reassoc double @llvm.vector.reduce.fadd.nxv4f64(double %s, <vscale x 4 x double> %v)
   ret double %red
 }
 
 define double @vreduce_ord_fadd_nxv4f64(<vscale x 4 x double> %v, double %s) {
 ; CHECK-LABEL: vreduce_ord_fadd_nxv4f64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v12, fa0
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m4, ta, mu
 ; CHECK-NEXT:    vfredosum.vs v8, v8, v12
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call double @llvm.vector.reduce.fadd.nxv4f64(double %s, <vscale x 4 x double> %v)
   ret double %red
 }
 
 declare half @llvm.vector.reduce.fmin.nxv1f16(<vscale x 1 x half>)
 
 define half @vreduce_fmin_nxv1f16(<vscale x 1 x half> %v) {
 ; CHECK-LABEL: vreduce_fmin_nxv1f16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, %hi(.LCPI18_0)
-; CHECK-NEXT:    flh ft0, %lo(.LCPI18_0)(a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI18_0)
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
+; CHECK-NEXT:    vlse16.v v9, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, mu
 ; CHECK-NEXT:    vfredmin.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call half @llvm.vector.reduce.fmin.nxv1f16(<vscale x 1 x half> %v)
   ret half %red
 }
 
 define half @vreduce_fmin_nxv1f16_nonans(<vscale x 1 x half> %v) #0 {
 ; CHECK-LABEL: vreduce_fmin_nxv1f16_nonans:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, %hi(.LCPI19_0)
-; CHECK-NEXT:    flh ft0, %lo(.LCPI19_0)(a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI19_0)
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
+; CHECK-NEXT:    vlse16.v v9, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, mu
 ; CHECK-NEXT:    vfredmin.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call nnan half @llvm.vector.reduce.fmin.nxv1f16(<vscale x 1 x half> %v)
   ret half %red
 }
 
 define half @vreduce_fmin_nxv1f16_nonans_noinfs(<vscale x 1 x half> %v) #1 {
 ; CHECK-LABEL: vreduce_fmin_nxv1f16_nonans_noinfs:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, %hi(.LCPI20_0)
-; CHECK-NEXT:    flh ft0, %lo(.LCPI20_0)(a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI20_0)
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
+; CHECK-NEXT:    vlse16.v v9, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, mu
 ; CHECK-NEXT:    vfredmin.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call nnan ninf half @llvm.vector.reduce.fmin.nxv1f16(<vscale x 1 x half> %v)
   ret half %red
 }
 
 declare half @llvm.vector.reduce.fmin.nxv2f16(<vscale x 2 x half>)
 
 define half @vreduce_fmin_nxv2f16(<vscale x 2 x half> %v) {
 ; CHECK-LABEL: vreduce_fmin_nxv2f16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, %hi(.LCPI21_0)
-; CHECK-NEXT:    flh ft0, %lo(.LCPI21_0)(a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI21_0)
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
+; CHECK-NEXT:    vlse16.v v9, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, mu
 ; CHECK-NEXT:    vfredmin.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call half @llvm.vector.reduce.fmin.nxv2f16(<vscale x 2 x half> %v)
   ret half %red
 }
 
 declare half @llvm.vector.reduce.fmin.nxv4f16(<vscale x 4 x half>)
 
 define half @vreduce_fmin_nxv4f16(<vscale x 4 x half> %v) {
 ; CHECK-LABEL: vreduce_fmin_nxv4f16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, %hi(.LCPI22_0)
-; CHECK-NEXT:    flh ft0, %lo(.LCPI22_0)(a0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI22_0)
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
+; CHECK-NEXT:    vlse16.v v9, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
 ; CHECK-NEXT:    vfredmin.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call half @llvm.vector.reduce.fmin.nxv4f16(<vscale x 4 x half> %v)
   ret half %red
 }
 
 declare half @llvm.vector.reduce.fmin.nxv64f16(<vscale x 64 x half>)
 
 define half @vreduce_fmin_nxv64f16(<vscale x 64 x half> %v) {
 ; CHECK-LABEL: vreduce_fmin_nxv64f16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a0, %hi(.LCPI23_0)
-; CHECK-NEXT:    flh ft0, %lo(.LCPI23_0)(a0)
 ; CHECK-NEXT:    vsetvli a0, zero, e16, m8, ta, mu
 ; CHECK-NEXT:    vfmin.vv v8, v8, v16
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v16, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI23_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI23_0)
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
+; CHECK-NEXT:    vlse16.v v16, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e16, m8, ta, mu
 ; CHECK-NEXT:    vfredmin.vs v8, v8, v16
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call half @llvm.vector.reduce.fmin.nxv64f16(<vscale x 64 x half> %v)
   ret half %red
 }
 
 declare float @llvm.vector.reduce.fmin.nxv1f32(<vscale x 1 x float>)
 
 define float @vreduce_fmin_nxv1f32(<vscale x 1 x float> %v) {
 ; CHECK-LABEL: vreduce_fmin_nxv1f32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, %hi(.LCPI24_0)
-; CHECK-NEXT:    flw ft0, %lo(.LCPI24_0)(a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI24_0)
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
+; CHECK-NEXT:    vlse32.v v9, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, mu
 ; CHECK-NEXT:    vfredmin.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call float @llvm.vector.reduce.fmin.nxv1f32(<vscale x 1 x float> %v)
   ret float %red
 }
 
 define float @vreduce_fmin_nxv1f32_nonans(<vscale x 1 x float> %v) {
 ; CHECK-LABEL: vreduce_fmin_nxv1f32_nonans:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, %hi(.LCPI25_0)
-; CHECK-NEXT:    flw ft0, %lo(.LCPI25_0)(a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI25_0)
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
+; CHECK-NEXT:    vlse32.v v9, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, mu
 ; CHECK-NEXT:    vfredmin.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call nnan float @llvm.vector.reduce.fmin.nxv1f32(<vscale x 1 x float> %v)
   ret float %red
 }
 
 define float @vreduce_fmin_nxv1f32_nonans_noinfs(<vscale x 1 x float> %v) {
 ; CHECK-LABEL: vreduce_fmin_nxv1f32_nonans_noinfs:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, %hi(.LCPI26_0)
-; CHECK-NEXT:    flw ft0, %lo(.LCPI26_0)(a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI26_0)
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
+; CHECK-NEXT:    vlse32.v v9, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, mu
 ; CHECK-NEXT:    vfredmin.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call nnan ninf float @llvm.vector.reduce.fmin.nxv1f32(<vscale x 1 x float> %v)
   ret float %red
 }
 
 declare float @llvm.vector.reduce.fmin.nxv2f32(<vscale x 2 x float>)
 
 define float @vreduce_fmin_nxv2f32(<vscale x 2 x float> %v) {
 ; CHECK-LABEL: vreduce_fmin_nxv2f32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, %hi(.LCPI27_0)
-; CHECK-NEXT:    flw ft0, %lo(.LCPI27_0)(a0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI27_0)
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
+; CHECK-NEXT:    vlse32.v v9, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
 ; CHECK-NEXT:    vfredmin.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call float @llvm.vector.reduce.fmin.nxv2f32(<vscale x 2 x float> %v)
   ret float %red
 }
 
 declare float @llvm.vector.reduce.fmin.nxv4f32(<vscale x 4 x float>)
 
 define float @vreduce_fmin_nxv4f32(<vscale x 4 x float> %v) {
 ; CHECK-LABEL: vreduce_fmin_nxv4f32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, %hi(.LCPI28_0)
-; CHECK-NEXT:    flw ft0, %lo(.LCPI28_0)(a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v10, ft0
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI28_0)
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
+; CHECK-NEXT:    vlse32.v v10, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, mu
 ; CHECK-NEXT:    vfredmin.vs v8, v8, v10
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call float @llvm.vector.reduce.fmin.nxv4f32(<vscale x 4 x float> %v)
   ret float %red
 }
 
 declare float @llvm.vector.reduce.fmin.nxv32f32(<vscale x 32 x float>)
 
 define float @vreduce_fmin_nxv32f32(<vscale x 32 x float> %v) {
 ; CHECK-LABEL: vreduce_fmin_nxv32f32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a0, %hi(.LCPI29_0)
-; CHECK-NEXT:    flw ft0, %lo(.LCPI29_0)(a0)
 ; CHECK-NEXT:    vsetvli a0, zero, e32, m8, ta, mu
 ; CHECK-NEXT:    vfmin.vv v8, v8, v16
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v16, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI29_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI29_0)
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
+; CHECK-NEXT:    vlse32.v v16, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e32, m8, ta, mu
 ; CHECK-NEXT:    vfredmin.vs v8, v8, v16
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call float @llvm.vector.reduce.fmin.nxv32f32(<vscale x 32 x float> %v)
   ret float %red
 }
 
 declare double @llvm.vector.reduce.fmin.nxv1f64(<vscale x 1 x double>)
 
 define double @vreduce_fmin_nxv1f64(<vscale x 1 x double> %v) {
 ; CHECK-LABEL: vreduce_fmin_nxv1f64:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, %hi(.LCPI30_0)
-; CHECK-NEXT:    fld ft0, %lo(.LCPI30_0)(a0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI30_0)
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
+; CHECK-NEXT:    vlse64.v v9, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
 ; CHECK-NEXT:    vfredmin.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call double @llvm.vector.reduce.fmin.nxv1f64(<vscale x 1 x double> %v)
   ret double %red
 }
 
 define double @vreduce_fmin_nxv1f64_nonans(<vscale x 1 x double> %v) {
 ; CHECK-LABEL: vreduce_fmin_nxv1f64_nonans:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, %hi(.LCPI31_0)
-; CHECK-NEXT:    fld ft0, %lo(.LCPI31_0)(a0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI31_0)
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
+; CHECK-NEXT:    vlse64.v v9, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
 ; CHECK-NEXT:    vfredmin.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call nnan double @llvm.vector.reduce.fmin.nxv1f64(<vscale x 1 x double> %v)
   ret double %red
 }
 
 define double @vreduce_fmin_nxv1f64_nonans_noinfs(<vscale x 1 x double> %v) {
 ; CHECK-LABEL: vreduce_fmin_nxv1f64_nonans_noinfs:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, %hi(.LCPI32_0)
-; CHECK-NEXT:    fld ft0, %lo(.LCPI32_0)(a0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI32_0)
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
+; CHECK-NEXT:    vlse64.v v9, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
 ; CHECK-NEXT:    vfredmin.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call nnan ninf double @llvm.vector.reduce.fmin.nxv1f64(<vscale x 1 x double> %v)
   ret double %red
 }
 
 declare double @llvm.vector.reduce.fmin.nxv2f64(<vscale x 2 x double>)
 
 define double @vreduce_fmin_nxv2f64(<vscale x 2 x double> %v) {
 ; CHECK-LABEL: vreduce_fmin_nxv2f64:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, %hi(.LCPI33_0)
-; CHECK-NEXT:    fld ft0, %lo(.LCPI33_0)(a0)
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v10, ft0
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI33_0)
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
+; CHECK-NEXT:    vlse64.v v10, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m2, ta, mu
 ; CHECK-NEXT:    vfredmin.vs v8, v8, v10
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call double @llvm.vector.reduce.fmin.nxv2f64(<vscale x 2 x double> %v)
   ret double %red
 }
 
 declare double @llvm.vector.reduce.fmin.nxv4f64(<vscale x 4 x double>)
 
 define double @vreduce_fmin_nxv4f64(<vscale x 4 x double> %v) {
 ; CHECK-LABEL: vreduce_fmin_nxv4f64:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, %hi(.LCPI34_0)
-; CHECK-NEXT:    fld ft0, %lo(.LCPI34_0)(a0)
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v12, ft0
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI34_0)
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
+; CHECK-NEXT:    vlse64.v v12, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m4, ta, mu
 ; CHECK-NEXT:    vfredmin.vs v8, v8, v12
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call double @llvm.vector.reduce.fmin.nxv4f64(<vscale x 4 x double> %v)
   ret double %red
 }
 
 declare double @llvm.vector.reduce.fmin.nxv16f64(<vscale x 16 x double>)
 
 define double @vreduce_fmin_nxv16f64(<vscale x 16 x double> %v) {
 ; CHECK-LABEL: vreduce_fmin_nxv16f64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a0, %hi(.LCPI35_0)
-; CHECK-NEXT:    fld ft0, %lo(.LCPI35_0)(a0)
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m8, ta, mu
 ; CHECK-NEXT:    vfmin.vv v8, v8, v16
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v16, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI35_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI35_0)
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
+; CHECK-NEXT:    vlse64.v v16, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m8, ta, mu
 ; CHECK-NEXT:    vfredmin.vs v8, v8, v16
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call double @llvm.vector.reduce.fmin.nxv16f64(<vscale x 16 x double> %v)
   ret double %red
 }
 
 declare half @llvm.vector.reduce.fmax.nxv1f16(<vscale x 1 x half>)
 
