[llvm] Insert vector elt shrink (PR #66087)
Luke Lau via llvm-commits
llvm-commits at lists.llvm.org
Tue Sep 12 06:46:53 PDT 2023
https://github.com/lukel97 created https://github.com/llvm/llvm-project/pull/66087:
Continuing on from #65997, if the index of insert_vector_elt is a constant then we can work out what the minimum number of registers will be needed for the slideup and choose a smaller type to operate on.
This reduces the LMUL for not just the slideup but also for the scalar insert.
>From dd344b6c3d2ab38609e33edf80e3398b0f4063b2 Mon Sep 17 00:00:00 2001
From: Luke Lau <luke at igalia.com>
Date: Mon, 11 Sep 2023 00:44:33 +0100
Subject: [PATCH 1/5] [TwoAddressInstruction] Check if segment was found when
updating subreg LIs
This patch fixes a crash I was seeing when working with subregisters in a
separate upcoming patch. Some of the RVV tests have -early-live-intervals
enabled, which triggers this subregister liveness updating code in
TwoAddressInstruction.
The INSERT_SUBREG in the test case gets converted to a COPY, so it tries to
update the subranges for %1 below:
********** INTERVALS **********
V8 [0B,16r:0) 0 at 0B-phi
%0 [16r,32r:0) 0 at 16r weight:0.000000e+00
%1 [32r,48r:0) 0 at 32r L000000000000003C [32r,48r:0) 0 at 32r L00000000000003C0 [32r,32d:0) 0 at 32r weight:0.000000e+00
%2 EMPTY weight:0.000000e+00
%3 [48r,48d:0) 0 at 48r weight:0.000000e+00
0B bb.0:
liveins: $v8
16B %0:vr = COPY $v8
32B undef %1.sub_vrm1_0:vrm8 = COPY %0:vr
48B dead %3:vrm4 = COPY %1.sub_vrm4_0:vrm8
64B PseudoRET
The subrange for %1 L00000000000003C0 [32r,32d:0) doesn't overlap with
sub_vrm1_0's lane mask, so it tries to merge 0 at 32r. It calls
FindSegmentContaining(32B), 32B being the instruction index for the
INSERT_SUBREG-turned-COPY. But 32B isn't actually in [32r,32d:0) so it returns
end(). The code wasn't checking for end() though so it crashed when
dereferencing DefSeg.
I'm not familiar with this area in the slightest, but I presume we wanted to
find the segment at 32r? And in this test case there's no other segment in the
subrange anyway, so we need to check if DefSeg exists.
I also did a quick check, nothing in llvm-check-codegen seems to trigger this
code. Not sure if that will change if -early-live-intervals is enabled by
default.
---
.../lib/CodeGen/TwoAddressInstructionPass.cpp | 5 +++--
...ressinstruction-subreg-liveness-update.mir | 22 +++++++++++++++++++
2 files changed, 25 insertions(+), 2 deletions(-)
create mode 100644 llvm/test/CodeGen/RISCV/rvv/twoaddressinstruction-subreg-liveness-update.mir
diff --git a/llvm/lib/CodeGen/TwoAddressInstructionPass.cpp b/llvm/lib/CodeGen/TwoAddressInstructionPass.cpp
index 45f61262faf9391..6ceb4b43522f00b 100644
--- a/llvm/lib/CodeGen/TwoAddressInstructionPass.cpp
+++ b/llvm/lib/CodeGen/TwoAddressInstructionPass.cpp
@@ -1868,12 +1868,13 @@ bool TwoAddressInstructionPass::runOnMachineFunction(MachineFunction &Func) {
// %reg.subidx.
LaneBitmask LaneMask =
TRI->getSubRegIndexLaneMask(mi->getOperand(0).getSubReg());
- SlotIndex Idx = LIS->getInstructionIndex(*mi);
+ SlotIndex Idx = LIS->getInstructionIndex(*mi).getRegSlot();
for (auto &S : LI.subranges()) {
if ((S.LaneMask & LaneMask).none()) {
LiveRange::iterator UseSeg = S.FindSegmentContaining(Idx);
LiveRange::iterator DefSeg = std::next(UseSeg);
- S.MergeValueNumberInto(DefSeg->valno, UseSeg->valno);
+ if (DefSeg != S.end())
+ S.MergeValueNumberInto(DefSeg->valno, UseSeg->valno);
}
}
diff --git a/llvm/test/CodeGen/RISCV/rvv/twoaddressinstruction-subreg-liveness-update.mir b/llvm/test/CodeGen/RISCV/rvv/twoaddressinstruction-subreg-liveness-update.mir
new file mode 100644
index 000000000000000..abafc580a968f9e
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/rvv/twoaddressinstruction-subreg-liveness-update.mir
@@ -0,0 +1,22 @@
+# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py UTC_ARGS: --version 3
+# RUN: llc %s -o - -mtriple=riscv64 -mattr=+v -run-pass=liveintervals,twoaddressinstruction,register-coalescer | FileCheck %s
+
+...
+---
+name: insert_v2i64_nxv16i64
+tracksRegLiveness: true
+body: |
+ bb.0:
+ liveins: $v8
+ ; CHECK-LABEL: name: insert_v2i64_nxv16i64
+ ; CHECK: liveins: $v8
+ ; CHECK-NEXT: {{ $}}
+ ; CHECK-NEXT: KILL $v8
+ ; CHECK-NEXT: PseudoRET
+ %0:vr = COPY killed $v8
+ %2:vrm8 = INSERT_SUBREG undef %1:vrm8, killed %0, %subreg.sub_vrm1_0
+ %3:vrm4 = COPY %2.sub_vrm4_0:vrm8
+
+ PseudoRET
+
+...
>From affce4ebc54b1b2678adab170674f0746659a37a Mon Sep 17 00:00:00 2001
From: Luke Lau <luke at igalia.com>
Date: Mon, 11 Sep 2023 14:40:21 +0100
Subject: [PATCH 2/5] Merge segments together if there's > 1, otherwise delete
the sole segment
---
.../lib/CodeGen/TwoAddressInstructionPass.cpp | 28 ++++++++++++++++---
...ressinstruction-subreg-liveness-update.mir | 26 +++++++++++++++--
2 files changed, 47 insertions(+), 7 deletions(-)
diff --git a/llvm/lib/CodeGen/TwoAddressInstructionPass.cpp b/llvm/lib/CodeGen/TwoAddressInstructionPass.cpp
index 6ceb4b43522f00b..3ea22eff91593d7 100644
--- a/llvm/lib/CodeGen/TwoAddressInstructionPass.cpp
+++ b/llvm/lib/CodeGen/TwoAddressInstructionPass.cpp
@@ -1868,15 +1868,35 @@ bool TwoAddressInstructionPass::runOnMachineFunction(MachineFunction &Func) {
// %reg.subidx.
LaneBitmask LaneMask =
TRI->getSubRegIndexLaneMask(mi->getOperand(0).getSubReg());
- SlotIndex Idx = LIS->getInstructionIndex(*mi).getRegSlot();
+ SlotIndex Idx = LIS->getInstructionIndex(*mi);
for (auto &S : LI.subranges()) {
if ((S.LaneMask & LaneMask).none()) {
+ // If Idx is 160B, and we have a subrange that isn't in
+ // %reg.subidx like so:
+ //
+ // [152r,160r)[160r,256r)
+ //
+ // Merge the two segments together so the subrange becomes:
+ //
+ // [152r,256r)
LiveRange::iterator UseSeg = S.FindSegmentContaining(Idx);
- LiveRange::iterator DefSeg = std::next(UseSeg);
- if (DefSeg != S.end())
- S.MergeValueNumberInto(DefSeg->valno, UseSeg->valno);
+ if (UseSeg != S.end()) {
+ LiveRange::iterator DefSeg = std::next(UseSeg);
+ if (DefSeg != S.end())
+ S.MergeValueNumberInto(DefSeg->valno, UseSeg->valno);
+ else
+ S.removeSegment(UseSeg);
+ }
+ // Otherwise, it should have only one segment that starts at
+ // 160r which we should remove.
+ else {
+ assert(S.containsOneValue());
+ assert(S.begin()->start == Idx.getRegSlot());
+ S.removeSegment(S.begin());
+ }
}
}
+ LI.removeEmptySubRanges();
// The COPY no longer has a use of %reg.
LIS->shrinkToUses(&LI);
diff --git a/llvm/test/CodeGen/RISCV/rvv/twoaddressinstruction-subreg-liveness-update.mir b/llvm/test/CodeGen/RISCV/rvv/twoaddressinstruction-subreg-liveness-update.mir
index abafc580a968f9e..03252038f03e2d2 100644
--- a/llvm/test/CodeGen/RISCV/rvv/twoaddressinstruction-subreg-liveness-update.mir
+++ b/llvm/test/CodeGen/RISCV/rvv/twoaddressinstruction-subreg-liveness-update.mir
@@ -1,14 +1,14 @@
# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py UTC_ARGS: --version 3
-# RUN: llc %s -o - -mtriple=riscv64 -mattr=+v -run-pass=liveintervals,twoaddressinstruction,register-coalescer | FileCheck %s
+# RUN: llc %s -o - -mtriple=riscv64 -mattr=+v -run-pass=liveintervals,twoaddressinstruction,register-coalescer -verify-machineinstrs | FileCheck %s
...
---
-name: insert_v2i64_nxv16i64
+name: f
tracksRegLiveness: true
body: |
bb.0:
liveins: $v8
- ; CHECK-LABEL: name: insert_v2i64_nxv16i64
+ ; CHECK-LABEL: name: f
; CHECK: liveins: $v8
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: KILL $v8
@@ -20,3 +20,23 @@ body: |
PseudoRET
...
+---
+name: g
+tracksRegLiveness: true
+body: |
+ bb.0:
+ liveins: $v8, $v16
+ ; CHECK-LABEL: name: g
+ ; CHECK: liveins: $v8, $v16
+ ; CHECK-NEXT: {{ $}}
+ ; CHECK-NEXT: KILL $v8
+ ; CHECK-NEXT: dead [[COPY:%[0-9]+]]:vrm8 = COPY $v16
+ ; CHECK-NEXT: PseudoRET
+ %0:vr = COPY killed $v8
+ %1:vrm8 = COPY killed $v16
+ %2:vrm8 = INSERT_SUBREG %1:vrm8, killed %0, %subreg.sub_vrm1_0
+ %3:vrm4 = COPY %2.sub_vrm4_0:vrm8
+
+ PseudoRET
+
+...