 define half @vreduce_fmax_nxv1f16(<vscale x 1 x half> %v) {
 ; CHECK-LABEL: vreduce_fmax_nxv1f16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, %hi(.LCPI36_0)
-; CHECK-NEXT:    flh ft0, %lo(.LCPI36_0)(a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI36_0)
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
+; CHECK-NEXT:    vlse16.v v9, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, mu
 ; CHECK-NEXT:    vfredmax.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call half @llvm.vector.reduce.fmax.nxv1f16(<vscale x 1 x half> %v)
   ret half %red
 }
 
 define half @vreduce_fmax_nxv1f16_nonans(<vscale x 1 x half> %v) #0 {
 ; CHECK-LABEL: vreduce_fmax_nxv1f16_nonans:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, %hi(.LCPI37_0)
-; CHECK-NEXT:    flh ft0, %lo(.LCPI37_0)(a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI37_0)
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
+; CHECK-NEXT:    vlse16.v v9, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, mu
 ; CHECK-NEXT:    vfredmax.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call nnan half @llvm.vector.reduce.fmax.nxv1f16(<vscale x 1 x half> %v)
   ret half %red
 }
 
 define half @vreduce_fmax_nxv1f16_nonans_noinfs(<vscale x 1 x half> %v) #1 {
 ; CHECK-LABEL: vreduce_fmax_nxv1f16_nonans_noinfs:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, %hi(.LCPI38_0)
-; CHECK-NEXT:    flh ft0, %lo(.LCPI38_0)(a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI38_0)
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
+; CHECK-NEXT:    vlse16.v v9, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, mu
 ; CHECK-NEXT:    vfredmax.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call nnan ninf half @llvm.vector.reduce.fmax.nxv1f16(<vscale x 1 x half> %v)
   ret half %red
 }
 
 declare half @llvm.vector.reduce.fmax.nxv2f16(<vscale x 2 x half>)
 
 define half @vreduce_fmax_nxv2f16(<vscale x 2 x half> %v) {
 ; CHECK-LABEL: vreduce_fmax_nxv2f16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, %hi(.LCPI39_0)
-; CHECK-NEXT:    flh ft0, %lo(.LCPI39_0)(a0)
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI39_0)
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
+; CHECK-NEXT:    vlse16.v v9, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, mu
 ; CHECK-NEXT:    vfredmax.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call half @llvm.vector.reduce.fmax.nxv2f16(<vscale x 2 x half> %v)
   ret half %red
 }
 
 declare half @llvm.vector.reduce.fmax.nxv4f16(<vscale x 4 x half>)
 
 define half @vreduce_fmax_nxv4f16(<vscale x 4 x half> %v) {
 ; CHECK-LABEL: vreduce_fmax_nxv4f16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, %hi(.LCPI40_0)
-; CHECK-NEXT:    flh ft0, %lo(.LCPI40_0)(a0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI40_0)
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
+; CHECK-NEXT:    vlse16.v v9, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
 ; CHECK-NEXT:    vfredmax.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call half @llvm.vector.reduce.fmax.nxv4f16(<vscale x 4 x half> %v)
   ret half %red
 }
 
 declare half @llvm.vector.reduce.fmax.nxv64f16(<vscale x 64 x half>)
 
 define half @vreduce_fmax_nxv64f16(<vscale x 64 x half> %v) {
 ; CHECK-LABEL: vreduce_fmax_nxv64f16:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a0, %hi(.LCPI41_0)
-; CHECK-NEXT:    flh ft0, %lo(.LCPI41_0)(a0)
 ; CHECK-NEXT:    vsetvli a0, zero, e16, m8, ta, mu
 ; CHECK-NEXT:    vfmax.vv v8, v8, v16
-; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v16, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI41_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI41_0)
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
+; CHECK-NEXT:    vlse16.v v16, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e16, m8, ta, mu
 ; CHECK-NEXT:    vfredmax.vs v8, v8, v16
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call half @llvm.vector.reduce.fmax.nxv64f16(<vscale x 64 x half> %v)
   ret half %red
 }
 
 declare float @llvm.vector.reduce.fmax.nxv1f32(<vscale x 1 x float>)
 
 define float @vreduce_fmax_nxv1f32(<vscale x 1 x float> %v) {
 ; CHECK-LABEL: vreduce_fmax_nxv1f32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, %hi(.LCPI42_0)
-; CHECK-NEXT:    flw ft0, %lo(.LCPI42_0)(a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI42_0)
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
+; CHECK-NEXT:    vlse32.v v9, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, mu
 ; CHECK-NEXT:    vfredmax.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call float @llvm.vector.reduce.fmax.nxv1f32(<vscale x 1 x float> %v)
   ret float %red
 }
 
 define float @vreduce_fmax_nxv1f32_nonans(<vscale x 1 x float> %v) {
 ; CHECK-LABEL: vreduce_fmax_nxv1f32_nonans:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, %hi(.LCPI43_0)
-; CHECK-NEXT:    flw ft0, %lo(.LCPI43_0)(a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI43_0)
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
+; CHECK-NEXT:    vlse32.v v9, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, mu
 ; CHECK-NEXT:    vfredmax.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call nnan float @llvm.vector.reduce.fmax.nxv1f32(<vscale x 1 x float> %v)
   ret float %red
 }
 
 define float @vreduce_fmax_nxv1f32_nonans_noinfs(<vscale x 1 x float> %v) {
 ; CHECK-LABEL: vreduce_fmax_nxv1f32_nonans_noinfs:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, %hi(.LCPI44_0)
-; CHECK-NEXT:    flw ft0, %lo(.LCPI44_0)(a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI44_0)
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
+; CHECK-NEXT:    vlse32.v v9, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, mu
 ; CHECK-NEXT:    vfredmax.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call nnan ninf float @llvm.vector.reduce.fmax.nxv1f32(<vscale x 1 x float> %v)
   ret float %red
 }
 
 declare float @llvm.vector.reduce.fmax.nxv2f32(<vscale x 2 x float>)
 
 define float @vreduce_fmax_nxv2f32(<vscale x 2 x float> %v) {
 ; CHECK-LABEL: vreduce_fmax_nxv2f32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, %hi(.LCPI45_0)
-; CHECK-NEXT:    flw ft0, %lo(.LCPI45_0)(a0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI45_0)
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
+; CHECK-NEXT:    vlse32.v v9, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
 ; CHECK-NEXT:    vfredmax.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call float @llvm.vector.reduce.fmax.nxv2f32(<vscale x 2 x float> %v)
   ret float %red
 }
 
 declare float @llvm.vector.reduce.fmax.nxv4f32(<vscale x 4 x float>)
 
 define float @vreduce_fmax_nxv4f32(<vscale x 4 x float> %v) {
 ; CHECK-LABEL: vreduce_fmax_nxv4f32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, %hi(.LCPI46_0)
-; CHECK-NEXT:    flw ft0, %lo(.LCPI46_0)(a0)
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v10, ft0
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI46_0)
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
+; CHECK-NEXT:    vlse32.v v10, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, mu
 ; CHECK-NEXT:    vfredmax.vs v8, v8, v10
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call float @llvm.vector.reduce.fmax.nxv4f32(<vscale x 4 x float> %v)
   ret float %red
 }
 
 declare float @llvm.vector.reduce.fmax.nxv32f32(<vscale x 32 x float>)
 
 define float @vreduce_fmax_nxv32f32(<vscale x 32 x float> %v) {
 ; CHECK-LABEL: vreduce_fmax_nxv32f32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a0, %hi(.LCPI47_0)
-; CHECK-NEXT:    flw ft0, %lo(.LCPI47_0)(a0)
 ; CHECK-NEXT:    vsetvli a0, zero, e32, m8, ta, mu
 ; CHECK-NEXT:    vfmax.vv v8, v8, v16
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v16, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI47_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI47_0)
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
+; CHECK-NEXT:    vlse32.v v16, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e32, m8, ta, mu
 ; CHECK-NEXT:    vfredmax.vs v8, v8, v16
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call float @llvm.vector.reduce.fmax.nxv32f32(<vscale x 32 x float> %v)
   ret float %red
 }
 
 declare double @llvm.vector.reduce.fmax.nxv1f64(<vscale x 1 x double>)
 
 define double @vreduce_fmax_nxv1f64(<vscale x 1 x double> %v) {
 ; CHECK-LABEL: vreduce_fmax_nxv1f64:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, %hi(.LCPI48_0)
-; CHECK-NEXT:    fld ft0, %lo(.LCPI48_0)(a0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI48_0)
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
+; CHECK-NEXT:    vlse64.v v9, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
 ; CHECK-NEXT:    vfredmax.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call double @llvm.vector.reduce.fmax.nxv1f64(<vscale x 1 x double> %v)
   ret double %red
 }
 
 define double @vreduce_fmax_nxv1f64_nonans(<vscale x 1 x double> %v) {
 ; CHECK-LABEL: vreduce_fmax_nxv1f64_nonans:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, %hi(.LCPI49_0)
-; CHECK-NEXT:    fld ft0, %lo(.LCPI49_0)(a0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI49_0)
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
+; CHECK-NEXT:    vlse64.v v9, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
 ; CHECK-NEXT:    vfredmax.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call nnan double @llvm.vector.reduce.fmax.nxv1f64(<vscale x 1 x double> %v)
   ret double %red
 }
 
 define double @vreduce_fmax_nxv1f64_nonans_noinfs(<vscale x 1 x double> %v) {
 ; CHECK-LABEL: vreduce_fmax_nxv1f64_nonans_noinfs:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, %hi(.LCPI50_0)
-; CHECK-NEXT:    fld ft0, %lo(.LCPI50_0)(a0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI50_0)
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
+; CHECK-NEXT:    vlse64.v v9, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v9, ft0
 ; CHECK-NEXT:    vfredmax.vs v8, v8, v9
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call nnan ninf double @llvm.vector.reduce.fmax.nxv1f64(<vscale x 1 x double> %v)
   ret double %red
 }
 
 declare double @llvm.vector.reduce.fmax.nxv2f64(<vscale x 2 x double>)
 
 define double @vreduce_fmax_nxv2f64(<vscale x 2 x double> %v) {
 ; CHECK-LABEL: vreduce_fmax_nxv2f64:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, %hi(.LCPI51_0)
-; CHECK-NEXT:    fld ft0, %lo(.LCPI51_0)(a0)
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v10, ft0
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI51_0)
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
+; CHECK-NEXT:    vlse64.v v10, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m2, ta, mu
 ; CHECK-NEXT:    vfredmax.vs v8, v8, v10
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call double @llvm.vector.reduce.fmax.nxv2f64(<vscale x 2 x double> %v)
   ret double %red
 }
 
 declare double @llvm.vector.reduce.fmax.nxv4f64(<vscale x 4 x double>)
 
 define double @vreduce_fmax_nxv4f64(<vscale x 4 x double> %v) {
 ; CHECK-LABEL: vreduce_fmax_nxv4f64:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, %hi(.LCPI52_0)
-; CHECK-NEXT:    fld ft0, %lo(.LCPI52_0)(a0)
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v12, ft0
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI52_0)
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
+; CHECK-NEXT:    vlse64.v v12, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m4, ta, mu
 ; CHECK-NEXT:    vfredmax.vs v8, v8, v12
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call double @llvm.vector.reduce.fmax.nxv4f64(<vscale x 4 x double> %v)
   ret double %red
 }
 
 declare double @llvm.vector.reduce.fmax.nxv16f64(<vscale x 16 x double>)
 
 define double @vreduce_fmax_nxv16f64(<vscale x 16 x double> %v) {
 ; CHECK-LABEL: vreduce_fmax_nxv16f64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    lui a0, %hi(.LCPI53_0)
-; CHECK-NEXT:    fld ft0, %lo(.LCPI53_0)(a0)
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m8, ta, mu
 ; CHECK-NEXT:    vfmax.vv v8, v8, v16
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
-; CHECK-NEXT:    vfmv.v.f v16, ft0
+; CHECK-NEXT:    lui a0, %hi(.LCPI53_0)
+; CHECK-NEXT:    addi a0, a0, %lo(.LCPI53_0)
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
+; CHECK-NEXT:    vlse64.v v16, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m8, ta, mu
 ; CHECK-NEXT:    vfredmax.vs v8, v8, v16
 ; CHECK-NEXT:    vfmv.f.s fa0, v8
 ; CHECK-NEXT:    ret
   %red = call double @llvm.vector.reduce.fmax.nxv16f64(<vscale x 16 x double> %v)
   ret double %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,+experimental-zfh,+experimental-v -target-abi=ilp32d \
 ; RUN:     -verify-machineinstrs < %s | FileCheck %s
 ; RUN: llc -mtriple=riscv64 -mattr=+d,+experimental-zfh,+experimental-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:    vsetvli a1, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v9, fa0
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, tu, mu
 ; 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:    vsetvli a1, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v9, fa0
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, tu, mu
 ; 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:    vsetvli a1, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v9, fa0
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, tu, mu
 ; 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:    vsetvli a1, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v9, fa0
 ; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, tu, mu
 ; 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:    vsetvli a1, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v9, fa0
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m1, tu, mu
 ; 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:    vsetvli a1, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v9, fa0
 ; CHECK-NEXT:    vsetvli zero, a0, e16, m1, tu, mu
 ; 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 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:    vsetvli a1, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v9, fa0
 ; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, tu, mu
 ; 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:    vsetvli a1, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v9, fa0
 ; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, tu, mu
 ; 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:    vsetvli a1, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v9, fa0
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m1, tu, mu
 ; 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:    vsetvli a1, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v9, fa0
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m1, tu, mu
 ; 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)
 
 define float @vpreduce_fadd_nxv4f32(float %s, <vscale x 4 x float> %v, <vscale x 4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_fadd_nxv4f32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a1, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v10, fa0
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m2, tu, mu
 ; CHECK-NEXT:    vfredusum.vs v10, v8, v10, v0.t
 ; CHECK-NEXT:    vfmv.f.s fa0, v10
 ; CHECK-NEXT:    ret
   %r = call reassoc float @llvm.vp.reduce.fadd.nxv4f32(float %s, <vscale x 4 x float> %v, <vscale x 4 x i1> %m, i32 %evl)
   ret float %r
 }
 
 define float @vpreduce_ord_fadd_nxv4f32(float %s, <vscale x 4 x float> %v, <vscale x 4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_ord_fadd_nxv4f32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a1, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v10, fa0
 ; CHECK-NEXT:    vsetvli zero, a0, e32, m2, tu, mu
 ; CHECK-NEXT:    vfredosum.vs v10, v8, v10, v0.t
 ; CHECK-NEXT:    vfmv.f.s fa0, v10
 ; CHECK-NEXT:    ret
   %r = call float @llvm.vp.reduce.fadd.nxv4f32(float %s, <vscale x 4 x float> %v, <vscale x 4 x i1> %m, i32 %evl)
   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:    vsetvli a1, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v9, fa0
 ; CHECK-NEXT:    vsetvli zero, a0, e64, m1, tu, mu
 ; 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:    vsetvli a1, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v9, fa0
 ; CHECK-NEXT:    vsetvli zero, a0, e64, m1, tu, mu
 ; 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)
 
 define double @vpreduce_fadd_nxv2f64(double %s, <vscale x 2 x double> %v, <vscale x 2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_fadd_nxv2f64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a1, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v10, fa0
 ; CHECK-NEXT:    vsetvli zero, a0, e64, m2, tu, mu
 ; CHECK-NEXT:    vfredusum.vs v10, v8, v10, v0.t
 ; CHECK-NEXT:    vfmv.f.s fa0, v10
 ; CHECK-NEXT:    ret
   %r = call reassoc double @llvm.vp.reduce.fadd.nxv2f64(double %s, <vscale x 2 x double> %v, <vscale x 2 x i1> %m, i32 %evl)
   ret double %r
 }
 
 define double @vpreduce_ord_fadd_nxv2f64(double %s, <vscale x 2 x double> %v, <vscale x 2 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_ord_fadd_nxv2f64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a1, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v10, fa0
 ; CHECK-NEXT:    vsetvli zero, a0, e64, m2, tu, mu
 ; CHECK-NEXT:    vfredosum.vs v10, v8, v10, v0.t
 ; CHECK-NEXT:    vfmv.f.s fa0, v10
 ; CHECK-NEXT:    ret
   %r = call double @llvm.vp.reduce.fadd.nxv2f64(double %s, <vscale x 2 x double> %v, <vscale x 2 x i1> %m, i32 %evl)
   ret double %r
 }
 
 declare double @llvm.vp.reduce.fadd.nxv4f64(double, <vscale x 4 x double>, <vscale x 4 x i1>, i32)
 
 define double @vpreduce_fadd_nxv4f64(double %s, <vscale x 4 x double> %v, <vscale x 4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_fadd_nxv4f64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a1, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v12, fa0
 ; CHECK-NEXT:    vsetvli zero, a0, e64, m4, tu, mu
 ; CHECK-NEXT:    vfredusum.vs v12, v8, v12, v0.t
 ; CHECK-NEXT:    vfmv.f.s fa0, v12
 ; CHECK-NEXT:    ret
   %r = call reassoc double @llvm.vp.reduce.fadd.nxv4f64(double %s, <vscale x 4 x double> %v, <vscale x 4 x i1> %m, i32 %evl)
   ret double %r
 }
 
 define double @vpreduce_ord_fadd_nxv4f64(double %s, <vscale x 4 x double> %v, <vscale x 4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_ord_fadd_nxv4f64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a1, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vfmv.v.f v12, fa0
 ; CHECK-NEXT:    vsetvli zero, a0, e64, m4, tu, mu
 ; CHECK-NEXT:    vfredosum.vs v12, v8, v12, v0.t
 ; CHECK-NEXT:    vfmv.f.s fa0, v12
 ; CHECK-NEXT:    ret
   %r = call double @llvm.vp.reduce.fadd.nxv4f64(double %s, <vscale x 4 x double> %v, <vscale x 4 x i1> %m, i32 %evl)
   ret double %r
 }

llvm/test/CodeGen/RISCV/rvv/vreductions-int-rv32.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc -mtriple=riscv32 -mattr=+experimental-v -verify-machineinstrs < %s | FileCheck %s
 
 declare i8 @llvm.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:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf8, ta, mu
 ; 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:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf8, ta, mu
 ; 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:    vsetvli a1, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf8, ta, mu
 ; 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:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf8, ta, mu
 ; 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:    vsetvli a1, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf8, ta, mu
 ; 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:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf8, ta, mu
 ; 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:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf8, ta, mu
 ; 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:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf8, ta, mu
 ; 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:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf4, ta, mu
 ; 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:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf4, ta, mu
 ; 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:    vsetvli a1, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf4, ta, mu
 ; 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:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf4, ta, mu
 ; 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:    vsetvli a1, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf4, ta, mu
 ; 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:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf4, ta, mu
 ; 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:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf4, ta, mu
 ; 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:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf4, ta, mu
 ; 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:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf2, ta, mu
 ; 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:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf2, ta, mu
 ; 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:    vsetvli a1, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf2, ta, mu
 ; 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:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf2, ta, mu
 ; 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:    vsetvli a1, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf2, ta, mu
 ; 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:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf2, ta, mu
 ; 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:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf2, ta, mu
 ; 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:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf2, ta, mu
 ; 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:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, mu
 ; 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
 }
 
 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:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, mu
 ; 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:    vsetvli a1, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, mu
 ; 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:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, mu
 ; 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) {
 ; CHECK-LABEL: vreduce_smin_nxv1i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, 8
 ; CHECK-NEXT:    addi a0, a0, -1
-; CHECK-NEXT:    vsetvli a1, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, mu
 ; CHECK-NEXT:    vredmin.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-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:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, mu
 ; 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:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, mu
 ; 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:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, mu
 ; 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:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, mu
 ; 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
 }
 
 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:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, mu
 ; 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:    vsetvli a1, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, mu
 ; 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:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, mu
 ; 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) {
 ; CHECK-LABEL: vreduce_smin_nxv2i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, 8
 ; CHECK-NEXT:    addi a0, a0, -1
-; CHECK-NEXT:    vsetvli a1, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, mu
 ; CHECK-NEXT:    vredmin.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-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:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, mu
 ; 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:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, mu
 ; 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:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, mu
 ; 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:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
 ; 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
 }
 
 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:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
 ; 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:    vsetvli a1, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
 ; 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:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
 ; 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) {
 ; CHECK-LABEL: vreduce_smin_nxv4i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, 8
 ; CHECK-NEXT:    addi a0, a0, -1
-; CHECK-NEXT:    vsetvli a1, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
 ; CHECK-NEXT:    vredmin.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-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:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
 ; 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:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
 ; 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:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
 ; 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 i32 @vreduce_add_nxv1i32(<vscale x 1 x i32> %v) {
 ; CHECK-LABEL: vreduce_add_nxv1i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, mu
 ; 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
 }
 
 declare i32 @llvm.vector.reduce.umax.nxv1i32(<vscale x 1 x i32>)
 
 define i32 @vreduce_umax_nxv1i32(<vscale x 1 x i32> %v) {
 ; CHECK-LABEL: vreduce_umax_nxv1i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, mu
 ; 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 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:    vsetvli a1, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, mu
 ; 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 i32 @vreduce_umin_nxv1i32(<vscale x 1 x i32> %v) {
 ; CHECK-LABEL: vreduce_umin_nxv1i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, mu
 ; 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 i32 @vreduce_smin_nxv1i32(<vscale x 1 x i32> %v) {
 ; CHECK-LABEL: vreduce_smin_nxv1i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, 524288
 ; CHECK-NEXT:    addi a0, a0, -1
-; CHECK-NEXT:    vsetvli a1, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, mu
 ; CHECK-NEXT:    vredmin.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-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 i32 @vreduce_and_nxv1i32(<vscale x 1 x i32> %v) {
 ; CHECK-LABEL: vreduce_and_nxv1i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, mu
 ; 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 i32 @vreduce_or_nxv1i32(<vscale x 1 x i32> %v) {
 ; CHECK-LABEL: vreduce_or_nxv1i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, mu
 ; 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 i32 @vreduce_xor_nxv1i32(<vscale x 1 x i32> %v) {
 ; CHECK-LABEL: vreduce_xor_nxv1i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, mu
 ; 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 i32 @vreduce_add_nxv2i32(<vscale x 2 x i32> %v) {
 ; CHECK-LABEL: vreduce_add_nxv2i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
 ; 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
 }
 
 declare i32 @llvm.vector.reduce.umax.nxv2i32(<vscale x 2 x i32>)
 
 define i32 @vreduce_umax_nxv2i32(<vscale x 2 x i32> %v) {
 ; CHECK-LABEL: vreduce_umax_nxv2i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
 ; 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 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:    vsetvli a1, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
 ; 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 i32 @vreduce_umin_nxv2i32(<vscale x 2 x i32> %v) {
 ; CHECK-LABEL: vreduce_umin_nxv2i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
 ; 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 i32 @vreduce_smin_nxv2i32(<vscale x 2 x i32> %v) {
 ; CHECK-LABEL: vreduce_smin_nxv2i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, 524288
 ; CHECK-NEXT:    addi a0, a0, -1
-; CHECK-NEXT:    vsetvli a1, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
 ; CHECK-NEXT:    vredmin.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-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 i32 @vreduce_and_nxv2i32(<vscale x 2 x i32> %v) {
 ; CHECK-LABEL: vreduce_and_nxv2i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
 ; 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 i32 @vreduce_or_nxv2i32(<vscale x 2 x i32> %v) {
 ; CHECK-LABEL: vreduce_or_nxv2i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
 ; 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 i32 @vreduce_xor_nxv2i32(<vscale x 2 x i32> %v) {
 ; CHECK-LABEL: vreduce_xor_nxv2i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
 ; 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>)
 
 define i32 @vreduce_add_nxv4i32(<vscale x 4 x i32> %v) {
 ; CHECK-LABEL: vreduce_add_nxv4i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, mu
 ; CHECK-NEXT:    vredsum.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i32 @llvm.vector.reduce.add.nxv4i32(<vscale x 4 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.umax.nxv4i32(<vscale x 4 x i32>)
 
 define i32 @vreduce_umax_nxv4i32(<vscale x 4 x i32> %v) {
 ; CHECK-LABEL: vreduce_umax_nxv4i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, mu
 ; CHECK-NEXT:    vredmaxu.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i32 @llvm.vector.reduce.umax.nxv4i32(<vscale x 4 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.smax.nxv4i32(<vscale x 4 x i32>)
 
 define i32 @vreduce_smax_nxv4i32(<vscale x 4 x i32> %v) {
 ; CHECK-LABEL: vreduce_smax_nxv4i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, 524288
-; CHECK-NEXT:    vsetvli a1, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v10, a0
 ; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, mu
 ; CHECK-NEXT:    vredmax.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i32 @llvm.vector.reduce.smax.nxv4i32(<vscale x 4 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.umin.nxv4i32(<vscale x 4 x i32>)
 
 define i32 @vreduce_umin_nxv4i32(<vscale x 4 x i32> %v) {
 ; CHECK-LABEL: vreduce_umin_nxv4i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, mu
 ; CHECK-NEXT:    vredminu.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i32 @llvm.vector.reduce.umin.nxv4i32(<vscale x 4 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.smin.nxv4i32(<vscale x 4 x i32>)
 
 define i32 @vreduce_smin_nxv4i32(<vscale x 4 x i32> %v) {
 ; CHECK-LABEL: vreduce_smin_nxv4i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, 524288
 ; CHECK-NEXT:    addi a0, a0, -1
-; CHECK-NEXT:    vsetvli a1, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v10, a0
 ; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, mu
 ; CHECK-NEXT:    vredmin.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i32 @llvm.vector.reduce.smin.nxv4i32(<vscale x 4 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.and.nxv4i32(<vscale x 4 x i32>)
 