>From d5b783334596be1d3772af15f82494fe13f868bf Mon Sep 17 00:00:00 2001
From: Luke Lau <luke at igalia.com>
Date: Mon, 11 Sep 2023 14:47:00 +0100
Subject: [PATCH 3/5] Change to assert that if we find a segment at the
instruction index, it should always have another segment after it
---
llvm/lib/CodeGen/TwoAddressInstructionPass.cpp | 6 ++----
1 file changed, 2 insertions(+), 4 deletions(-)
diff --git a/llvm/lib/CodeGen/TwoAddressInstructionPass.cpp b/llvm/lib/CodeGen/TwoAddressInstructionPass.cpp
index 3ea22eff91593d7..e36bffc91b91d95 100644
--- a/llvm/lib/CodeGen/TwoAddressInstructionPass.cpp
+++ b/llvm/lib/CodeGen/TwoAddressInstructionPass.cpp
@@ -1882,10 +1882,8 @@ bool TwoAddressInstructionPass::runOnMachineFunction(MachineFunction &Func) {
LiveRange::iterator UseSeg = S.FindSegmentContaining(Idx);
if (UseSeg != S.end()) {
LiveRange::iterator DefSeg = std::next(UseSeg);
- if (DefSeg != S.end())
- S.MergeValueNumberInto(DefSeg->valno, UseSeg->valno);
- else
- S.removeSegment(UseSeg);
+ assert(DefSeg != S.end());
+ S.MergeValueNumberInto(DefSeg->valno, UseSeg->valno);
}
// Otherwise, it should have only one segment that starts at
// 160r which we should remove.
>From dd0dd7a664a97a1f32032d91ea189ca633b55285 Mon Sep 17 00:00:00 2001
From: Luke Lau <luke at igalia.com>
Date: Mon, 11 Sep 2023 19:51:13 +0100
Subject: [PATCH 4/5] [RISCV] Shrink vslideup's LMUL when lowering fixed
insert_subvector
Similar to #65598, if we're using a vslideup to insert a fixed length vector
into another vector, then we can work out the minimum number of registers it
will need to slide up across given the minimum VLEN, and shrink the type
operated on to reduce LMUL accordingly.
This is somewhat dependent on #65916, since it introduces a subregister copy
that triggers a crash with -early-live-intervals in one of the tests.
---
llvm/lib/Target/RISCV/RISCVISelLowering.cpp | 18 +
.../rvv/fixed-vectors-insert-subvector.ll | 45 ++-
.../rvv/fixed-vectors-strided-load-combine.ll | 80 ++---
.../CodeGen/RISCV/rvv/fpclamptosat_vec.ll | 330 ++++++++----------
4 files changed, 229 insertions(+), 244 deletions(-)
diff --git a/llvm/lib/Target/RISCV/RISCVISelLowering.cpp b/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
index 4ff264635cda248..3a81179d1cfc120 100644
--- a/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
+++ b/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
@@ -8606,6 +8606,18 @@ SDValue RISCVTargetLowering::lowerINSERT_SUBVECTOR(SDValue Op,
ContainerVT = getContainerForFixedLengthVector(VecVT);
Vec = convertToScalableVector(ContainerVT, Vec, DAG, Subtarget);
}
+
+ // Shrink down Vec so we're performing the slideup on a smaller LMUL.
+ unsigned LastIdx = OrigIdx + SubVecVT.getVectorNumElements() - 1;
+ MVT OrigContainerVT = ContainerVT;
+ SDValue OrigVec = Vec;
+ if (auto ShrunkVT =
+ getSmallestVTForIndex(ContainerVT, LastIdx, DL, DAG, Subtarget)) {
+ ContainerVT = *ShrunkVT;
+ Vec = DAG.getNode(ISD::EXTRACT_SUBVECTOR, DL, ContainerVT, Vec,
+ DAG.getVectorIdxConstant(0, DL));
+ }
+
SubVec = DAG.getNode(ISD::INSERT_SUBVECTOR, DL, ContainerVT,
DAG.getUNDEF(ContainerVT), SubVec,
DAG.getConstant(0, DL, XLenVT));
@@ -8636,6 +8648,12 @@ SDValue RISCVTargetLowering::lowerINSERT_SUBVECTOR(SDValue Op,
SlideupAmt, Mask, VL, Policy);
}
+ // If we performed the slideup on a smaller LMUL, insert the result back
+ // into the rest of the vector.
+ if (ContainerVT != OrigContainerVT)
+ SubVec = DAG.getNode(ISD::INSERT_SUBVECTOR, DL, OrigContainerVT, OrigVec,
+ SubVec, DAG.getVectorIdxConstant(0, DL));
+
if (VecVT.isFixedLengthVector())
SubVec = convertFromScalableVector(VecVT, SubVec, DAG, Subtarget);
return DAG.getBitcast(Op.getValueType(), SubVec);
diff --git a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-insert-subvector.ll b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-insert-subvector.ll
index 1d6a45ed36f335c..6a9212ed309a8ef 100644
--- a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-insert-subvector.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-insert-subvector.ll
@@ -14,7 +14,7 @@ define <vscale x 8 x i32> @insert_nxv8i32_v2i32_0(<vscale x 8 x i32> %vec, ptr %
; CHECK: # %bb.0:
; CHECK-NEXT: vsetivli zero, 2, e32, mf2, ta, ma
; CHECK-NEXT: vle32.v v12, (a0)
-; CHECK-NEXT: vsetivli zero, 2, e32, m4, tu, ma
+; CHECK-NEXT: vsetivli zero, 2, e32, m1, tu, ma
; CHECK-NEXT: vmv.v.v v8, v12
; CHECK-NEXT: ret
%sv = load <2 x i32>, ptr %svp
@@ -27,7 +27,7 @@ define <vscale x 8 x i32> @insert_nxv8i32_v2i32_2(<vscale x 8 x i32> %vec, ptr %
; CHECK: # %bb.0:
; CHECK-NEXT: vsetivli zero, 2, e32, mf2, ta, ma
; CHECK-NEXT: vle32.v v12, (a0)
-; CHECK-NEXT: vsetivli zero, 4, e32, m4, tu, ma
+; CHECK-NEXT: vsetivli zero, 4, e32, m1, tu, ma
; CHECK-NEXT: vslideup.vi v8, v12, 2
; CHECK-NEXT: ret
%sv = load <2 x i32>, ptr %svp
@@ -40,7 +40,7 @@ define <vscale x 8 x i32> @insert_nxv8i32_v2i32_6(<vscale x 8 x i32> %vec, ptr %
; CHECK: # %bb.0:
; CHECK-NEXT: vsetivli zero, 2, e32, mf2, ta, ma
; CHECK-NEXT: vle32.v v12, (a0)
-; CHECK-NEXT: vsetivli zero, 8, e32, m4, tu, ma
+; CHECK-NEXT: vsetivli zero, 8, e32, m2, tu, ma
; CHECK-NEXT: vslideup.vi v8, v12, 6
; CHECK-NEXT: ret
%sv = load <2 x i32>, ptr %svp
@@ -51,22 +51,19 @@ define <vscale x 8 x i32> @insert_nxv8i32_v2i32_6(<vscale x 8 x i32> %vec, ptr %
define <vscale x 8 x i32> @insert_nxv8i32_v8i32_0(<vscale x 8 x i32> %vec, ptr %svp) {
; LMULMAX2-LABEL: insert_nxv8i32_v8i32_0:
; LMULMAX2: # %bb.0:
-; LMULMAX2-NEXT: vsetivli zero, 8, e32, m2, ta, ma
-; LMULMAX2-NEXT: vle32.v v12, (a0)
-; LMULMAX2-NEXT: vsetivli zero, 8, e32, m4, tu, ma
-; LMULMAX2-NEXT: vmv.v.v v8, v12
+; LMULMAX2-NEXT: vsetivli zero, 8, e32, m2, tu, ma
+; LMULMAX2-NEXT: vle32.v v8, (a0)
; LMULMAX2-NEXT: ret
;
; LMULMAX1-LABEL: insert_nxv8i32_v8i32_0:
; LMULMAX1: # %bb.0:
+; LMULMAX1-NEXT: addi a1, a0, 16
; LMULMAX1-NEXT: vsetivli zero, 4, e32, m1, ta, ma
-; LMULMAX1-NEXT: vle32.v v12, (a0)
-; LMULMAX1-NEXT: addi a0, a0, 16
-; LMULMAX1-NEXT: vle32.v v16, (a0)
-; LMULMAX1-NEXT: vsetivli zero, 4, e32, m4, tu, ma
-; LMULMAX1-NEXT: vmv.v.v v8, v12
-; LMULMAX1-NEXT: vsetivli zero, 8, e32, m4, tu, ma
-; LMULMAX1-NEXT: vslideup.vi v8, v16, 4
+; LMULMAX1-NEXT: vle32.v v12, (a1)
+; LMULMAX1-NEXT: vsetvli zero, zero, e32, m1, tu, ma
+; LMULMAX1-NEXT: vle32.v v8, (a0)
+; LMULMAX1-NEXT: vsetivli zero, 8, e32, m2, tu, ma
+; LMULMAX1-NEXT: vslideup.vi v8, v12, 4
; LMULMAX1-NEXT: ret
%sv = load <8 x i32>, ptr %svp
%v = call <vscale x 8 x i32> @llvm.vector.insert.v8i32.nxv8i32(<vscale x 8 x i32> %vec, <8 x i32> %sv, i64 0)
@@ -84,14 +81,14 @@ define <vscale x 8 x i32> @insert_nxv8i32_v8i32_8(<vscale x 8 x i32> %vec, ptr %
;
; LMULMAX1-LABEL: insert_nxv8i32_v8i32_8:
; LMULMAX1: # %bb.0:
-; LMULMAX1-NEXT: addi a1, a0, 16
; LMULMAX1-NEXT: vsetivli zero, 4, e32, m1, ta, ma
-; LMULMAX1-NEXT: vle32.v v12, (a1)
+; LMULMAX1-NEXT: vle32.v v12, (a0)
+; LMULMAX1-NEXT: addi a0, a0, 16
; LMULMAX1-NEXT: vle32.v v16, (a0)
; LMULMAX1-NEXT: vsetivli zero, 12, e32, m4, tu, ma
-; LMULMAX1-NEXT: vslideup.vi v8, v16, 8
+; LMULMAX1-NEXT: vslideup.vi v8, v12, 8
; LMULMAX1-NEXT: vsetivli zero, 16, e32, m4, tu, ma
-; LMULMAX1-NEXT: vslideup.vi v8, v12, 12
+; LMULMAX1-NEXT: vslideup.vi v8, v16, 12
; LMULMAX1-NEXT: ret
%sv = load <8 x i32>, ptr %svp
%v = call <vscale x 8 x i32> @llvm.vector.insert.v8i32.nxv8i32(<vscale x 8 x i32> %vec, <8 x i32> %sv, i64 8)
@@ -166,7 +163,7 @@ define void @insert_v8i32_v2i32_0(ptr %vp, ptr %svp) {
; LMULMAX2-NEXT: vle32.v v8, (a1)
; LMULMAX2-NEXT: vsetivli zero, 8, e32, m2, ta, ma
; LMULMAX2-NEXT: vle32.v v10, (a0)
-; LMULMAX2-NEXT: vsetivli zero, 2, e32, m2, tu, ma
+; LMULMAX2-NEXT: vsetivli zero, 2, e32, m1, tu, ma
; LMULMAX2-NEXT: vmv.v.v v10, v8
; LMULMAX2-NEXT: vsetivli zero, 8, e32, m2, ta, ma
; LMULMAX2-NEXT: vse32.v v10, (a0)
@@ -197,7 +194,7 @@ define void @insert_v8i32_v2i32_2(ptr %vp, ptr %svp) {
; LMULMAX2-NEXT: vle32.v v8, (a1)
; LMULMAX2-NEXT: vsetivli zero, 8, e32, m2, ta, ma
; LMULMAX2-NEXT: vle32.v v10, (a0)
-; LMULMAX2-NEXT: vsetivli zero, 4, e32, m2, tu, ma
+; LMULMAX2-NEXT: vsetivli zero, 4, e32, m1, tu, ma
; LMULMAX2-NEXT: vslideup.vi v10, v8, 2
; LMULMAX2-NEXT: vsetivli zero, 8, e32, m2, ta, ma
; LMULMAX2-NEXT: vse32.v v10, (a0)
@@ -508,9 +505,9 @@ define void @insert_v2i64_nxv16i64(ptr %psv0, ptr %psv1, <vscale x 16 x i64>* %o
; CHECK: # %bb.0:
; CHECK-NEXT: vsetivli zero, 2, e64, m1, ta, ma
; CHECK-NEXT: vle64.v v8, (a0)
-; CHECK-NEXT: vle64.v v16, (a1)
-; CHECK-NEXT: vsetivli zero, 6, e64, m8, tu, ma
-; CHECK-NEXT: vslideup.vi v8, v16, 4
+; CHECK-NEXT: vle64.v v12, (a1)
+; CHECK-NEXT: vsetivli zero, 6, e64, m4, tu, ma
+; CHECK-NEXT: vslideup.vi v8, v12, 4
; CHECK-NEXT: vs8r.v v8, (a2)
; CHECK-NEXT: ret
%sv0 = load <2 x i64>, ptr %psv0
@@ -539,7 +536,7 @@ define void @insert_v2i64_nxv16i64_lo2(ptr %psv, <vscale x 16 x i64>* %out) {
; CHECK: # %bb.0:
; CHECK-NEXT: vsetivli zero, 2, e64, m1, ta, ma
; CHECK-NEXT: vle64.v v8, (a0)
-; CHECK-NEXT: vsetivli zero, 4, e64, m8, ta, ma
+; CHECK-NEXT: vsetivli zero, 4, e64, m2, ta, ma
; CHECK-NEXT: vslideup.vi v16, v8, 2
; CHECK-NEXT: vs8r.v v16, (a1)
; CHECK-NEXT: ret
diff --git a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-strided-load-combine.ll b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-strided-load-combine.ll
index f52ba6f51d5c897..805557905117add 100644
--- a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-strided-load-combine.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-strided-load-combine.ll
@@ -27,13 +27,13 @@ define void @widen_3xv4i16(ptr %x, ptr %z) {
; CHECK-NEXT: vsetivli zero, 4, e16, mf2, ta, ma
; CHECK-NEXT: vle16.v v8, (a0)
; CHECK-NEXT: addi a2, a0, 8
-; CHECK-NEXT: vle16.v v10, (a2)
+; CHECK-NEXT: vle16.v v9, (a2)
; CHECK-NEXT: addi a0, a0, 16
-; CHECK-NEXT: vle16.v v12, (a0)
-; CHECK-NEXT: vsetivli zero, 8, e16, m2, tu, ma
-; CHECK-NEXT: vslideup.vi v8, v10, 4
+; CHECK-NEXT: vle16.v v10, (a0)
+; CHECK-NEXT: vsetivli zero, 8, e16, m1, tu, ma
+; CHECK-NEXT: vslideup.vi v8, v9, 4
; CHECK-NEXT: vsetivli zero, 12, e16, m2, tu, ma
-; CHECK-NEXT: vslideup.vi v8, v12, 8
+; CHECK-NEXT: vslideup.vi v8, v10, 8
; CHECK-NEXT: vse16.v v8, (a1)
; CHECK-NEXT: ret
%a = load <4 x i16>, ptr %x
@@ -75,17 +75,17 @@ define void @widen_4xv4i16_unaligned(ptr %x, ptr %z) {
; CHECK-NO-MISALIGN-NEXT: vsetivli zero, 8, e8, mf2, ta, ma
; CHECK-NO-MISALIGN-NEXT: vle8.v v8, (a0)
; CHECK-NO-MISALIGN-NEXT: addi a2, a0, 8
-; CHECK-NO-MISALIGN-NEXT: vle8.v v10, (a2)
+; CHECK-NO-MISALIGN-NEXT: vle8.v v9, (a2)
; CHECK-NO-MISALIGN-NEXT: addi a2, a0, 16
-; CHECK-NO-MISALIGN-NEXT: vle8.v v12, (a2)
+; CHECK-NO-MISALIGN-NEXT: vle8.v v10, (a2)
; CHECK-NO-MISALIGN-NEXT: addi a0, a0, 24
-; CHECK-NO-MISALIGN-NEXT: vle8.v v14, (a0)
-; CHECK-NO-MISALIGN-NEXT: vsetivli zero, 8, e16, m2, tu, ma
-; CHECK-NO-MISALIGN-NEXT: vslideup.vi v8, v10, 4
+; CHECK-NO-MISALIGN-NEXT: vle8.v v12, (a0)
+; CHECK-NO-MISALIGN-NEXT: vsetvli zero, zero, e16, m1, tu, ma
+; CHECK-NO-MISALIGN-NEXT: vslideup.vi v8, v9, 4
; CHECK-NO-MISALIGN-NEXT: vsetivli zero, 12, e16, m2, tu, ma
-; CHECK-NO-MISALIGN-NEXT: vslideup.vi v8, v12, 8
+; CHECK-NO-MISALIGN-NEXT: vslideup.vi v8, v10, 8
; CHECK-NO-MISALIGN-NEXT: vsetivli zero, 16, e16, m2, ta, ma
-; CHECK-NO-MISALIGN-NEXT: vslideup.vi v8, v14, 12
+; CHECK-NO-MISALIGN-NEXT: vslideup.vi v8, v12, 12
; CHECK-NO-MISALIGN-NEXT: vse16.v v8, (a1)
; CHECK-NO-MISALIGN-NEXT: ret
;
@@ -188,17 +188,17 @@ define void @strided_constant_mismatch_4xv4i16(ptr %x, ptr %z) {
; CHECK-NEXT: vsetivli zero, 4, e16, mf2, ta, ma
; CHECK-NEXT: vle16.v v8, (a0)
; CHECK-NEXT: addi a2, a0, 2
-; CHECK-NEXT: vle16.v v10, (a2)
+; CHECK-NEXT: vle16.v v9, (a2)
; CHECK-NEXT: addi a2, a0, 6
-; CHECK-NEXT: vle16.v v12, (a2)
+; CHECK-NEXT: vle16.v v10, (a2)
; CHECK-NEXT: addi a0, a0, 8
-; CHECK-NEXT: vle16.v v14, (a0)
-; CHECK-NEXT: vsetivli zero, 8, e16, m2, tu, ma
-; CHECK-NEXT: vslideup.vi v8, v10, 4
+; CHECK-NEXT: vle16.v v12, (a0)
+; CHECK-NEXT: vsetivli zero, 8, e16, m1, tu, ma
+; CHECK-NEXT: vslideup.vi v8, v9, 4
; CHECK-NEXT: vsetivli zero, 12, e16, m2, tu, ma
-; CHECK-NEXT: vslideup.vi v8, v12, 8
+; CHECK-NEXT: vslideup.vi v8, v10, 8
; CHECK-NEXT: vsetivli zero, 16, e16, m2, ta, ma
-; CHECK-NEXT: vslideup.vi v8, v14, 12
+; CHECK-NEXT: vslideup.vi v8, v12, 12
; CHECK-NEXT: vse16.v v8, (a1)
; CHECK-NEXT: ret
%a = load <4 x i16>, ptr %x
@@ -258,17 +258,17 @@ define void @strided_runtime_mismatch_4xv4i16(ptr %x, ptr %z, i64 %s, i64 %t) {