 define i32 @vreduce_and_nxv4i32(<vscale x 4 x i32> %v) {
 ; CHECK-LABEL: vreduce_and_nxv4i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, mu
 ; CHECK-NEXT:    vredand.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i32 @llvm.vector.reduce.and.nxv4i32(<vscale x 4 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.or.nxv4i32(<vscale x 4 x i32>)
 
 define i32 @vreduce_or_nxv4i32(<vscale x 4 x i32> %v) {
 ; CHECK-LABEL: vreduce_or_nxv4i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, mu
 ; CHECK-NEXT:    vredor.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i32 @llvm.vector.reduce.or.nxv4i32(<vscale x 4 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.xor.nxv4i32(<vscale x 4 x i32>)
 
 define i32 @vreduce_xor_nxv4i32(<vscale x 4 x i32> %v) {
 ; CHECK-LABEL: vreduce_xor_nxv4i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, mu
 ; CHECK-NEXT:    vredxor.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i32 @llvm.vector.reduce.xor.nxv4i32(<vscale x 4 x i32> %v)
   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) {
 ; CHECK-LABEL: vreduce_add_nxv1i64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
+; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
 ; CHECK-NEXT:    vredsum.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vsrl.vx v8, v8, a1
 ; CHECK-NEXT:    vmv.x.s a1, v8
 ; CHECK-NEXT:    ret
   %red = call i64 @llvm.vector.reduce.add.nxv1i64(<vscale x 1 x i64> %v)
   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) {
 ; CHECK-LABEL: vreduce_umax_nxv1i64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
+; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
 ; CHECK-NEXT:    vredmaxu.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vsrl.vx v8, v8, a1
 ; CHECK-NEXT:    vmv.x.s a1, v8
 ; CHECK-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) {
 ; CHECK-LABEL: vreduce_smax_nxv1i64:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    addi sp, sp, -16
 ; CHECK-NEXT:    .cfi_def_cfa_offset 16
 ; CHECK-NEXT:    lui a0, 524288
 ; CHECK-NEXT:    sw a0, 12(sp)
 ; CHECK-NEXT:    sw zero, 8(sp)
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    addi a0, sp, 8
 ; CHECK-NEXT:    vlse64.v v9, (a0), zero
+; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
 ; CHECK-NEXT:    vredmax.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vsrl.vx v8, v8, a1
 ; CHECK-NEXT:    vmv.x.s a1, v8
 ; CHECK-NEXT:    addi sp, sp, 16
 ; CHECK-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) {
 ; CHECK-LABEL: vreduce_umin_nxv1i64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
+; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
 ; CHECK-NEXT:    vredminu.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vsrl.vx v8, v8, a1
 ; CHECK-NEXT:    vmv.x.s a1, v8
 ; CHECK-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) {
 ; CHECK-LABEL: vreduce_smin_nxv1i64:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    addi sp, sp, -16
 ; CHECK-NEXT:    .cfi_def_cfa_offset 16
 ; CHECK-NEXT:    li a0, -1
 ; CHECK-NEXT:    sw a0, 8(sp)
 ; CHECK-NEXT:    lui a0, 524288
 ; CHECK-NEXT:    addi a0, a0, -1
 ; CHECK-NEXT:    sw a0, 12(sp)
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    addi a0, sp, 8
 ; CHECK-NEXT:    vlse64.v v9, (a0), zero
+; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
 ; CHECK-NEXT:    vredmin.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vsrl.vx v8, v8, a1
 ; CHECK-NEXT:    vmv.x.s a1, v8
 ; CHECK-NEXT:    addi sp, sp, 16
 ; CHECK-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) {
 ; CHECK-LABEL: vreduce_and_nxv1i64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
+; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
 ; CHECK-NEXT:    vredand.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vsrl.vx v8, v8, a1
 ; CHECK-NEXT:    vmv.x.s a1, v8
 ; CHECK-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) {
 ; CHECK-LABEL: vreduce_or_nxv1i64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
+; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
 ; CHECK-NEXT:    vredor.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vsrl.vx v8, v8, a1
 ; CHECK-NEXT:    vmv.x.s a1, v8
 ; CHECK-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) {
 ; CHECK-LABEL: vreduce_xor_nxv1i64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
+; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
 ; CHECK-NEXT:    vredxor.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vsrl.vx v8, v8, a1
 ; CHECK-NEXT:    vmv.x.s a1, v8
 ; CHECK-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>)
 
 define i64 @vreduce_add_nxv2i64(<vscale x 2 x i64> %v) {
 ; CHECK-LABEL: vreduce_add_nxv2i64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m2, ta, mu
 ; CHECK-NEXT:    vredsum.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vsrl.vx v8, v8, a1
 ; CHECK-NEXT:    vmv.x.s a1, v8
 ; CHECK-NEXT:    ret
   %red = call i64 @llvm.vector.reduce.add.nxv2i64(<vscale x 2 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.umax.nxv2i64(<vscale x 2 x i64>)
 
 define i64 @vreduce_umax_nxv2i64(<vscale x 2 x i64> %v) {
 ; CHECK-LABEL: vreduce_umax_nxv2i64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m2, ta, mu
 ; CHECK-NEXT:    vredmaxu.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vsrl.vx v8, v8, a1
 ; CHECK-NEXT:    vmv.x.s a1, v8
 ; CHECK-NEXT:    ret
   %red = call i64 @llvm.vector.reduce.umax.nxv2i64(<vscale x 2 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.smax.nxv2i64(<vscale x 2 x i64>)
 
 define i64 @vreduce_smax_nxv2i64(<vscale x 2 x i64> %v) {
 ; CHECK-LABEL: vreduce_smax_nxv2i64:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    addi sp, sp, -16
 ; CHECK-NEXT:    .cfi_def_cfa_offset 16
 ; CHECK-NEXT:    lui a0, 524288
 ; CHECK-NEXT:    sw a0, 12(sp)
 ; CHECK-NEXT:    sw zero, 8(sp)
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    addi a0, sp, 8
 ; CHECK-NEXT:    vlse64.v v10, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m2, ta, mu
 ; CHECK-NEXT:    vredmax.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vsrl.vx v8, v8, a1
 ; CHECK-NEXT:    vmv.x.s a1, v8
 ; CHECK-NEXT:    addi sp, sp, 16
 ; CHECK-NEXT:    ret
   %red = call i64 @llvm.vector.reduce.smax.nxv2i64(<vscale x 2 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.umin.nxv2i64(<vscale x 2 x i64>)
 
 define i64 @vreduce_umin_nxv2i64(<vscale x 2 x i64> %v) {
 ; CHECK-LABEL: vreduce_umin_nxv2i64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m2, ta, mu
 ; CHECK-NEXT:    vredminu.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vsrl.vx v8, v8, a1
 ; CHECK-NEXT:    vmv.x.s a1, v8

 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    addi sp, sp, -16
 ; CHECK-NEXT:    .cfi_def_cfa_offset 16
 ; CHECK-NEXT:    li a0, -1
 ; CHECK-NEXT:    sw a0, 8(sp)
 ; CHECK-NEXT:    lui a0, 524288
 ; CHECK-NEXT:    addi a0, a0, -1
 ; CHECK-NEXT:    sw a0, 12(sp)
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    addi a0, sp, 8
 ; CHECK-NEXT:    vlse64.v v10, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m2, ta, mu
 ; CHECK-NEXT:    vredmin.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vsrl.vx v8, v8, a1
 ; CHECK-NEXT:    vmv.x.s a1, v8
 ; CHECK-NEXT:    addi sp, sp, 16
 ; CHECK-NEXT:    ret
   %red = call i64 @llvm.vector.reduce.smin.nxv2i64(<vscale x 2 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.and.nxv2i64(<vscale x 2 x i64>)
 
 define i64 @vreduce_and_nxv2i64(<vscale x 2 x i64> %v) {
 ; CHECK-LABEL: vreduce_and_nxv2i64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m2, ta, mu
 ; CHECK-NEXT:    vredand.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vsrl.vx v8, v8, a1
 ; CHECK-NEXT:    vmv.x.s a1, v8
 ; CHECK-NEXT:    ret
   %red = call i64 @llvm.vector.reduce.and.nxv2i64(<vscale x 2 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.or.nxv2i64(<vscale x 2 x i64>)
 
 define i64 @vreduce_or_nxv2i64(<vscale x 2 x i64> %v) {
 ; CHECK-LABEL: vreduce_or_nxv2i64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m2, ta, mu
 ; CHECK-NEXT:    vredor.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vsrl.vx v8, v8, a1
 ; CHECK-NEXT:    vmv.x.s a1, v8
 ; CHECK-NEXT:    ret
   %red = call i64 @llvm.vector.reduce.or.nxv2i64(<vscale x 2 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.xor.nxv2i64(<vscale x 2 x i64>)
 
 define i64 @vreduce_xor_nxv2i64(<vscale x 2 x i64> %v) {
 ; CHECK-LABEL: vreduce_xor_nxv2i64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m2, ta, mu
 ; CHECK-NEXT:    vredxor.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vsrl.vx v8, v8, a1
 ; CHECK-NEXT:    vmv.x.s a1, v8
 ; CHECK-NEXT:    ret
   %red = call i64 @llvm.vector.reduce.xor.nxv2i64(<vscale x 2 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.add.nxv4i64(<vscale x 4 x i64>)
 
 define i64 @vreduce_add_nxv4i64(<vscale x 4 x i64> %v) {
 ; CHECK-LABEL: vreduce_add_nxv4i64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v12, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m4, ta, mu
 ; CHECK-NEXT:    vredsum.vs v8, v8, v12
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vsrl.vx v8, v8, a1
 ; CHECK-NEXT:    vmv.x.s a1, v8
 ; CHECK-NEXT:    ret
   %red = call i64 @llvm.vector.reduce.add.nxv4i64(<vscale x 4 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.umax.nxv4i64(<vscale x 4 x i64>)
 
 define i64 @vreduce_umax_nxv4i64(<vscale x 4 x i64> %v) {
 ; CHECK-LABEL: vreduce_umax_nxv4i64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v12, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m4, ta, mu
 ; CHECK-NEXT:    vredmaxu.vs v8, v8, v12
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vsrl.vx v8, v8, a1
 ; CHECK-NEXT:    vmv.x.s a1, v8
 ; CHECK-NEXT:    ret
   %red = call i64 @llvm.vector.reduce.umax.nxv4i64(<vscale x 4 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.smax.nxv4i64(<vscale x 4 x i64>)
 
 define i64 @vreduce_smax_nxv4i64(<vscale x 4 x i64> %v) {
 ; CHECK-LABEL: vreduce_smax_nxv4i64:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    addi sp, sp, -16
 ; CHECK-NEXT:    .cfi_def_cfa_offset 16
 ; CHECK-NEXT:    lui a0, 524288
 ; CHECK-NEXT:    sw a0, 12(sp)
 ; CHECK-NEXT:    sw zero, 8(sp)
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    addi a0, sp, 8
 ; CHECK-NEXT:    vlse64.v v12, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m4, ta, mu
 ; CHECK-NEXT:    vredmax.vs v8, v8, v12
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vsrl.vx v8, v8, a1
 ; CHECK-NEXT:    vmv.x.s a1, v8
 ; CHECK-NEXT:    addi sp, sp, 16
 ; CHECK-NEXT:    ret
   %red = call i64 @llvm.vector.reduce.smax.nxv4i64(<vscale x 4 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.umin.nxv4i64(<vscale x 4 x i64>)
 
 define i64 @vreduce_umin_nxv4i64(<vscale x 4 x i64> %v) {
 ; CHECK-LABEL: vreduce_umin_nxv4i64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v12, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m4, ta, mu
 ; CHECK-NEXT:    vredminu.vs v8, v8, v12
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vsrl.vx v8, v8, a1
 ; CHECK-NEXT:    vmv.x.s a1, v8