; RV32-NEXT: vsetivli zero, 4, e16, mf2, ta, ma
; RV32-NEXT: vle16.v v8, (a0)
; RV32-NEXT: add a0, a0, a2
-; RV32-NEXT: vle16.v v10, (a0)
+; RV32-NEXT: vle16.v v9, (a0)
; RV32-NEXT: add a0, a0, a4
-; RV32-NEXT: vle16.v v12, (a0)
+; RV32-NEXT: vle16.v v10, (a0)
; RV32-NEXT: add a0, a0, a2
-; RV32-NEXT: vle16.v v14, (a0)
-; RV32-NEXT: vsetivli zero, 8, e16, m2, tu, ma
-; RV32-NEXT: vslideup.vi v8, v10, 4
+; RV32-NEXT: vle16.v v12, (a0)
+; RV32-NEXT: vsetivli zero, 8, e16, m1, tu, ma
+; RV32-NEXT: vslideup.vi v8, v9, 4
; RV32-NEXT: vsetivli zero, 12, e16, m2, tu, ma
-; RV32-NEXT: vslideup.vi v8, v12, 8
+; RV32-NEXT: vslideup.vi v8, v10, 8
; RV32-NEXT: vsetivli zero, 16, e16, m2, ta, ma
-; RV32-NEXT: vslideup.vi v8, v14, 12
+; RV32-NEXT: vslideup.vi v8, v12, 12
; RV32-NEXT: vse16.v v8, (a1)
; RV32-NEXT: ret
;
@@ -277,17 +277,17 @@ define void @strided_runtime_mismatch_4xv4i16(ptr %x, ptr %z, i64 %s, i64 %t) {
; RV64-NEXT: vsetivli zero, 4, e16, mf2, ta, ma
; RV64-NEXT: vle16.v v8, (a0)
; RV64-NEXT: add a0, a0, a2
-; RV64-NEXT: vle16.v v10, (a0)
+; RV64-NEXT: vle16.v v9, (a0)
; RV64-NEXT: add a0, a0, a3
-; RV64-NEXT: vle16.v v12, (a0)
+; RV64-NEXT: vle16.v v10, (a0)
; RV64-NEXT: add a0, a0, a2
-; RV64-NEXT: vle16.v v14, (a0)
-; RV64-NEXT: vsetivli zero, 8, e16, m2, tu, ma
-; RV64-NEXT: vslideup.vi v8, v10, 4
+; RV64-NEXT: vle16.v v12, (a0)
+; RV64-NEXT: vsetivli zero, 8, e16, m1, tu, ma
+; RV64-NEXT: vslideup.vi v8, v9, 4
; RV64-NEXT: vsetivli zero, 12, e16, m2, tu, ma
-; RV64-NEXT: vslideup.vi v8, v12, 8
+; RV64-NEXT: vslideup.vi v8, v10, 8
; RV64-NEXT: vsetivli zero, 16, e16, m2, ta, ma
-; RV64-NEXT: vslideup.vi v8, v14, 12
+; RV64-NEXT: vslideup.vi v8, v12, 12
; RV64-NEXT: vse16.v v8, (a1)
; RV64-NEXT: ret
;
@@ -296,17 +296,17 @@ define void @strided_runtime_mismatch_4xv4i16(ptr %x, ptr %z, i64 %s, i64 %t) {
; ZVE64F-NEXT: vsetivli zero, 4, e16, mf2, ta, ma
; ZVE64F-NEXT: vle16.v v8, (a0)
; ZVE64F-NEXT: add a0, a0, a2
-; ZVE64F-NEXT: vle16.v v10, (a0)
+; ZVE64F-NEXT: vle16.v v9, (a0)
; ZVE64F-NEXT: add a0, a0, a3
-; ZVE64F-NEXT: vle16.v v12, (a0)
+; ZVE64F-NEXT: vle16.v v10, (a0)
; ZVE64F-NEXT: add a0, a0, a2
-; ZVE64F-NEXT: vle16.v v14, (a0)
-; ZVE64F-NEXT: vsetivli zero, 8, e16, m2, tu, ma
-; ZVE64F-NEXT: vslideup.vi v8, v10, 4
+; ZVE64F-NEXT: vle16.v v12, (a0)
+; ZVE64F-NEXT: vsetivli zero, 8, e16, m1, tu, ma
+; ZVE64F-NEXT: vslideup.vi v8, v9, 4
; ZVE64F-NEXT: vsetivli zero, 12, e16, m2, tu, ma
-; ZVE64F-NEXT: vslideup.vi v8, v12, 8
+; ZVE64F-NEXT: vslideup.vi v8, v10, 8
; ZVE64F-NEXT: vsetivli zero, 16, e16, m2, ta, ma
-; ZVE64F-NEXT: vslideup.vi v8, v14, 12
+; ZVE64F-NEXT: vslideup.vi v8, v12, 12
; ZVE64F-NEXT: vse16.v v8, (a1)
; ZVE64F-NEXT: ret
%a = load <4 x i16>, ptr %x
diff --git a/llvm/test/CodeGen/RISCV/rvv/fpclamptosat_vec.ll b/llvm/test/CodeGen/RISCV/rvv/fpclamptosat_vec.ll
index 31e7e7be76c89b1..f598118c18aff9f 100644
--- a/llvm/test/CodeGen/RISCV/rvv/fpclamptosat_vec.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/fpclamptosat_vec.ll
@@ -460,54 +460,49 @@ define <4 x i32> @stest_f16i32(<4 x half> %x) {
; CHECK-V-NEXT: sub sp, sp, a1
; CHECK-V-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x30, 0x22, 0x11, 0x04, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 48 + 4 * vlenb
; CHECK-V-NEXT: lhu s0, 24(a0)
-; CHECK-V-NEXT: lhu s1, 16(a0)
-; CHECK-V-NEXT: lhu s2, 0(a0)
-; CHECK-V-NEXT: lhu a0, 8(a0)
+; CHECK-V-NEXT: lhu s1, 0(a0)
+; CHECK-V-NEXT: lhu s2, 8(a0)
+; CHECK-V-NEXT: lhu a0, 16(a0)
; CHECK-V-NEXT: fmv.w.x fa0, a0
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.l.s a0, fa0, rtz
; CHECK-V-NEXT: vsetivli zero, 1, e64, m1, ta, ma
; CHECK-V-NEXT: vmv.s.x v8, a0
-; CHECK-V-NEXT: addi a0, sp, 16
-; CHECK-V-NEXT: vs2r.v v8, (a0) # Unknown-size Folded Spill
-; CHECK-V-NEXT: fmv.w.x fa0, s2
-; CHECK-V-NEXT: call __extendhfsf2 at plt
-; CHECK-V-NEXT: fcvt.l.s a0, fa0, rtz
-; CHECK-V-NEXT: vsetivli zero, 2, e64, m2, tu, ma
-; CHECK-V-NEXT: vmv.s.x v8, a0
-; CHECK-V-NEXT: addi a0, sp, 16
-; CHECK-V-NEXT: vl2r.v v10, (a0) # Unknown-size Folded Reload
-; CHECK-V-NEXT: vslideup.vi v8, v10, 1
; CHECK-V-NEXT: csrr a0, vlenb
; CHECK-V-NEXT: slli a0, a0, 1
; CHECK-V-NEXT: add a0, sp, a0
; CHECK-V-NEXT: addi a0, a0, 16
; CHECK-V-NEXT: vs2r.v v8, (a0) # Unknown-size Folded Spill
+; CHECK-V-NEXT: fmv.w.x fa0, s2
+; CHECK-V-NEXT: call __extendhfsf2 at plt
+; CHECK-V-NEXT: fcvt.l.s a0, fa0, rtz
+; CHECK-V-NEXT: vsetivli zero, 1, e64, m1, ta, ma
; CHECK-V-NEXT: fmv.w.x fa0, s1
+; CHECK-V-NEXT: vmv.s.x v8, a0
+; CHECK-V-NEXT: addi a0, sp, 16
+; CHECK-V-NEXT: vs1r.v v8, (a0) # Unknown-size Folded Spill
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.l.s a0, fa0, rtz
+; CHECK-V-NEXT: vsetivli zero, 2, e64, m1, tu, ma
+; CHECK-V-NEXT: vmv.s.x v10, a0
+; CHECK-V-NEXT: addi a0, sp, 16
+; CHECK-V-NEXT: vl1r.v v8, (a0) # Unknown-size Folded Reload
+; CHECK-V-NEXT: vslideup.vi v10, v8, 1
; CHECK-V-NEXT: vsetivli zero, 3, e64, m2, tu, ma
-; CHECK-V-NEXT: vmv.s.x v8, a0
; CHECK-V-NEXT: csrr a0, vlenb
; CHECK-V-NEXT: slli a0, a0, 1
; CHECK-V-NEXT: add a0, sp, a0
; CHECK-V-NEXT: addi a0, a0, 16
-; CHECK-V-NEXT: vl2r.v v10, (a0) # Unknown-size Folded Reload
+; CHECK-V-NEXT: vl2r.v v8, (a0) # Unknown-size Folded Reload
; CHECK-V-NEXT: vslideup.vi v10, v8, 2
-; CHECK-V-NEXT: csrr a0, vlenb
-; CHECK-V-NEXT: slli a0, a0, 1
-; CHECK-V-NEXT: add a0, sp, a0
-; CHECK-V-NEXT: addi a0, a0, 16
+; CHECK-V-NEXT: addi a0, sp, 16
; CHECK-V-NEXT: vs2r.v v10, (a0) # Unknown-size Folded Spill
; CHECK-V-NEXT: fmv.w.x fa0, s0
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.l.s a0, fa0, rtz
; CHECK-V-NEXT: vsetivli zero, 4, e64, m2, ta, ma
; CHECK-V-NEXT: vmv.s.x v8, a0
-; CHECK-V-NEXT: csrr a0, vlenb
-; CHECK-V-NEXT: slli a0, a0, 1
-; CHECK-V-NEXT: add a0, sp, a0
-; CHECK-V-NEXT: addi a0, a0, 16
+; CHECK-V-NEXT: addi a0, sp, 16
; CHECK-V-NEXT: vl2r.v v10, (a0) # Unknown-size Folded Reload
; CHECK-V-NEXT: vslideup.vi v10, v8, 3
; CHECK-V-NEXT: lui a0, 524288
@@ -632,54 +627,49 @@ define <4 x i32> @utesth_f16i32(<4 x half> %x) {
; CHECK-V-NEXT: sub sp, sp, a1
; CHECK-V-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x30, 0x22, 0x11, 0x04, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 48 + 4 * vlenb
; CHECK-V-NEXT: lhu s0, 24(a0)
-; CHECK-V-NEXT: lhu s1, 16(a0)
-; CHECK-V-NEXT: lhu s2, 0(a0)
-; CHECK-V-NEXT: lhu a0, 8(a0)
+; CHECK-V-NEXT: lhu s1, 0(a0)
+; CHECK-V-NEXT: lhu s2, 8(a0)
+; CHECK-V-NEXT: lhu a0, 16(a0)
; CHECK-V-NEXT: fmv.w.x fa0, a0
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.lu.s a0, fa0, rtz
; CHECK-V-NEXT: vsetivli zero, 1, e64, m1, ta, ma
; CHECK-V-NEXT: vmv.s.x v8, a0
-; CHECK-V-NEXT: addi a0, sp, 16
-; CHECK-V-NEXT: vs2r.v v8, (a0) # Unknown-size Folded Spill
-; CHECK-V-NEXT: fmv.w.x fa0, s2
-; CHECK-V-NEXT: call __extendhfsf2 at plt
-; CHECK-V-NEXT: fcvt.lu.s a0, fa0, rtz
-; CHECK-V-NEXT: vsetivli zero, 2, e64, m2, tu, ma
-; CHECK-V-NEXT: vmv.s.x v8, a0
-; CHECK-V-NEXT: addi a0, sp, 16
-; CHECK-V-NEXT: vl2r.v v10, (a0) # Unknown-size Folded Reload
-; CHECK-V-NEXT: vslideup.vi v8, v10, 1
; CHECK-V-NEXT: csrr a0, vlenb
; CHECK-V-NEXT: slli a0, a0, 1
; CHECK-V-NEXT: add a0, sp, a0
; CHECK-V-NEXT: addi a0, a0, 16
; CHECK-V-NEXT: vs2r.v v8, (a0) # Unknown-size Folded Spill
+; CHECK-V-NEXT: fmv.w.x fa0, s2
+; CHECK-V-NEXT: call __extendhfsf2 at plt
+; CHECK-V-NEXT: fcvt.lu.s a0, fa0, rtz
+; CHECK-V-NEXT: vsetivli zero, 1, e64, m1, ta, ma
; CHECK-V-NEXT: fmv.w.x fa0, s1
+; CHECK-V-NEXT: vmv.s.x v8, a0
+; CHECK-V-NEXT: addi a0, sp, 16
+; CHECK-V-NEXT: vs1r.v v8, (a0) # Unknown-size Folded Spill
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.lu.s a0, fa0, rtz
+; CHECK-V-NEXT: vsetivli zero, 2, e64, m1, tu, ma
+; CHECK-V-NEXT: vmv.s.x v10, a0
+; CHECK-V-NEXT: addi a0, sp, 16
+; CHECK-V-NEXT: vl1r.v v8, (a0) # Unknown-size Folded Reload
+; CHECK-V-NEXT: vslideup.vi v10, v8, 1
; CHECK-V-NEXT: vsetivli zero, 3, e64, m2, tu, ma
-; CHECK-V-NEXT: vmv.s.x v8, a0
; CHECK-V-NEXT: csrr a0, vlenb
; CHECK-V-NEXT: slli a0, a0, 1
; CHECK-V-NEXT: add a0, sp, a0
; CHECK-V-NEXT: addi a0, a0, 16
-; CHECK-V-NEXT: vl2r.v v10, (a0) # Unknown-size Folded Reload
+; CHECK-V-NEXT: vl2r.v v8, (a0) # Unknown-size Folded Reload
; CHECK-V-NEXT: vslideup.vi v10, v8, 2
-; CHECK-V-NEXT: csrr a0, vlenb
-; CHECK-V-NEXT: slli a0, a0, 1
-; CHECK-V-NEXT: add a0, sp, a0
-; CHECK-V-NEXT: addi a0, a0, 16
+; CHECK-V-NEXT: addi a0, sp, 16
; CHECK-V-NEXT: vs2r.v v10, (a0) # Unknown-size Folded Spill
; CHECK-V-NEXT: fmv.w.x fa0, s0
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.lu.s a0, fa0, rtz
; CHECK-V-NEXT: vsetivli zero, 4, e64, m2, ta, ma
; CHECK-V-NEXT: vmv.s.x v8, a0
-; CHECK-V-NEXT: csrr a0, vlenb
-; CHECK-V-NEXT: slli a0, a0, 1
-; CHECK-V-NEXT: add a0, sp, a0
-; CHECK-V-NEXT: addi a0, a0, 16
+; CHECK-V-NEXT: addi a0, sp, 16
; CHECK-V-NEXT: vl2r.v v10, (a0) # Unknown-size Folded Reload
; CHECK-V-NEXT: vslideup.vi v10, v8, 3
; CHECK-V-NEXT: li a0, -1
@@ -813,54 +803,49 @@ define <4 x i32> @ustest_f16i32(<4 x half> %x) {
; CHECK-V-NEXT: sub sp, sp, a1
; CHECK-V-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x30, 0x22, 0x11, 0x04, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 48 + 4 * vlenb
; CHECK-V-NEXT: lhu s0, 24(a0)
-; CHECK-V-NEXT: lhu s1, 16(a0)
-; CHECK-V-NEXT: lhu s2, 0(a0)
-; CHECK-V-NEXT: lhu a0, 8(a0)
+; CHECK-V-NEXT: lhu s1, 0(a0)
+; CHECK-V-NEXT: lhu s2, 8(a0)
+; CHECK-V-NEXT: lhu a0, 16(a0)
; CHECK-V-NEXT: fmv.w.x fa0, a0
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.l.s a0, fa0, rtz
; CHECK-V-NEXT: vsetivli zero, 1, e64, m1, ta, ma
; CHECK-V-NEXT: vmv.s.x v8, a0
-; CHECK-V-NEXT: addi a0, sp, 16
-; CHECK-V-NEXT: vs2r.v v8, (a0) # Unknown-size Folded Spill
-; CHECK-V-NEXT: fmv.w.x fa0, s2
-; CHECK-V-NEXT: call __extendhfsf2 at plt
-; CHECK-V-NEXT: fcvt.l.s a0, fa0, rtz
-; CHECK-V-NEXT: vsetivli zero, 2, e64, m2, tu, ma
-; CHECK-V-NEXT: vmv.s.x v8, a0
-; CHECK-V-NEXT: addi a0, sp, 16
-; CHECK-V-NEXT: vl2r.v v10, (a0) # Unknown-size Folded Reload
-; CHECK-V-NEXT: vslideup.vi v8, v10, 1
; CHECK-V-NEXT: csrr a0, vlenb
; CHECK-V-NEXT: slli a0, a0, 1
; CHECK-V-NEXT: add a0, sp, a0
; CHECK-V-NEXT: addi a0, a0, 16
; CHECK-V-NEXT: vs2r.v v8, (a0) # Unknown-size Folded Spill
+; CHECK-V-NEXT: fmv.w.x fa0, s2
+; CHECK-V-NEXT: call __extendhfsf2 at plt
+; CHECK-V-NEXT: fcvt.l.s a0, fa0, rtz
+; CHECK-V-NEXT: vsetivli zero, 1, e64, m1, ta, ma
; CHECK-V-NEXT: fmv.w.x fa0, s1
+; CHECK-V-NEXT: vmv.s.x v8, a0
+; CHECK-V-NEXT: addi a0, sp, 16
+; CHECK-V-NEXT: vs1r.v v8, (a0) # Unknown-size Folded Spill
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.l.s a0, fa0, rtz
+; CHECK-V-NEXT: vsetivli zero, 2, e64, m1, tu, ma
+; CHECK-V-NEXT: vmv.s.x v10, a0
+; CHECK-V-NEXT: addi a0, sp, 16
+; CHECK-V-NEXT: vl1r.v v8, (a0) # Unknown-size Folded Reload
+; CHECK-V-NEXT: vslideup.vi v10, v8, 1
; CHECK-V-NEXT: vsetivli zero, 3, e64, m2, tu, ma
-; CHECK-V-NEXT: vmv.s.x v8, a0
; CHECK-V-NEXT: csrr a0, vlenb
; CHECK-V-NEXT: slli a0, a0, 1
; CHECK-V-NEXT: add a0, sp, a0
; CHECK-V-NEXT: addi a0, a0, 16
-; CHECK-V-NEXT: vl2r.v v10, (a0) # Unknown-size Folded Reload
+; CHECK-V-NEXT: vl2r.v v8, (a0) # Unknown-size Folded Reload
; CHECK-V-NEXT: vslideup.vi v10, v8, 2
-; CHECK-V-NEXT: csrr a0, vlenb
-; CHECK-V-NEXT: slli a0, a0, 1
-; CHECK-V-NEXT: add a0, sp, a0
-; CHECK-V-NEXT: addi a0, a0, 16
+; CHECK-V-NEXT: addi a0, sp, 16
; CHECK-V-NEXT: vs2r.v v10, (a0) # Unknown-size Folded Spill
; CHECK-V-NEXT: fmv.w.x fa0, s0
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.l.s a0, fa0, rtz
; CHECK-V-NEXT: vsetivli zero, 4, e64, m2, ta, ma
; CHECK-V-NEXT: vmv.s.x v8, a0
-; CHECK-V-NEXT: csrr a0, vlenb
-; CHECK-V-NEXT: slli a0, a0, 1
-; CHECK-V-NEXT: add a0, sp, a0
-; CHECK-V-NEXT: addi a0, a0, 16
+; CHECK-V-NEXT: addi a0, sp, 16
; CHECK-V-NEXT: vl2r.v v10, (a0) # Unknown-size Folded Reload
; CHECK-V-NEXT: vslideup.vi v10, v8, 3
; CHECK-V-NEXT: li a0, -1
@@ -1454,8 +1439,8 @@ define <8 x i16> @stest_f16i16(<8 x half> %x) {
; CHECK-V-NEXT: lhu s3, 32(a0)
; CHECK-V-NEXT: lhu s4, 24(a0)
; CHECK-V-NEXT: lhu s5, 16(a0)
-; CHECK-V-NEXT: lhu s6, 0(a0)
-; CHECK-V-NEXT: lhu a0, 8(a0)
+; CHECK-V-NEXT: lhu s6, 8(a0)
+; CHECK-V-NEXT: lhu a0, 0(a0)
; CHECK-V-NEXT: fmv.w.x fa0, a0
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.l.s a0, fa0, rtz
@@ -1466,16 +1451,16 @@ define <8 x i16> @stest_f16i16(<8 x half> %x) {
; CHECK-V-NEXT: fmv.w.x fa0, s6
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.l.s a0, fa0, rtz