 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    addi sp, sp, -16
 ; CHECK-NEXT:    .cfi_def_cfa_offset 16
 ; CHECK-NEXT:    li a0, -1
 ; CHECK-NEXT:    sw a0, 8(sp)
 ; CHECK-NEXT:    lui a0, 524288
 ; CHECK-NEXT:    addi a0, a0, -1
 ; CHECK-NEXT:    sw a0, 12(sp)
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    addi a0, sp, 8
 ; CHECK-NEXT:    vlse64.v v12, (a0), zero
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m4, ta, mu
 ; CHECK-NEXT:    vredmin.vs v8, v8, v12
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vsrl.vx v8, v8, a1
 ; CHECK-NEXT:    vmv.x.s a1, v8
 ; CHECK-NEXT:    addi sp, sp, 16
 ; CHECK-NEXT:    ret
   %red = call i64 @llvm.vector.reduce.smin.nxv4i64(<vscale x 4 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.and.nxv4i64(<vscale x 4 x i64>)
 
 define i64 @vreduce_and_nxv4i64(<vscale x 4 x i64> %v) {
 ; CHECK-LABEL: vreduce_and_nxv4i64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v12, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m4, ta, mu
 ; CHECK-NEXT:    vredand.vs v8, v8, v12
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vsrl.vx v8, v8, a1
 ; CHECK-NEXT:    vmv.x.s a1, v8
 ; CHECK-NEXT:    ret
   %red = call i64 @llvm.vector.reduce.and.nxv4i64(<vscale x 4 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.or.nxv4i64(<vscale x 4 x i64>)
 
 define i64 @vreduce_or_nxv4i64(<vscale x 4 x i64> %v) {
 ; CHECK-LABEL: vreduce_or_nxv4i64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v12, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m4, ta, mu
 ; CHECK-NEXT:    vredor.vs v8, v8, v12
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vsrl.vx v8, v8, a1
 ; CHECK-NEXT:    vmv.x.s a1, v8
 ; CHECK-NEXT:    ret
   %red = call i64 @llvm.vector.reduce.or.nxv4i64(<vscale x 4 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.xor.nxv4i64(<vscale x 4 x i64>)
 
 define i64 @vreduce_xor_nxv4i64(<vscale x 4 x i64> %v) {
 ; CHECK-LABEL: vreduce_xor_nxv4i64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v12, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m4, ta, mu
 ; CHECK-NEXT:    vredxor.vs v8, v8, v12
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    li a1, 32
 ; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vsrl.vx v8, v8, a1
 ; CHECK-NEXT:    vmv.x.s a1, v8
 ; CHECK-NEXT:    ret
   %red = call i64 @llvm.vector.reduce.xor.nxv4i64(<vscale x 4 x i64> %v)
   ret i64 %red
 }

llvm/test/CodeGen/RISCV/rvv/vreductions-int-rv64.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc -mtriple=riscv64 -mattr=+experimental-v -verify-machineinstrs < %s | FileCheck %s
 
 declare i8 @llvm.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:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf8, ta, mu
 ; 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:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf8, ta, mu
 ; 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:    vsetvli a1, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf8, ta, mu
 ; 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:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf8, ta, mu
 ; 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:    vsetvli a1, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf8, ta, mu
 ; 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:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf8, ta, mu
 ; 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:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf8, ta, mu
 ; 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:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf8, ta, mu
 ; 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:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf4, ta, mu
 ; 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:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf4, ta, mu
 ; 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:    vsetvli a1, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf4, ta, mu
 ; 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:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf4, ta, mu
 ; 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:    vsetvli a1, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf4, ta, mu
 ; 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:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf4, ta, mu
 ; 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:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf4, ta, mu
 ; 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:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf4, ta, mu
 ; 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:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf2, ta, mu
 ; 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:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf2, ta, mu
 ; 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:    vsetvli a1, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf2, ta, mu
 ; 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:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf2, ta, mu
 ; 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:    vsetvli a1, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf2, ta, mu
 ; 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:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf2, ta, mu
 ; 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:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf2, ta, mu
 ; 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:    vsetvli a0, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e8, mf2, ta, mu
 ; 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:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, mu
 ; 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
 }
 
 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:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, mu
 ; 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:    vsetvli a1, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, mu
 ; 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:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, mu
 ; 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) {
 ; CHECK-LABEL: vreduce_smin_nxv1i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, 8
 ; CHECK-NEXT:    addiw a0, a0, -1
-; CHECK-NEXT:    vsetvli a1, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, mu
 ; CHECK-NEXT:    vredmin.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-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:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, mu
 ; 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:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, mu
 ; 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:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf4, ta, mu
 ; 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:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, mu
 ; 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
 }
 
 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:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, mu
 ; 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:    vsetvli a1, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, mu
 ; 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:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, mu
 ; 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) {
 ; CHECK-LABEL: vreduce_smin_nxv2i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, 8
 ; CHECK-NEXT:    addiw a0, a0, -1
-; CHECK-NEXT:    vsetvli a1, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, mu
 ; CHECK-NEXT:    vredmin.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-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:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, mu
 ; 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:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, mu
 ; 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:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e16, mf2, ta, mu
 ; 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:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
 ; 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
 }
 
 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:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
 ; 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:    vsetvli a1, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
 ; 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:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
 ; 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) {
 ; CHECK-LABEL: vreduce_smin_nxv4i16:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, 8
 ; CHECK-NEXT:    addiw a0, a0, -1
-; CHECK-NEXT:    vsetvli a1, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
 ; CHECK-NEXT:    vredmin.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-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:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
 ; 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:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
 ; 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:    vsetvli a0, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
+; CHECK-NEXT:    vsetvli a0, zero, e16, m1, ta, mu
 ; 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:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, mu
 ; 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
 }
 
 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:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, mu
 ; 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:    vsetvli a1, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, mu
 ; 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:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, mu
 ; 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) {
 ; CHECK-LABEL: vreduce_smin_nxv1i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, 524288
 ; CHECK-NEXT:    addiw a0, a0, -1
-; CHECK-NEXT:    vsetvli a1, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, mu
 ; CHECK-NEXT:    vredmin.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-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:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, mu
 ; 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:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, mu
 ; 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:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e32, mf2, ta, mu
 ; 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:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
 ; 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
 }
 
 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:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
 ; 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:    vsetvli a1, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
 ; 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:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
 ; 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) {
 ; CHECK-LABEL: vreduce_smin_nxv2i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, 524288
 ; CHECK-NEXT:    addiw a0, a0, -1
-; CHECK-NEXT:    vsetvli a1, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
 ; CHECK-NEXT:    vredmin.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-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:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
 ; 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:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
 ; 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:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
+; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
 ; 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>)
 
 define signext i32 @vreduce_add_nxv4i32(<vscale x 4 x i32> %v) {
 ; CHECK-LABEL: vreduce_add_nxv4i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, mu
 ; CHECK-NEXT:    vredsum.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i32 @llvm.vector.reduce.add.nxv4i32(<vscale x 4 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.umax.nxv4i32(<vscale x 4 x i32>)
 
 define signext i32 @vreduce_umax_nxv4i32(<vscale x 4 x i32> %v) {
 ; CHECK-LABEL: vreduce_umax_nxv4i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, mu
 ; CHECK-NEXT:    vredmaxu.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i32 @llvm.vector.reduce.umax.nxv4i32(<vscale x 4 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.smax.nxv4i32(<vscale x 4 x i32>)
 
 define signext i32 @vreduce_smax_nxv4i32(<vscale x 4 x i32> %v) {
 ; CHECK-LABEL: vreduce_smax_nxv4i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, 524288
-; CHECK-NEXT:    vsetvli a1, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v10, a0
 ; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, mu
 ; CHECK-NEXT:    vredmax.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i32 @llvm.vector.reduce.smax.nxv4i32(<vscale x 4 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.umin.nxv4i32(<vscale x 4 x i32>)
 
 define signext i32 @vreduce_umin_nxv4i32(<vscale x 4 x i32> %v) {
 ; CHECK-LABEL: vreduce_umin_nxv4i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, mu
 ; CHECK-NEXT:    vredminu.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i32 @llvm.vector.reduce.umin.nxv4i32(<vscale x 4 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.smin.nxv4i32(<vscale x 4 x i32>)
 
 define signext i32 @vreduce_smin_nxv4i32(<vscale x 4 x i32> %v) {
 ; CHECK-LABEL: vreduce_smin_nxv4i32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a0, 524288
 ; CHECK-NEXT:    addiw a0, a0, -1
-; CHECK-NEXT:    vsetvli a1, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v10, a0
 ; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, mu
 ; CHECK-NEXT:    vredmin.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i32 @llvm.vector.reduce.smin.nxv4i32(<vscale x 4 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.and.nxv4i32(<vscale x 4 x i32>)
 
 define signext i32 @vreduce_and_nxv4i32(<vscale x 4 x i32> %v) {
 ; CHECK-LABEL: vreduce_and_nxv4i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, mu
 ; CHECK-NEXT:    vredand.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i32 @llvm.vector.reduce.and.nxv4i32(<vscale x 4 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.or.nxv4i32(<vscale x 4 x i32>)
 
 define signext i32 @vreduce_or_nxv4i32(<vscale x 4 x i32> %v) {
 ; CHECK-LABEL: vreduce_or_nxv4i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, mu
 ; CHECK-NEXT:    vredor.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i32 @llvm.vector.reduce.or.nxv4i32(<vscale x 4 x i32> %v)
   ret i32 %red
 }
 
 declare i32 @llvm.vector.reduce.xor.nxv4i32(<vscale x 4 x i32>)
 
 define signext i32 @vreduce_xor_nxv4i32(<vscale x 4 x i32> %v) {
 ; CHECK-LABEL: vreduce_xor_nxv4i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e32, m2, ta, mu
 ; CHECK-NEXT:    vredxor.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i32 @llvm.vector.reduce.xor.nxv4i32(<vscale x 4 x i32> %v)
   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) {
 ; CHECK-LABEL: vreduce_add_nxv1i64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
+; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
 ; CHECK-NEXT:    vredsum.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i64 @llvm.vector.reduce.add.nxv1i64(<vscale x 1 x i64> %v)
   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) {
 ; CHECK-LABEL: vreduce_umax_nxv1i64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
+; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
 ; CHECK-NEXT:    vredmaxu.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-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) {
 ; CHECK-LABEL: vreduce_smax_nxv1i64:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a0, -1
 ; CHECK-NEXT:    slli a0, a0, 63
-; CHECK-NEXT:    vsetvli a1, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
 ; CHECK-NEXT:    vredmax.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-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) {
 ; CHECK-LABEL: vreduce_umin_nxv1i64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
+; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
 ; CHECK-NEXT:    vredminu.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-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) {
 ; CHECK-LABEL: vreduce_smin_nxv1i64:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a0, -1
 ; CHECK-NEXT:    srli a0, a0, 1
-; CHECK-NEXT:    vsetvli a1, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
+; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
 ; CHECK-NEXT:    vredmin.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-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) {
 ; CHECK-LABEL: vreduce_and_nxv1i64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, -1
+; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
 ; CHECK-NEXT:    vredand.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-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) {
 ; CHECK-LABEL: vreduce_or_nxv1i64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
+; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
 ; CHECK-NEXT:    vredor.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-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) {
 ; CHECK-LABEL: vreduce_xor_nxv1i64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v9, 0
+; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
 ; CHECK-NEXT:    vredxor.vs v8, v8, v9
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-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>)
 
 define i64 @vreduce_add_nxv2i64(<vscale x 2 x i64> %v) {
 ; CHECK-LABEL: vreduce_add_nxv2i64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m2, ta, mu
 ; CHECK-NEXT:    vredsum.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i64 @llvm.vector.reduce.add.nxv2i64(<vscale x 2 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.umax.nxv2i64(<vscale x 2 x i64>)
 
 define i64 @vreduce_umax_nxv2i64(<vscale x 2 x i64> %v) {
 ; CHECK-LABEL: vreduce_umax_nxv2i64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m2, ta, mu
 ; CHECK-NEXT:    vredmaxu.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i64 @llvm.vector.reduce.umax.nxv2i64(<vscale x 2 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.smax.nxv2i64(<vscale x 2 x i64>)
 
 define i64 @vreduce_smax_nxv2i64(<vscale x 2 x i64> %v) {
 ; CHECK-LABEL: vreduce_smax_nxv2i64:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a0, -1
 ; CHECK-NEXT:    slli a0, a0, 63
-; CHECK-NEXT:    vsetvli a1, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v10, a0
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m2, ta, mu
 ; CHECK-NEXT:    vredmax.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i64 @llvm.vector.reduce.smax.nxv2i64(<vscale x 2 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.umin.nxv2i64(<vscale x 2 x i64>)
 
 define i64 @vreduce_umin_nxv2i64(<vscale x 2 x i64> %v) {
 ; CHECK-LABEL: vreduce_umin_nxv2i64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m2, ta, mu
 ; CHECK-NEXT:    vredminu.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i64 @llvm.vector.reduce.umin.nxv2i64(<vscale x 2 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.smin.nxv2i64(<vscale x 2 x i64>)
 
 define i64 @vreduce_smin_nxv2i64(<vscale x 2 x i64> %v) {
 ; CHECK-LABEL: vreduce_smin_nxv2i64:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a0, -1
 ; CHECK-NEXT:    srli a0, a0, 1
-; CHECK-NEXT:    vsetvli a1, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v10, a0
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m2, ta, mu
 ; CHECK-NEXT:    vredmin.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i64 @llvm.vector.reduce.smin.nxv2i64(<vscale x 2 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.and.nxv2i64(<vscale x 2 x i64>)
 
 define i64 @vreduce_and_nxv2i64(<vscale x 2 x i64> %v) {
 ; CHECK-LABEL: vreduce_and_nxv2i64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m2, ta, mu
 ; CHECK-NEXT:    vredand.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i64 @llvm.vector.reduce.and.nxv2i64(<vscale x 2 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.or.nxv2i64(<vscale x 2 x i64>)
 