-; CHECK-V-NEXT: vsetivli zero, 2, e32, m2, tu, ma
+; CHECK-V-NEXT: vsetivli zero, 2, e32, m1, tu, ma
; CHECK-V-NEXT: vmv.s.x v8, a0
; CHECK-V-NEXT: addi a0, sp, 16
; CHECK-V-NEXT: vl2r.v v10, (a0) # Unknown-size Folded Reload
-; CHECK-V-NEXT: vslideup.vi v8, v10, 1
-; CHECK-V-NEXT: vs2r.v v8, (a0) # Unknown-size Folded Spill
+; CHECK-V-NEXT: vslideup.vi v10, v8, 1
+; CHECK-V-NEXT: vs2r.v v10, (a0) # Unknown-size Folded Spill
; CHECK-V-NEXT: fmv.w.x fa0, s5
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.l.s a0, fa0, rtz
-; CHECK-V-NEXT: vsetivli zero, 3, e32, m2, tu, ma
+; CHECK-V-NEXT: vsetivli zero, 3, e32, m1, tu, ma
; CHECK-V-NEXT: vmv.s.x v8, a0
; CHECK-V-NEXT: addi a0, sp, 16
; CHECK-V-NEXT: vl2r.v v10, (a0) # Unknown-size Folded Reload
@@ -1484,7 +1469,7 @@ define <8 x i16> @stest_f16i16(<8 x half> %x) {
; CHECK-V-NEXT: fmv.w.x fa0, s4
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.l.s a0, fa0, rtz
-; CHECK-V-NEXT: vsetivli zero, 4, e32, m2, tu, ma
+; CHECK-V-NEXT: vsetivli zero, 4, e32, m1, tu, ma
; CHECK-V-NEXT: vmv.s.x v8, a0
; CHECK-V-NEXT: addi a0, sp, 16
; CHECK-V-NEXT: vl2r.v v10, (a0) # Unknown-size Folded Reload
@@ -1737,8 +1722,8 @@ define <8 x i16> @utesth_f16i16(<8 x half> %x) {
; CHECK-V-NEXT: lhu s3, 32(a0)
; CHECK-V-NEXT: lhu s4, 24(a0)
; CHECK-V-NEXT: lhu s5, 16(a0)
-; CHECK-V-NEXT: lhu s6, 0(a0)
-; CHECK-V-NEXT: lhu a0, 8(a0)
+; CHECK-V-NEXT: lhu s6, 8(a0)
+; CHECK-V-NEXT: lhu a0, 0(a0)
; CHECK-V-NEXT: fmv.w.x fa0, a0
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.lu.s a0, fa0, rtz
@@ -1749,16 +1734,16 @@ define <8 x i16> @utesth_f16i16(<8 x half> %x) {
; CHECK-V-NEXT: fmv.w.x fa0, s6
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.lu.s a0, fa0, rtz
-; CHECK-V-NEXT: vsetivli zero, 2, e32, m2, tu, ma
+; CHECK-V-NEXT: vsetivli zero, 2, e32, m1, tu, ma
; CHECK-V-NEXT: vmv.s.x v8, a0
; CHECK-V-NEXT: addi a0, sp, 16
; CHECK-V-NEXT: vl2r.v v10, (a0) # Unknown-size Folded Reload
-; CHECK-V-NEXT: vslideup.vi v8, v10, 1
-; CHECK-V-NEXT: vs2r.v v8, (a0) # Unknown-size Folded Spill
+; CHECK-V-NEXT: vslideup.vi v10, v8, 1
+; CHECK-V-NEXT: vs2r.v v10, (a0) # Unknown-size Folded Spill
; CHECK-V-NEXT: fmv.w.x fa0, s5
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.lu.s a0, fa0, rtz
-; CHECK-V-NEXT: vsetivli zero, 3, e32, m2, tu, ma
+; CHECK-V-NEXT: vsetivli zero, 3, e32, m1, tu, ma
; CHECK-V-NEXT: vmv.s.x v8, a0
; CHECK-V-NEXT: addi a0, sp, 16
; CHECK-V-NEXT: vl2r.v v10, (a0) # Unknown-size Folded Reload
@@ -1767,7 +1752,7 @@ define <8 x i16> @utesth_f16i16(<8 x half> %x) {
; CHECK-V-NEXT: fmv.w.x fa0, s4
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.lu.s a0, fa0, rtz
-; CHECK-V-NEXT: vsetivli zero, 4, e32, m2, tu, ma
+; CHECK-V-NEXT: vsetivli zero, 4, e32, m1, tu, ma
; CHECK-V-NEXT: vmv.s.x v8, a0
; CHECK-V-NEXT: addi a0, sp, 16
; CHECK-V-NEXT: vl2r.v v10, (a0) # Unknown-size Folded Reload
@@ -2040,8 +2025,8 @@ define <8 x i16> @ustest_f16i16(<8 x half> %x) {
; CHECK-V-NEXT: lhu s3, 32(a0)
; CHECK-V-NEXT: lhu s4, 24(a0)
; CHECK-V-NEXT: lhu s5, 16(a0)
-; CHECK-V-NEXT: lhu s6, 0(a0)
-; CHECK-V-NEXT: lhu a0, 8(a0)
+; CHECK-V-NEXT: lhu s6, 8(a0)
+; CHECK-V-NEXT: lhu a0, 0(a0)
; CHECK-V-NEXT: fmv.w.x fa0, a0
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.l.s a0, fa0, rtz
@@ -2052,16 +2037,16 @@ define <8 x i16> @ustest_f16i16(<8 x half> %x) {
; CHECK-V-NEXT: fmv.w.x fa0, s6
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.l.s a0, fa0, rtz
-; CHECK-V-NEXT: vsetivli zero, 2, e32, m2, tu, ma
+; CHECK-V-NEXT: vsetivli zero, 2, e32, m1, tu, ma
; CHECK-V-NEXT: vmv.s.x v8, a0
; CHECK-V-NEXT: addi a0, sp, 16
; CHECK-V-NEXT: vl2r.v v10, (a0) # Unknown-size Folded Reload
-; CHECK-V-NEXT: vslideup.vi v8, v10, 1
-; CHECK-V-NEXT: vs2r.v v8, (a0) # Unknown-size Folded Spill
+; CHECK-V-NEXT: vslideup.vi v10, v8, 1
+; CHECK-V-NEXT: vs2r.v v10, (a0) # Unknown-size Folded Spill
; CHECK-V-NEXT: fmv.w.x fa0, s5
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.l.s a0, fa0, rtz
-; CHECK-V-NEXT: vsetivli zero, 3, e32, m2, tu, ma
+; CHECK-V-NEXT: vsetivli zero, 3, e32, m1, tu, ma
; CHECK-V-NEXT: vmv.s.x v8, a0
; CHECK-V-NEXT: addi a0, sp, 16
; CHECK-V-NEXT: vl2r.v v10, (a0) # Unknown-size Folded Reload
@@ -2070,7 +2055,7 @@ define <8 x i16> @ustest_f16i16(<8 x half> %x) {
; CHECK-V-NEXT: fmv.w.x fa0, s4
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.l.s a0, fa0, rtz
-; CHECK-V-NEXT: vsetivli zero, 4, e32, m2, tu, ma
+; CHECK-V-NEXT: vsetivli zero, 4, e32, m1, tu, ma
; CHECK-V-NEXT: vmv.s.x v8, a0
; CHECK-V-NEXT: addi a0, sp, 16
; CHECK-V-NEXT: vl2r.v v10, (a0) # Unknown-size Folded Reload
@@ -3796,54 +3781,49 @@ define <4 x i32> @stest_f16i32_mm(<4 x half> %x) {
; CHECK-V-NEXT: sub sp, sp, a1
; CHECK-V-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x30, 0x22, 0x11, 0x04, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 48 + 4 * vlenb
; CHECK-V-NEXT: lhu s0, 24(a0)
-; CHECK-V-NEXT: lhu s1, 16(a0)
-; CHECK-V-NEXT: lhu s2, 0(a0)
-; CHECK-V-NEXT: lhu a0, 8(a0)
+; CHECK-V-NEXT: lhu s1, 0(a0)
+; CHECK-V-NEXT: lhu s2, 8(a0)
+; CHECK-V-NEXT: lhu a0, 16(a0)
; CHECK-V-NEXT: fmv.w.x fa0, a0
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.l.s a0, fa0, rtz
; CHECK-V-NEXT: vsetivli zero, 1, e64, m1, ta, ma
; CHECK-V-NEXT: vmv.s.x v8, a0
-; CHECK-V-NEXT: addi a0, sp, 16
-; CHECK-V-NEXT: vs2r.v v8, (a0) # Unknown-size Folded Spill
-; CHECK-V-NEXT: fmv.w.x fa0, s2
-; CHECK-V-NEXT: call __extendhfsf2 at plt
-; CHECK-V-NEXT: fcvt.l.s a0, fa0, rtz
-; CHECK-V-NEXT: vsetivli zero, 2, e64, m2, tu, ma
-; CHECK-V-NEXT: vmv.s.x v8, a0
-; CHECK-V-NEXT: addi a0, sp, 16
-; CHECK-V-NEXT: vl2r.v v10, (a0) # Unknown-size Folded Reload
-; CHECK-V-NEXT: vslideup.vi v8, v10, 1
; CHECK-V-NEXT: csrr a0, vlenb
; CHECK-V-NEXT: slli a0, a0, 1
; CHECK-V-NEXT: add a0, sp, a0
; CHECK-V-NEXT: addi a0, a0, 16
; CHECK-V-NEXT: vs2r.v v8, (a0) # Unknown-size Folded Spill
+; CHECK-V-NEXT: fmv.w.x fa0, s2
+; CHECK-V-NEXT: call __extendhfsf2 at plt
+; CHECK-V-NEXT: fcvt.l.s a0, fa0, rtz
+; CHECK-V-NEXT: vsetivli zero, 1, e64, m1, ta, ma
; CHECK-V-NEXT: fmv.w.x fa0, s1
+; CHECK-V-NEXT: vmv.s.x v8, a0
+; CHECK-V-NEXT: addi a0, sp, 16
+; CHECK-V-NEXT: vs1r.v v8, (a0) # Unknown-size Folded Spill