 define i64 @vreduce_or_nxv2i64(<vscale x 2 x i64> %v) {
 ; CHECK-LABEL: vreduce_or_nxv2i64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m2, ta, mu
 ; CHECK-NEXT:    vredor.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i64 @llvm.vector.reduce.or.nxv2i64(<vscale x 2 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.xor.nxv2i64(<vscale x 2 x i64>)
 
 define i64 @vreduce_xor_nxv2i64(<vscale x 2 x i64> %v) {
 ; CHECK-LABEL: vreduce_xor_nxv2i64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v10, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m2, ta, mu
 ; CHECK-NEXT:    vredxor.vs v8, v8, v10
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i64 @llvm.vector.reduce.xor.nxv2i64(<vscale x 2 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.add.nxv4i64(<vscale x 4 x i64>)
 
 define i64 @vreduce_add_nxv4i64(<vscale x 4 x i64> %v) {
 ; CHECK-LABEL: vreduce_add_nxv4i64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v12, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m4, ta, mu
 ; CHECK-NEXT:    vredsum.vs v8, v8, v12
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i64 @llvm.vector.reduce.add.nxv4i64(<vscale x 4 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.umax.nxv4i64(<vscale x 4 x i64>)
 
 define i64 @vreduce_umax_nxv4i64(<vscale x 4 x i64> %v) {
 ; CHECK-LABEL: vreduce_umax_nxv4i64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v12, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m4, ta, mu
 ; CHECK-NEXT:    vredmaxu.vs v8, v8, v12
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i64 @llvm.vector.reduce.umax.nxv4i64(<vscale x 4 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.smax.nxv4i64(<vscale x 4 x i64>)
 
 define i64 @vreduce_smax_nxv4i64(<vscale x 4 x i64> %v) {
 ; CHECK-LABEL: vreduce_smax_nxv4i64:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a0, -1
 ; CHECK-NEXT:    slli a0, a0, 63
-; CHECK-NEXT:    vsetvli a1, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v12, a0
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m4, ta, mu
 ; CHECK-NEXT:    vredmax.vs v8, v8, v12
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i64 @llvm.vector.reduce.smax.nxv4i64(<vscale x 4 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.umin.nxv4i64(<vscale x 4 x i64>)
 
 define i64 @vreduce_umin_nxv4i64(<vscale x 4 x i64> %v) {
 ; CHECK-LABEL: vreduce_umin_nxv4i64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v12, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m4, ta, mu
 ; CHECK-NEXT:    vredminu.vs v8, v8, v12
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i64 @llvm.vector.reduce.umin.nxv4i64(<vscale x 4 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.smin.nxv4i64(<vscale x 4 x i64>)
 
 define i64 @vreduce_smin_nxv4i64(<vscale x 4 x i64> %v) {
 ; CHECK-LABEL: vreduce_smin_nxv4i64:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    li a0, -1
 ; CHECK-NEXT:    srli a0, a0, 1
-; CHECK-NEXT:    vsetvli a1, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v12, a0
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m4, ta, mu
 ; CHECK-NEXT:    vredmin.vs v8, v8, v12
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i64 @llvm.vector.reduce.smin.nxv4i64(<vscale x 4 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.and.nxv4i64(<vscale x 4 x i64>)
 
 define i64 @vreduce_and_nxv4i64(<vscale x 4 x i64> %v) {
 ; CHECK-LABEL: vreduce_and_nxv4i64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v12, -1
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m4, ta, mu
 ; CHECK-NEXT:    vredand.vs v8, v8, v12
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i64 @llvm.vector.reduce.and.nxv4i64(<vscale x 4 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.or.nxv4i64(<vscale x 4 x i64>)
 
 define i64 @vreduce_or_nxv4i64(<vscale x 4 x i64> %v) {
 ; CHECK-LABEL: vreduce_or_nxv4i64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v12, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m4, ta, mu
 ; CHECK-NEXT:    vredor.vs v8, v8, v12
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i64 @llvm.vector.reduce.or.nxv4i64(<vscale x 4 x i64> %v)
   ret i64 %red
 }
 
 declare i64 @llvm.vector.reduce.xor.nxv4i64(<vscale x 4 x i64>)
 
 define i64 @vreduce_xor_nxv4i64(<vscale x 4 x i64> %v) {
 ; CHECK-LABEL: vreduce_xor_nxv4i64:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a0, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.i v12, 0
 ; CHECK-NEXT:    vsetvli a0, zero, e64, m4, ta, mu
 ; CHECK-NEXT:    vredxor.vs v8, v8, v12
 ; CHECK-NEXT:    vmv.x.s a0, v8
 ; CHECK-NEXT:    ret
   %red = call i64 @llvm.vector.reduce.xor.nxv4i64(<vscale x 4 x i64> %v)
   ret i64 %red
 }

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=+experimental-v -verify-machineinstrs < %s \
 ; RUN:   | FileCheck %s --check-prefixes=CHECK,RV32
 ; RUN: llc -mtriple=riscv64 -mattr=+experimental-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:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, tu, mu
 ; 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:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, tu, mu
 ; 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:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, tu, mu
 ; 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:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, tu, mu
 ; 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:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, tu, mu
 ; 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:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, tu, mu
 ; 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:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, tu, mu
 ; 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:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, mf8, tu, mu
 ; 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:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, tu, mu
 ; 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:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, tu, mu
 ; 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:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, tu, mu
 ; 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:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, tu, mu
 ; 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:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, tu, mu
 ; 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:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, tu, mu
 ; 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:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, tu, mu
 ; 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:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, mf4, tu, mu
 ; 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.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:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, tu, mu
 ; 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:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, tu, mu
 ; 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:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, tu, mu
 ; 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:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, tu, mu
 ; 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:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, tu, mu
 ; 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:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, tu, mu
 ; 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:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, tu, mu
 ; 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:    vsetvli a2, zero, e8, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e8, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e8, mf2, tu, mu
 ; 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:    vsetvli a2, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, mf4, tu, mu
 ; 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:    lui a2, 16
 ; RV32-NEXT:    addi a2, a2, -1
 ; RV32-NEXT:    and a0, a0, a2
-; RV32-NEXT:    vsetvli a2, zero, e16, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; RV32-NEXT:    vmv.v.x v9, a0
 ; RV32-NEXT:    vsetvli zero, a1, e16, mf4, tu, mu
 ; 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:    lui a2, 16
 ; RV64-NEXT:    addiw a2, a2, -1
 ; RV64-NEXT:    and a0, a0, a2
-; RV64-NEXT:    vsetvli a2, zero, e16, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vsetvli zero, a1, e16, mf4, tu, mu
 ; RV64-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v9
 ; RV64-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:    vsetvli a2, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, mf4, tu, mu
 ; 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:    lui a2, 16
 ; RV32-NEXT:    addi a2, a2, -1
 ; RV32-NEXT:    and a0, a0, a2
-; RV32-NEXT:    vsetvli a2, zero, e16, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; RV32-NEXT:    vmv.v.x v9, a0
 ; RV32-NEXT:    vsetvli zero, a1, e16, mf4, tu, mu
 ; 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:    lui a2, 16
 ; RV64-NEXT:    addiw a2, a2, -1
 ; RV64-NEXT:    and a0, a0, a2
-; RV64-NEXT:    vsetvli a2, zero, e16, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vsetvli zero, a1, e16, mf4, tu, mu
 ; RV64-NEXT:    vredminu.vs v9, v8, v9, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v9
 ; RV64-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:    vsetvli a2, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, mf4, tu, mu
 ; 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:    vsetvli a2, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, mf4, tu, mu
 ; 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:    vsetvli a2, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, mf4, tu, mu
 ; 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:    vsetvli a2, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, mf4, tu, mu
 ; 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:    vsetvli a2, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, mf2, tu, mu
 ; 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:    lui a2, 16
 ; RV32-NEXT:    addi a2, a2, -1
 ; RV32-NEXT:    and a0, a0, a2
-; RV32-NEXT:    vsetvli a2, zero, e16, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; RV32-NEXT:    vmv.v.x v9, a0
 ; RV32-NEXT:    vsetvli zero, a1, e16, mf2, tu, mu
 ; 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:    lui a2, 16
 ; RV64-NEXT:    addiw a2, a2, -1
 ; RV64-NEXT:    and a0, a0, a2
-; RV64-NEXT:    vsetvli a2, zero, e16, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vsetvli zero, a1, e16, mf2, tu, mu
 ; RV64-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v9
 ; RV64-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:    vsetvli a2, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, mf2, tu, mu
 ; 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:    lui a2, 16
 ; RV32-NEXT:    addi a2, a2, -1
 ; RV32-NEXT:    and a0, a0, a2
-; RV32-NEXT:    vsetvli a2, zero, e16, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; RV32-NEXT:    vmv.v.x v9, a0
 ; RV32-NEXT:    vsetvli zero, a1, e16, mf2, tu, mu
 ; 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:    lui a2, 16
 ; RV64-NEXT:    addiw a2, a2, -1
 ; RV64-NEXT:    and a0, a0, a2
-; RV64-NEXT:    vsetvli a2, zero, e16, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vsetvli zero, a1, e16, mf2, tu, mu
 ; RV64-NEXT:    vredminu.vs v9, v8, v9, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v9
 ; RV64-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:    vsetvli a2, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, mf2, tu, mu
 ; 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:    vsetvli a2, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, mf2, tu, mu
 ; 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:    vsetvli a2, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, mf2, tu, mu
 ; 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:    vsetvli a2, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, mf2, tu, mu
 ; 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:    vsetvli a2, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m1, tu, mu
 ; 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:    lui a2, 16
 ; RV32-NEXT:    addi a2, a2, -1
 ; RV32-NEXT:    and a0, a0, a2
-; RV32-NEXT:    vsetvli a2, zero, e16, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; RV32-NEXT:    vmv.v.x v9, a0
 ; RV32-NEXT:    vsetvli zero, a1, e16, m1, tu, mu
 ; 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:    lui a2, 16
 ; RV64-NEXT:    addiw a2, a2, -1
 ; RV64-NEXT:    and a0, a0, a2
-; RV64-NEXT:    vsetvli a2, zero, e16, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vsetvli zero, a1, e16, m1, tu, mu
 ; RV64-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v9
 ; RV64-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:    vsetvli a2, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m1, tu, mu
 ; 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:    lui a2, 16
 ; RV32-NEXT:    addi a2, a2, -1
 ; RV32-NEXT:    and a0, a0, a2
-; RV32-NEXT:    vsetvli a2, zero, e16, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; RV32-NEXT:    vmv.v.x v9, a0
 ; RV32-NEXT:    vsetvli zero, a1, e16, m1, tu, mu
 ; 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:    lui a2, 16
 ; RV64-NEXT:    addiw a2, a2, -1
 ; RV64-NEXT:    and a0, a0, a2
-; RV64-NEXT:    vsetvli a2, zero, e16, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vsetvli zero, a1, e16, m1, tu, mu
 ; RV64-NEXT:    vredminu.vs v9, v8, v9, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v9
 ; RV64-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:    vsetvli a2, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m1, tu, mu
 ; 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:    vsetvli a2, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m1, tu, mu
 ; 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:    vsetvli a2, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m1, tu, mu
 ; 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:    vsetvli a2, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e16, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e16, m1, tu, mu
 ; 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:    vsetvli a2, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, tu, mu
 ; 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:    vsetvli a2, zero, e32, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; RV32-NEXT:    vmv.v.x v9, a0
 ; RV32-NEXT:    vsetvli zero, a1, e32, mf2, tu, mu
 ; 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:    vsetvli a2, zero, e32, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vsetvli zero, a1, e32, mf2, tu, mu
 ; RV64-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v9
 ; RV64-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:    vsetvli a2, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, tu, mu
 ; 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:    vsetvli a2, zero, e32, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; RV32-NEXT:    vmv.v.x v9, a0
 ; RV32-NEXT:    vsetvli zero, a1, e32, mf2, tu, mu
 ; 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:    vsetvli a2, zero, e32, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vsetvli zero, a1, e32, mf2, tu, mu
 ; RV64-NEXT:    vredminu.vs v9, v8, v9, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v9
 ; RV64-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:    vsetvli a2, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, tu, mu
 ; 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:    vsetvli a2, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, tu, mu
 ; 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:    vsetvli a2, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, tu, mu
 ; 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:    vsetvli a2, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e32, mf2, tu, mu
 ; 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:    vsetvli a2, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m1, tu, mu
 ; 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:    vsetvli a2, zero, e32, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; RV32-NEXT:    vmv.v.x v9, a0
 ; RV32-NEXT:    vsetvli zero, a1, e32, m1, tu, mu
 ; 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:    vsetvli a2, zero, e32, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vsetvli zero, a1, e32, m1, tu, mu
 ; RV64-NEXT:    vredmaxu.vs v9, v8, v9, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v9
 ; RV64-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:    vsetvli a2, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m1, tu, mu
 ; 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:    vsetvli a2, zero, e32, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; RV32-NEXT:    vmv.v.x v9, a0
 ; RV32-NEXT:    vsetvli zero, a1, e32, m1, tu, mu
 ; 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:    vsetvli a2, zero, e32, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vsetvli zero, a1, e32, m1, tu, mu
 ; RV64-NEXT:    vredminu.vs v9, v8, v9, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v9
 ; RV64-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:    vsetvli a2, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m1, tu, mu
 ; 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:    vsetvli a2, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m1, tu, mu
 ; 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:    vsetvli a2, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m1, tu, mu
 ; 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:    vsetvli a2, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v9, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m1, tu, mu
 ; 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 i32 @vpreduce_add_nxv4i32(i32 signext %s, <vscale x 4 x i32> %v, <vscale x 4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_add_nxv4i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a2, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v10, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m2, tu, mu
 ; CHECK-NEXT:    vredsum.vs v10, v8, v10, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v10
 ; CHECK-NEXT:    ret
   %r = call i32 @llvm.vp.reduce.add.nxv4i32(i32 %s, <vscale x 4 x i32> %v, <vscale x 4 x i1> %m, i32 %evl)
   ret i32 %r
 }
 