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.l.s a0, fa0, rtz
+; CHECK-V-NEXT: vsetivli zero, 2, e64, m1, tu, ma
+; CHECK-V-NEXT: vmv.s.x v10, a0
+; CHECK-V-NEXT: addi a0, sp, 16
+; CHECK-V-NEXT: vl1r.v v8, (a0) # Unknown-size Folded Reload
+; CHECK-V-NEXT: vslideup.vi v10, v8, 1
; CHECK-V-NEXT: vsetivli zero, 3, e64, m2, tu, ma
-; CHECK-V-NEXT: vmv.s.x v8, a0
; CHECK-V-NEXT: csrr a0, vlenb
; CHECK-V-NEXT: slli a0, a0, 1
; CHECK-V-NEXT: add a0, sp, a0
; CHECK-V-NEXT: addi a0, a0, 16
-; CHECK-V-NEXT: vl2r.v v10, (a0) # Unknown-size Folded Reload
+; CHECK-V-NEXT: vl2r.v v8, (a0) # Unknown-size Folded Reload
; CHECK-V-NEXT: vslideup.vi v10, v8, 2
-; CHECK-V-NEXT: csrr a0, vlenb
-; CHECK-V-NEXT: slli a0, a0, 1
-; CHECK-V-NEXT: add a0, sp, a0
-; CHECK-V-NEXT: addi a0, a0, 16
+; CHECK-V-NEXT: addi a0, sp, 16
; CHECK-V-NEXT: vs2r.v v10, (a0) # Unknown-size Folded Spill
; CHECK-V-NEXT: fmv.w.x fa0, s0
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.l.s a0, fa0, rtz
; CHECK-V-NEXT: vsetivli zero, 4, e64, m2, ta, ma
; CHECK-V-NEXT: vmv.s.x v8, a0
-; CHECK-V-NEXT: csrr a0, vlenb
-; CHECK-V-NEXT: slli a0, a0, 1
-; CHECK-V-NEXT: add a0, sp, a0
-; CHECK-V-NEXT: addi a0, a0, 16
+; CHECK-V-NEXT: addi a0, sp, 16
; CHECK-V-NEXT: vl2r.v v10, (a0) # Unknown-size Folded Reload
; CHECK-V-NEXT: vslideup.vi v10, v8, 3
; CHECK-V-NEXT: lui a0, 524288
@@ -3966,54 +3946,49 @@ define <4 x i32> @utesth_f16i32_mm(<4 x half> %x) {
; CHECK-V-NEXT: sub sp, sp, a1
; CHECK-V-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x30, 0x22, 0x11, 0x04, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 48 + 4 * vlenb
; CHECK-V-NEXT: lhu s0, 24(a0)
-; CHECK-V-NEXT: lhu s1, 16(a0)
-; CHECK-V-NEXT: lhu s2, 0(a0)
-; CHECK-V-NEXT: lhu a0, 8(a0)
+; CHECK-V-NEXT: lhu s1, 0(a0)
+; CHECK-V-NEXT: lhu s2, 8(a0)
+; CHECK-V-NEXT: lhu a0, 16(a0)
; CHECK-V-NEXT: fmv.w.x fa0, a0
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.lu.s a0, fa0, rtz
; CHECK-V-NEXT: vsetivli zero, 1, e64, m1, ta, ma
; CHECK-V-NEXT: vmv.s.x v8, a0
-; CHECK-V-NEXT: addi a0, sp, 16
-; CHECK-V-NEXT: vs2r.v v8, (a0) # Unknown-size Folded Spill
-; CHECK-V-NEXT: fmv.w.x fa0, s2
-; CHECK-V-NEXT: call __extendhfsf2 at plt
-; CHECK-V-NEXT: fcvt.lu.s a0, fa0, rtz
-; CHECK-V-NEXT: vsetivli zero, 2, e64, m2, tu, ma
-; CHECK-V-NEXT: vmv.s.x v8, a0
-; CHECK-V-NEXT: addi a0, sp, 16
-; CHECK-V-NEXT: vl2r.v v10, (a0) # Unknown-size Folded Reload
-; CHECK-V-NEXT: vslideup.vi v8, v10, 1
; CHECK-V-NEXT: csrr a0, vlenb
; CHECK-V-NEXT: slli a0, a0, 1
; CHECK-V-NEXT: add a0, sp, a0
; CHECK-V-NEXT: addi a0, a0, 16
; CHECK-V-NEXT: vs2r.v v8, (a0) # Unknown-size Folded Spill
+; CHECK-V-NEXT: fmv.w.x fa0, s2
+; CHECK-V-NEXT: call __extendhfsf2 at plt
+; CHECK-V-NEXT: fcvt.lu.s a0, fa0, rtz
+; CHECK-V-NEXT: vsetivli zero, 1, e64, m1, ta, ma
; CHECK-V-NEXT: fmv.w.x fa0, s1
+; CHECK-V-NEXT: vmv.s.x v8, a0
+; CHECK-V-NEXT: addi a0, sp, 16
+; CHECK-V-NEXT: vs1r.v v8, (a0) # Unknown-size Folded Spill
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.lu.s a0, fa0, rtz
+; CHECK-V-NEXT: vsetivli zero, 2, e64, m1, tu, ma
+; CHECK-V-NEXT: vmv.s.x v10, a0
+; CHECK-V-NEXT: addi a0, sp, 16
+; CHECK-V-NEXT: vl1r.v v8, (a0) # Unknown-size Folded Reload
+; CHECK-V-NEXT: vslideup.vi v10, v8, 1
; CHECK-V-NEXT: vsetivli zero, 3, e64, m2, tu, ma
-; CHECK-V-NEXT: vmv.s.x v8, a0
; CHECK-V-NEXT: csrr a0, vlenb
; CHECK-V-NEXT: slli a0, a0, 1
; CHECK-V-NEXT: add a0, sp, a0
; CHECK-V-NEXT: addi a0, a0, 16
-; CHECK-V-NEXT: vl2r.v v10, (a0) # Unknown-size Folded Reload
+; CHECK-V-NEXT: vl2r.v v8, (a0) # Unknown-size Folded Reload
; CHECK-V-NEXT: vslideup.vi v10, v8, 2
-; CHECK-V-NEXT: csrr a0, vlenb
-; CHECK-V-NEXT: slli a0, a0, 1
-; CHECK-V-NEXT: add a0, sp, a0
-; CHECK-V-NEXT: addi a0, a0, 16
+; CHECK-V-NEXT: addi a0, sp, 16
; CHECK-V-NEXT: vs2r.v v10, (a0) # Unknown-size Folded Spill
; CHECK-V-NEXT: fmv.w.x fa0, s0
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.lu.s a0, fa0, rtz
; CHECK-V-NEXT: vsetivli zero, 4, e64, m2, ta, ma
; CHECK-V-NEXT: vmv.s.x v8, a0
-; CHECK-V-NEXT: csrr a0, vlenb
-; CHECK-V-NEXT: slli a0, a0, 1
-; CHECK-V-NEXT: add a0, sp, a0
-; CHECK-V-NEXT: addi a0, a0, 16
+; CHECK-V-NEXT: addi a0, sp, 16
; CHECK-V-NEXT: vl2r.v v10, (a0) # Unknown-size Folded Reload
; CHECK-V-NEXT: vslideup.vi v10, v8, 3
; CHECK-V-NEXT: li a0, -1
@@ -4146,54 +4121,49 @@ define <4 x i32> @ustest_f16i32_mm(<4 x half> %x) {
; CHECK-V-NEXT: sub sp, sp, a1
; CHECK-V-NEXT: .cfi_escape 0x0f, 0x0d, 0x72, 0x00, 0x11, 0x30, 0x22, 0x11, 0x04, 0x92, 0xa2, 0x38, 0x00, 0x1e, 0x22 # sp + 48 + 4 * vlenb
; CHECK-V-NEXT: lhu s0, 24(a0)
-; CHECK-V-NEXT: lhu s1, 16(a0)
-; CHECK-V-NEXT: lhu s2, 0(a0)
-; CHECK-V-NEXT: lhu a0, 8(a0)
+; CHECK-V-NEXT: lhu s1, 0(a0)
+; CHECK-V-NEXT: lhu s2, 8(a0)
+; CHECK-V-NEXT: lhu a0, 16(a0)
; CHECK-V-NEXT: fmv.w.x fa0, a0
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.l.s a0, fa0, rtz
; CHECK-V-NEXT: vsetivli zero, 1, e64, m1, ta, ma
; CHECK-V-NEXT: vmv.s.x v8, a0
-; CHECK-V-NEXT: addi a0, sp, 16
-; CHECK-V-NEXT: vs2r.v v8, (a0) # Unknown-size Folded Spill
-; CHECK-V-NEXT: fmv.w.x fa0, s2
-; CHECK-V-NEXT: call __extendhfsf2 at plt
-; CHECK-V-NEXT: fcvt.l.s a0, fa0, rtz
-; CHECK-V-NEXT: vsetivli zero, 2, e64, m2, tu, ma
-; CHECK-V-NEXT: vmv.s.x v8, a0
-; CHECK-V-NEXT: addi a0, sp, 16
-; CHECK-V-NEXT: vl2r.v v10, (a0) # Unknown-size Folded Reload
-; CHECK-V-NEXT: vslideup.vi v8, v10, 1
; CHECK-V-NEXT: csrr a0, vlenb
; CHECK-V-NEXT: slli a0, a0, 1
; CHECK-V-NEXT: add a0, sp, a0
; CHECK-V-NEXT: addi a0, a0, 16
; CHECK-V-NEXT: vs2r.v v8, (a0) # Unknown-size Folded Spill
+; CHECK-V-NEXT: fmv.w.x fa0, s2
+; CHECK-V-NEXT: call __extendhfsf2 at plt
+; CHECK-V-NEXT: fcvt.l.s a0, fa0, rtz
+; CHECK-V-NEXT: vsetivli zero, 1, e64, m1, ta, ma
; CHECK-V-NEXT: fmv.w.x fa0, s1
+; CHECK-V-NEXT: vmv.s.x v8, a0
+; CHECK-V-NEXT: addi a0, sp, 16
+; CHECK-V-NEXT: vs1r.v v8, (a0) # Unknown-size Folded Spill
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.l.s a0, fa0, rtz
+; CHECK-V-NEXT: vsetivli zero, 2, e64, m1, tu, ma
+; CHECK-V-NEXT: vmv.s.x v10, a0
+; CHECK-V-NEXT: addi a0, sp, 16
+; CHECK-V-NEXT: vl1r.v v8, (a0) # Unknown-size Folded Reload
+; CHECK-V-NEXT: vslideup.vi v10, v8, 1