 declare i32 @llvm.vp.reduce.umax.nxv4i32(i32, <vscale x 4 x i32>, <vscale x 4 x i1>, i32)
 
 define signext i32 @vpreduce_umax_nxv4i32(i32 signext %s, <vscale x 4 x i32> %v, <vscale x 4 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_umax_nxv4i32:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetvli a2, zero, e32, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; RV32-NEXT:    vmv.v.x v10, a0
 ; RV32-NEXT:    vsetvli zero, a1, e32, m2, tu, mu
 ; RV32-NEXT:    vredmaxu.vs v10, v8, v10, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v10
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_umax_nxv4i32:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    slli a0, a0, 32
 ; RV64-NEXT:    srli a0, a0, 32
-; RV64-NEXT:    vsetvli a2, zero, e32, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v10, a0
 ; RV64-NEXT:    vsetvli zero, a1, e32, m2, tu, mu
 ; RV64-NEXT:    vredmaxu.vs v10, v8, v10, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v10
 ; RV64-NEXT:    ret
   %r = call i32 @llvm.vp.reduce.umax.nxv4i32(i32 %s, <vscale x 4 x i32> %v, <vscale x 4 x i1> %m, i32 %evl)
   ret i32 %r
 }
 
 declare i32 @llvm.vp.reduce.smax.nxv4i32(i32, <vscale x 4 x i32>, <vscale x 4 x i1>, i32)
 
 define signext i32 @vpreduce_smax_nxv4i32(i32 signext %s, <vscale x 4 x i32> %v, <vscale x 4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_smax_nxv4i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a2, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v10, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m2, tu, mu
 ; CHECK-NEXT:    vredmax.vs v10, v8, v10, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v10
 ; CHECK-NEXT:    ret
   %r = call i32 @llvm.vp.reduce.smax.nxv4i32(i32 %s, <vscale x 4 x i32> %v, <vscale x 4 x i1> %m, i32 %evl)
   ret i32 %r
 }
 
 declare i32 @llvm.vp.reduce.umin.nxv4i32(i32, <vscale x 4 x i32>, <vscale x 4 x i1>, i32)
 
 define signext i32 @vpreduce_umin_nxv4i32(i32 signext %s, <vscale x 4 x i32> %v, <vscale x 4 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_umin_nxv4i32:
 ; RV32:       # %bb.0:
-; RV32-NEXT:    vsetvli a2, zero, e32, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; RV32-NEXT:    vmv.v.x v10, a0
 ; RV32-NEXT:    vsetvli zero, a1, e32, m2, tu, mu
 ; RV32-NEXT:    vredminu.vs v10, v8, v10, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v10
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_umin_nxv4i32:
 ; RV64:       # %bb.0:
 ; RV64-NEXT:    slli a0, a0, 32
 ; RV64-NEXT:    srli a0, a0, 32
-; RV64-NEXT:    vsetvli a2, zero, e32, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v10, a0
 ; RV64-NEXT:    vsetvli zero, a1, e32, m2, tu, mu
 ; RV64-NEXT:    vredminu.vs v10, v8, v10, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v10
 ; RV64-NEXT:    ret
   %r = call i32 @llvm.vp.reduce.umin.nxv4i32(i32 %s, <vscale x 4 x i32> %v, <vscale x 4 x i1> %m, i32 %evl)
   ret i32 %r
 }
 
 declare i32 @llvm.vp.reduce.smin.nxv4i32(i32, <vscale x 4 x i32>, <vscale x 4 x i1>, i32)
 
 define signext i32 @vpreduce_smin_nxv4i32(i32 signext %s, <vscale x 4 x i32> %v, <vscale x 4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_smin_nxv4i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a2, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v10, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m2, tu, mu
 ; CHECK-NEXT:    vredmin.vs v10, v8, v10, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v10
 ; CHECK-NEXT:    ret
   %r = call i32 @llvm.vp.reduce.smin.nxv4i32(i32 %s, <vscale x 4 x i32> %v, <vscale x 4 x i1> %m, i32 %evl)
   ret i32 %r
 }
 
 declare i32 @llvm.vp.reduce.and.nxv4i32(i32, <vscale x 4 x i32>, <vscale x 4 x i1>, i32)
 
 define signext i32 @vpreduce_and_nxv4i32(i32 signext %s, <vscale x 4 x i32> %v, <vscale x 4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_and_nxv4i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a2, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v10, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m2, tu, mu
 ; CHECK-NEXT:    vredand.vs v10, v8, v10, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v10
 ; CHECK-NEXT:    ret
   %r = call i32 @llvm.vp.reduce.and.nxv4i32(i32 %s, <vscale x 4 x i32> %v, <vscale x 4 x i1> %m, i32 %evl)
   ret i32 %r
 }
 
 declare i32 @llvm.vp.reduce.or.nxv4i32(i32, <vscale x 4 x i32>, <vscale x 4 x i1>, i32)
 
 define signext i32 @vpreduce_or_nxv4i32(i32 signext %s, <vscale x 4 x i32> %v, <vscale x 4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_or_nxv4i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a2, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v10, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m2, tu, mu
 ; CHECK-NEXT:    vredor.vs v10, v8, v10, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v10
 ; CHECK-NEXT:    ret
   %r = call i32 @llvm.vp.reduce.or.nxv4i32(i32 %s, <vscale x 4 x i32> %v, <vscale x 4 x i1> %m, i32 %evl)
   ret i32 %r
 }
 
 declare i32 @llvm.vp.reduce.xor.nxv4i32(i32, <vscale x 4 x i32>, <vscale x 4 x i1>, i32)
 
 define signext i32 @vpreduce_xor_nxv4i32(i32 signext %s, <vscale x 4 x i32> %v, <vscale x 4 x i1> %m, i32 zeroext %evl) {
 ; CHECK-LABEL: vpreduce_xor_nxv4i32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vsetvli a2, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vsetivli zero, 1, e32, m1, ta, mu
 ; CHECK-NEXT:    vmv.v.x v10, a0
 ; CHECK-NEXT:    vsetvli zero, a1, e32, m2, tu, mu
 ; CHECK-NEXT:    vredxor.vs v10, v8, v10, v0.t
 ; CHECK-NEXT:    vmv.x.s a0, v10
 ; CHECK-NEXT:    ret
   %r = call i32 @llvm.vp.reduce.xor.nxv4i32(i32 %s, <vscale x 4 x i32> %v, <vscale x 4 x i1> %m, i32 %evl)
   ret i32 %r
 }
 
 declare i64 @llvm.vp.reduce.add.nxv1i64(i64, <vscale x 1 x i64>, <vscale x 1 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v9, (a0), zero
 ; RV32-NEXT:    vsetvli zero, a2, e64, m1, tu, mu
 ; 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, mu
 ; 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:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vsetvli zero, a1, e64, m1, tu, mu
 ; 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
 }
 
 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v9, (a0), zero
 ; RV32-NEXT:    vsetvli zero, a2, e64, m1, tu, mu
 ; 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, mu
 ; 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:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vsetvli zero, a1, e64, m1, tu, mu
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v9, (a0), zero
 ; RV32-NEXT:    vsetvli zero, a2, e64, m1, tu, mu
 ; 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, mu
 ; 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:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vsetvli zero, a1, e64, m1, tu, mu
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v9, (a0), zero
 ; RV32-NEXT:    vsetvli zero, a2, e64, m1, tu, mu
 ; 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, mu
 ; 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:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vsetvli zero, a1, e64, m1, tu, mu
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v9, (a0), zero
 ; RV32-NEXT:    vsetvli zero, a2, e64, m1, tu, mu
 ; 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, mu
 ; 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:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vsetvli zero, a1, e64, m1, tu, mu
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v9, (a0), zero
 ; RV32-NEXT:    vsetvli zero, a2, e64, m1, tu, mu
 ; 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, mu
 ; 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:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vsetvli zero, a1, e64, m1, tu, mu
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v9, (a0), zero
 ; RV32-NEXT:    vsetvli zero, a2, e64, m1, tu, mu
 ; 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, mu
 ; 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:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vsetvli zero, a1, e64, m1, tu, mu
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v9, (a0), zero
 ; RV32-NEXT:    vsetvli zero, a2, e64, m1, tu, mu
 ; 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, mu
 ; 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:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v9, a0
 ; RV64-NEXT:    vsetvli zero, a1, e64, m1, tu, mu
 ; 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)
 
 define signext i64 @vpreduce_add_nxv2i64(i64 signext %s, <vscale x 2 x i64> %v, <vscale x 2 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_add_nxv2i64:
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v10, (a0), zero
 ; RV32-NEXT:    vsetvli zero, a2, e64, m2, tu, mu
 ; RV32-NEXT:    vredsum.vs v10, v8, v10, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v10
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v10, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_add_nxv2i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v10, a0
 ; RV64-NEXT:    vsetvli zero, a1, e64, m2, tu, mu
 ; RV64-NEXT:    vredsum.vs v10, v8, v10, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v10
 ; RV64-NEXT:    ret
   %r = call i64 @llvm.vp.reduce.add.nxv2i64(i64 %s, <vscale x 2 x i64> %v, <vscale x 2 x i1> %m, i32 %evl)
   ret i64 %r
 }
 
 declare i64 @llvm.vp.reduce.umax.nxv2i64(i64, <vscale x 2 x i64>, <vscale x 2 x i1>, i32)
 
 define signext i64 @vpreduce_umax_nxv2i64(i64 signext %s, <vscale x 2 x i64> %v, <vscale x 2 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_umax_nxv2i64:
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v10, (a0), zero
 ; RV32-NEXT:    vsetvli zero, a2, e64, m2, tu, mu
 ; RV32-NEXT:    vredmaxu.vs v10, v8, v10, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v10
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v10, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_umax_nxv2i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v10, a0
 ; RV64-NEXT:    vsetvli zero, a1, e64, m2, tu, mu
 ; RV64-NEXT:    vredmaxu.vs v10, v8, v10, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v10
 ; RV64-NEXT:    ret
   %r = call i64 @llvm.vp.reduce.umax.nxv2i64(i64 %s, <vscale x 2 x i64> %v, <vscale x 2 x i1> %m, i32 %evl)
   ret i64 %r
 }
 
 declare i64 @llvm.vp.reduce.smax.nxv2i64(i64, <vscale x 2 x i64>, <vscale x 2 x i1>, i32)
 
 define signext i64 @vpreduce_smax_nxv2i64(i64 signext %s, <vscale x 2 x i64> %v, <vscale x 2 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_smax_nxv2i64:
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v10, (a0), zero
 ; RV32-NEXT:    vsetvli zero, a2, e64, m2, tu, mu
 ; RV32-NEXT:    vredmax.vs v10, v8, v10, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v10
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v10, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_smax_nxv2i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v10, a0
 ; RV64-NEXT:    vsetvli zero, a1, e64, m2, tu, mu
 ; RV64-NEXT:    vredmax.vs v10, v8, v10, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v10
 ; RV64-NEXT:    ret
   %r = call i64 @llvm.vp.reduce.smax.nxv2i64(i64 %s, <vscale x 2 x i64> %v, <vscale x 2 x i1> %m, i32 %evl)
   ret i64 %r
 }
 
 declare i64 @llvm.vp.reduce.umin.nxv2i64(i64, <vscale x 2 x i64>, <vscale x 2 x i1>, i32)
 
 define signext i64 @vpreduce_umin_nxv2i64(i64 signext %s, <vscale x 2 x i64> %v, <vscale x 2 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_umin_nxv2i64:
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v10, (a0), zero
 ; RV32-NEXT:    vsetvli zero, a2, e64, m2, tu, mu
 ; RV32-NEXT:    vredminu.vs v10, v8, v10, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v10
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v10, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_umin_nxv2i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v10, a0
 ; RV64-NEXT:    vsetvli zero, a1, e64, m2, tu, mu
 ; RV64-NEXT:    vredminu.vs v10, v8, v10, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v10
 ; RV64-NEXT:    ret
   %r = call i64 @llvm.vp.reduce.umin.nxv2i64(i64 %s, <vscale x 2 x i64> %v, <vscale x 2 x i1> %m, i32 %evl)
   ret i64 %r
 }
 
 declare i64 @llvm.vp.reduce.smin.nxv2i64(i64, <vscale x 2 x i64>, <vscale x 2 x i1>, i32)
 
 define signext i64 @vpreduce_smin_nxv2i64(i64 signext %s, <vscale x 2 x i64> %v, <vscale x 2 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_smin_nxv2i64:
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v10, (a0), zero
 ; RV32-NEXT:    vsetvli zero, a2, e64, m2, tu, mu
 ; RV32-NEXT:    vredmin.vs v10, v8, v10, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v10
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v10, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_smin_nxv2i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v10, a0
 ; RV64-NEXT:    vsetvli zero, a1, e64, m2, tu, mu
 ; RV64-NEXT:    vredmin.vs v10, v8, v10, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v10
 ; RV64-NEXT:    ret
   %r = call i64 @llvm.vp.reduce.smin.nxv2i64(i64 %s, <vscale x 2 x i64> %v, <vscale x 2 x i1> %m, i32 %evl)
   ret i64 %r
 }
 
 declare i64 @llvm.vp.reduce.and.nxv2i64(i64, <vscale x 2 x i64>, <vscale x 2 x i1>, i32)
 
 define signext i64 @vpreduce_and_nxv2i64(i64 signext %s, <vscale x 2 x i64> %v, <vscale x 2 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_and_nxv2i64:
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v10, (a0), zero
 ; RV32-NEXT:    vsetvli zero, a2, e64, m2, tu, mu
 ; RV32-NEXT:    vredand.vs v10, v8, v10, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v10
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v10, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_and_nxv2i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v10, a0
 ; RV64-NEXT:    vsetvli zero, a1, e64, m2, tu, mu
 ; RV64-NEXT:    vredand.vs v10, v8, v10, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v10
 ; RV64-NEXT:    ret
   %r = call i64 @llvm.vp.reduce.and.nxv2i64(i64 %s, <vscale x 2 x i64> %v, <vscale x 2 x i1> %m, i32 %evl)
   ret i64 %r
 }
 
 declare i64 @llvm.vp.reduce.or.nxv2i64(i64, <vscale x 2 x i64>, <vscale x 2 x i1>, i32)
 
 define signext i64 @vpreduce_or_nxv2i64(i64 signext %s, <vscale x 2 x i64> %v, <vscale x 2 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_or_nxv2i64:
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v10, (a0), zero
 ; RV32-NEXT:    vsetvli zero, a2, e64, m2, tu, mu
 ; RV32-NEXT:    vredor.vs v10, v8, v10, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v10
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v10, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_or_nxv2i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v10, a0
 ; RV64-NEXT:    vsetvli zero, a1, e64, m2, tu, mu
 ; RV64-NEXT:    vredor.vs v10, v8, v10, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v10
 ; RV64-NEXT:    ret
   %r = call i64 @llvm.vp.reduce.or.nxv2i64(i64 %s, <vscale x 2 x i64> %v, <vscale x 2 x i1> %m, i32 %evl)
   ret i64 %r
 }
 
 declare i64 @llvm.vp.reduce.xor.nxv2i64(i64, <vscale x 2 x i64>, <vscale x 2 x i1>, i32)
 
 define signext i64 @vpreduce_xor_nxv2i64(i64 signext %s, <vscale x 2 x i64> %v, <vscale x 2 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_xor_nxv2i64:
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v10, (a0), zero
 ; RV32-NEXT:    vsetvli zero, a2, e64, m2, tu, mu
 ; RV32-NEXT:    vredxor.vs v10, v8, v10, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v10
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v10, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_xor_nxv2i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v10, a0
 ; RV64-NEXT:    vsetvli zero, a1, e64, m2, tu, mu
 ; RV64-NEXT:    vredxor.vs v10, v8, v10, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v10
 ; RV64-NEXT:    ret
   %r = call i64 @llvm.vp.reduce.xor.nxv2i64(i64 %s, <vscale x 2 x i64> %v, <vscale x 2 x i1> %m, i32 %evl)
   ret i64 %r
 }
 
 declare i64 @llvm.vp.reduce.add.nxv4i64(i64, <vscale x 4 x i64>, <vscale x 4 x i1>, i32)
 
 define signext i64 @vpreduce_add_nxv4i64(i64 signext %s, <vscale x 4 x i64> %v, <vscale x 4 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_add_nxv4i64:
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v12, (a0), zero
 ; RV32-NEXT:    vsetvli zero, a2, e64, m4, tu, mu
 ; RV32-NEXT:    vredsum.vs v12, v8, v12, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v12
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v12, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_add_nxv4i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v12, a0
 ; RV64-NEXT:    vsetvli zero, a1, e64, m4, tu, mu
 ; RV64-NEXT:    vredsum.vs v12, v8, v12, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v12
 ; RV64-NEXT:    ret
   %r = call i64 @llvm.vp.reduce.add.nxv4i64(i64 %s, <vscale x 4 x i64> %v, <vscale x 4 x i1> %m, i32 %evl)
   ret i64 %r
 }
 
 declare i64 @llvm.vp.reduce.umax.nxv4i64(i64, <vscale x 4 x i64>, <vscale x 4 x i1>, i32)
 
 define signext i64 @vpreduce_umax_nxv4i64(i64 signext %s, <vscale x 4 x i64> %v, <vscale x 4 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_umax_nxv4i64:
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v12, (a0), zero
 ; RV32-NEXT:    vsetvli zero, a2, e64, m4, tu, mu
 ; RV32-NEXT:    vredmaxu.vs v12, v8, v12, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v12
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v12, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_umax_nxv4i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v12, a0
 ; RV64-NEXT:    vsetvli zero, a1, e64, m4, tu, mu
 ; RV64-NEXT:    vredmaxu.vs v12, v8, v12, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v12
 ; RV64-NEXT:    ret
   %r = call i64 @llvm.vp.reduce.umax.nxv4i64(i64 %s, <vscale x 4 x i64> %v, <vscale x 4 x i1> %m, i32 %evl)
   ret i64 %r
 }
 
 declare i64 @llvm.vp.reduce.smax.nxv4i64(i64, <vscale x 4 x i64>, <vscale x 4 x i1>, i32)
 
 define signext i64 @vpreduce_smax_nxv4i64(i64 signext %s, <vscale x 4 x i64> %v, <vscale x 4 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_smax_nxv4i64:
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v12, (a0), zero
 ; RV32-NEXT:    vsetvli zero, a2, e64, m4, tu, mu
 ; RV32-NEXT:    vredmax.vs v12, v8, v12, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v12
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v12, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_smax_nxv4i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v12, a0
 ; RV64-NEXT:    vsetvli zero, a1, e64, m4, tu, mu
 ; RV64-NEXT:    vredmax.vs v12, v8, v12, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v12
 ; RV64-NEXT:    ret
   %r = call i64 @llvm.vp.reduce.smax.nxv4i64(i64 %s, <vscale x 4 x i64> %v, <vscale x 4 x i1> %m, i32 %evl)
   ret i64 %r
 }
 
 declare i64 @llvm.vp.reduce.umin.nxv4i64(i64, <vscale x 4 x i64>, <vscale x 4 x i1>, i32)
 
 define signext i64 @vpreduce_umin_nxv4i64(i64 signext %s, <vscale x 4 x i64> %v, <vscale x 4 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_umin_nxv4i64:
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v12, (a0), zero
 ; RV32-NEXT:    vsetvli zero, a2, e64, m4, tu, mu
 ; RV32-NEXT:    vredminu.vs v12, v8, v12, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v12
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v12, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_umin_nxv4i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v12, a0
 ; RV64-NEXT:    vsetvli zero, a1, e64, m4, tu, mu
 ; RV64-NEXT:    vredminu.vs v12, v8, v12, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v12
 ; RV64-NEXT:    ret
   %r = call i64 @llvm.vp.reduce.umin.nxv4i64(i64 %s, <vscale x 4 x i64> %v, <vscale x 4 x i1> %m, i32 %evl)
   ret i64 %r
 }
 
 declare i64 @llvm.vp.reduce.smin.nxv4i64(i64, <vscale x 4 x i64>, <vscale x 4 x i1>, i32)
 
 define signext i64 @vpreduce_smin_nxv4i64(i64 signext %s, <vscale x 4 x i64> %v, <vscale x 4 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_smin_nxv4i64:
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v12, (a0), zero
 ; RV32-NEXT:    vsetvli zero, a2, e64, m4, tu, mu
 ; RV32-NEXT:    vredmin.vs v12, v8, v12, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v12
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v12, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_smin_nxv4i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v12, a0
 ; RV64-NEXT:    vsetvli zero, a1, e64, m4, tu, mu
 ; RV64-NEXT:    vredmin.vs v12, v8, v12, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v12
 ; RV64-NEXT:    ret
   %r = call i64 @llvm.vp.reduce.smin.nxv4i64(i64 %s, <vscale x 4 x i64> %v, <vscale x 4 x i1> %m, i32 %evl)
   ret i64 %r
 }
 
 declare i64 @llvm.vp.reduce.and.nxv4i64(i64, <vscale x 4 x i64>, <vscale x 4 x i1>, i32)
 
 define signext i64 @vpreduce_and_nxv4i64(i64 signext %s, <vscale x 4 x i64> %v, <vscale x 4 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_and_nxv4i64:
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v12, (a0), zero
 ; RV32-NEXT:    vsetvli zero, a2, e64, m4, tu, mu
 ; RV32-NEXT:    vredand.vs v12, v8, v12, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v12
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v12, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_and_nxv4i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v12, a0
 ; RV64-NEXT:    vsetvli zero, a1, e64, m4, tu, mu
 ; RV64-NEXT:    vredand.vs v12, v8, v12, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v12
 ; RV64-NEXT:    ret
   %r = call i64 @llvm.vp.reduce.and.nxv4i64(i64 %s, <vscale x 4 x i64> %v, <vscale x 4 x i1> %m, i32 %evl)
   ret i64 %r
 }
 
 declare i64 @llvm.vp.reduce.or.nxv4i64(i64, <vscale x 4 x i64>, <vscale x 4 x i1>, i32)
 
 define signext i64 @vpreduce_or_nxv4i64(i64 signext %s, <vscale x 4 x i64> %v, <vscale x 4 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_or_nxv4i64:
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v12, (a0), zero
 ; RV32-NEXT:    vsetvli zero, a2, e64, m4, tu, mu
 ; RV32-NEXT:    vredor.vs v12, v8, v12, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v12
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v12, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_or_nxv4i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v12, a0
 ; RV64-NEXT:    vsetvli zero, a1, e64, m4, tu, mu
 ; RV64-NEXT:    vredor.vs v12, v8, v12, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v12
 ; RV64-NEXT:    ret
   %r = call i64 @llvm.vp.reduce.or.nxv4i64(i64 %s, <vscale x 4 x i64> %v, <vscale x 4 x i1> %m, i32 %evl)
   ret i64 %r
 }
 
 declare i64 @llvm.vp.reduce.xor.nxv4i64(i64, <vscale x 4 x i64>, <vscale x 4 x i1>, i32)
 
 define signext i64 @vpreduce_xor_nxv4i64(i64 signext %s, <vscale x 4 x i64> %v, <vscale x 4 x i1> %m, i32 zeroext %evl) {
 ; RV32-LABEL: vpreduce_xor_nxv4i64:
 ; 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:    vsetvli a0, zero, e64, m1, ta, mu
+; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    addi a0, sp, 8
 ; RV32-NEXT:    vlse64.v v12, (a0), zero
 ; RV32-NEXT:    vsetvli zero, a2, e64, m4, tu, mu
 ; RV32-NEXT:    vredxor.vs v12, v8, v12, v0.t
 ; RV32-NEXT:    vmv.x.s a0, v12
 ; RV32-NEXT:    li a1, 32
 ; RV32-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV32-NEXT:    vsrl.vx v8, v12, a1
 ; RV32-NEXT:    vmv.x.s a1, v8
 ; RV32-NEXT:    addi sp, sp, 16
 ; RV32-NEXT:    ret
 ;
 ; RV64-LABEL: vpreduce_xor_nxv4i64:
 ; RV64:       # %bb.0:
-; RV64-NEXT:    vsetvli a2, zero, e64, m1, ta, mu
+; RV64-NEXT:    vsetivli zero, 1, e64, m1, ta, mu
 ; RV64-NEXT:    vmv.v.x v12, a0
 ; RV64-NEXT:    vsetvli zero, a1, e64, m4, tu, mu
 ; RV64-NEXT:    vredxor.vs v12, v8, v12, v0.t
 ; RV64-NEXT:    vmv.x.s a0, v12
 ; RV64-NEXT:    ret
   %r = call i64 @llvm.vp.reduce.xor.nxv4i64(i64 %s, <vscale x 4 x i64> %v, <vscale x 4 x i1> %m, i32 %evl)
   ret i64 %r
 }