; CHECK-V-NEXT: vsetivli zero, 3, e64, m2, tu, ma
-; CHECK-V-NEXT: vmv.s.x v8, a0
; CHECK-V-NEXT: csrr a0, vlenb
; CHECK-V-NEXT: slli a0, a0, 1
; CHECK-V-NEXT: add a0, sp, a0
; CHECK-V-NEXT: addi a0, a0, 16
-; CHECK-V-NEXT: vl2r.v v10, (a0) # Unknown-size Folded Reload
+; CHECK-V-NEXT: vl2r.v v8, (a0) # Unknown-size Folded Reload
; CHECK-V-NEXT: vslideup.vi v10, v8, 2
-; CHECK-V-NEXT: csrr a0, vlenb
-; CHECK-V-NEXT: slli a0, a0, 1
-; CHECK-V-NEXT: add a0, sp, a0
-; CHECK-V-NEXT: addi a0, a0, 16
+; CHECK-V-NEXT: addi a0, sp, 16
; CHECK-V-NEXT: vs2r.v v10, (a0) # Unknown-size Folded Spill
; CHECK-V-NEXT: fmv.w.x fa0, s0
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.l.s a0, fa0, rtz
; CHECK-V-NEXT: vsetivli zero, 4, e64, m2, ta, ma
; CHECK-V-NEXT: vmv.s.x v8, a0
-; CHECK-V-NEXT: csrr a0, vlenb
-; CHECK-V-NEXT: slli a0, a0, 1
-; CHECK-V-NEXT: add a0, sp, a0
-; CHECK-V-NEXT: addi a0, a0, 16
+; CHECK-V-NEXT: addi a0, sp, 16
; CHECK-V-NEXT: vl2r.v v10, (a0) # Unknown-size Folded Reload
; CHECK-V-NEXT: vslideup.vi v10, v8, 3
; CHECK-V-NEXT: li a0, -1
@@ -4775,8 +4745,8 @@ define <8 x i16> @stest_f16i16_mm(<8 x half> %x) {
; CHECK-V-NEXT: lhu s3, 32(a0)
; CHECK-V-NEXT: lhu s4, 24(a0)
; CHECK-V-NEXT: lhu s5, 16(a0)
-; CHECK-V-NEXT: lhu s6, 0(a0)
-; CHECK-V-NEXT: lhu a0, 8(a0)
+; CHECK-V-NEXT: lhu s6, 8(a0)
+; CHECK-V-NEXT: lhu a0, 0(a0)
; CHECK-V-NEXT: fmv.w.x fa0, a0
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.l.s a0, fa0, rtz
@@ -4787,16 +4757,16 @@ define <8 x i16> @stest_f16i16_mm(<8 x half> %x) {
; CHECK-V-NEXT: fmv.w.x fa0, s6
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.l.s a0, fa0, rtz
-; CHECK-V-NEXT: vsetivli zero, 2, e32, m2, tu, ma
+; CHECK-V-NEXT: vsetivli zero, 2, e32, m1, tu, ma
; CHECK-V-NEXT: vmv.s.x v8, a0
; CHECK-V-NEXT: addi a0, sp, 16
; CHECK-V-NEXT: vl2r.v v10, (a0) # Unknown-size Folded Reload
-; CHECK-V-NEXT: vslideup.vi v8, v10, 1
-; CHECK-V-NEXT: vs2r.v v8, (a0) # Unknown-size Folded Spill
+; CHECK-V-NEXT: vslideup.vi v10, v8, 1
+; CHECK-V-NEXT: vs2r.v v10, (a0) # Unknown-size Folded Spill
; CHECK-V-NEXT: fmv.w.x fa0, s5
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.l.s a0, fa0, rtz
-; CHECK-V-NEXT: vsetivli zero, 3, e32, m2, tu, ma
+; CHECK-V-NEXT: vsetivli zero, 3, e32, m1, tu, ma
; CHECK-V-NEXT: vmv.s.x v8, a0
; CHECK-V-NEXT: addi a0, sp, 16
; CHECK-V-NEXT: vl2r.v v10, (a0) # Unknown-size Folded Reload
@@ -4805,7 +4775,7 @@ define <8 x i16> @stest_f16i16_mm(<8 x half> %x) {
; CHECK-V-NEXT: fmv.w.x fa0, s4
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.l.s a0, fa0, rtz
-; CHECK-V-NEXT: vsetivli zero, 4, e32, m2, tu, ma
+; CHECK-V-NEXT: vsetivli zero, 4, e32, m1, tu, ma
; CHECK-V-NEXT: vmv.s.x v8, a0
; CHECK-V-NEXT: addi a0, sp, 16
; CHECK-V-NEXT: vl2r.v v10, (a0) # Unknown-size Folded Reload
@@ -5054,8 +5024,8 @@ define <8 x i16> @utesth_f16i16_mm(<8 x half> %x) {
; CHECK-V-NEXT: lhu s3, 32(a0)
; CHECK-V-NEXT: lhu s4, 24(a0)
; CHECK-V-NEXT: lhu s5, 16(a0)
-; CHECK-V-NEXT: lhu s6, 0(a0)
-; CHECK-V-NEXT: lhu a0, 8(a0)
+; CHECK-V-NEXT: lhu s6, 8(a0)
+; CHECK-V-NEXT: lhu a0, 0(a0)
; CHECK-V-NEXT: fmv.w.x fa0, a0
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.lu.s a0, fa0, rtz
@@ -5066,16 +5036,16 @@ define <8 x i16> @utesth_f16i16_mm(<8 x half> %x) {
; CHECK-V-NEXT: fmv.w.x fa0, s6
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.lu.s a0, fa0, rtz
-; CHECK-V-NEXT: vsetivli zero, 2, e32, m2, tu, ma
+; CHECK-V-NEXT: vsetivli zero, 2, e32, m1, tu, ma
; CHECK-V-NEXT: vmv.s.x v8, a0
; CHECK-V-NEXT: addi a0, sp, 16
; CHECK-V-NEXT: vl2r.v v10, (a0) # Unknown-size Folded Reload
-; CHECK-V-NEXT: vslideup.vi v8, v10, 1
-; CHECK-V-NEXT: vs2r.v v8, (a0) # Unknown-size Folded Spill
+; CHECK-V-NEXT: vslideup.vi v10, v8, 1
+; CHECK-V-NEXT: vs2r.v v10, (a0) # Unknown-size Folded Spill
; CHECK-V-NEXT: fmv.w.x fa0, s5
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.lu.s a0, fa0, rtz
-; CHECK-V-NEXT: vsetivli zero, 3, e32, m2, tu, ma
+; CHECK-V-NEXT: vsetivli zero, 3, e32, m1, tu, ma
; CHECK-V-NEXT: vmv.s.x v8, a0
; CHECK-V-NEXT: addi a0, sp, 16
; CHECK-V-NEXT: vl2r.v v10, (a0) # Unknown-size Folded Reload
@@ -5084,7 +5054,7 @@ define <8 x i16> @utesth_f16i16_mm(<8 x half> %x) {
; CHECK-V-NEXT: fmv.w.x fa0, s4
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.lu.s a0, fa0, rtz
-; CHECK-V-NEXT: vsetivli zero, 4, e32, m2, tu, ma
+; CHECK-V-NEXT: vsetivli zero, 4, e32, m1, tu, ma
; CHECK-V-NEXT: vmv.s.x v8, a0
; CHECK-V-NEXT: addi a0, sp, 16
; CHECK-V-NEXT: vl2r.v v10, (a0) # Unknown-size Folded Reload
@@ -5356,8 +5326,8 @@ define <8 x i16> @ustest_f16i16_mm(<8 x half> %x) {
; CHECK-V-NEXT: lhu s3, 32(a0)
; CHECK-V-NEXT: lhu s4, 24(a0)
; CHECK-V-NEXT: lhu s5, 16(a0)
-; CHECK-V-NEXT: lhu s6, 0(a0)
-; CHECK-V-NEXT: lhu a0, 8(a0)
+; CHECK-V-NEXT: lhu s6, 8(a0)
+; CHECK-V-NEXT: lhu a0, 0(a0)
; CHECK-V-NEXT: fmv.w.x fa0, a0
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.l.s a0, fa0, rtz
@@ -5368,16 +5338,16 @@ define <8 x i16> @ustest_f16i16_mm(<8 x half> %x) {
; CHECK-V-NEXT: fmv.w.x fa0, s6
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.l.s a0, fa0, rtz
-; CHECK-V-NEXT: vsetivli zero, 2, e32, m2, tu, ma
+; CHECK-V-NEXT: vsetivli zero, 2, e32, m1, tu, ma
; CHECK-V-NEXT: vmv.s.x v8, a0
; CHECK-V-NEXT: addi a0, sp, 16
; CHECK-V-NEXT: vl2r.v v10, (a0) # Unknown-size Folded Reload
-; CHECK-V-NEXT: vslideup.vi v8, v10, 1
-; CHECK-V-NEXT: vs2r.v v8, (a0) # Unknown-size Folded Spill
+; CHECK-V-NEXT: vslideup.vi v10, v8, 1
+; CHECK-V-NEXT: vs2r.v v10, (a0) # Unknown-size Folded Spill
; CHECK-V-NEXT: fmv.w.x fa0, s5
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.l.s a0, fa0, rtz
-; CHECK-V-NEXT: vsetivli zero, 3, e32, m2, tu, ma
+; CHECK-V-NEXT: vsetivli zero, 3, e32, m1, tu, ma
; CHECK-V-NEXT: vmv.s.x v8, a0
; CHECK-V-NEXT: addi a0, sp, 16
; CHECK-V-NEXT: vl2r.v v10, (a0) # Unknown-size Folded Reload
@@ -5386,7 +5356,7 @@ define <8 x i16> @ustest_f16i16_mm(<8 x half> %x) {
; CHECK-V-NEXT: fmv.w.x fa0, s4
; CHECK-V-NEXT: call __extendhfsf2 at plt
; CHECK-V-NEXT: fcvt.l.s a0, fa0, rtz
-; CHECK-V-NEXT: vsetivli zero, 4, e32, m2, tu, ma
+; CHECK-V-NEXT: vsetivli zero, 4, e32, m1, tu, ma
; CHECK-V-NEXT: vmv.s.x v8, a0
; CHECK-V-NEXT: addi a0, sp, 16
; CHECK-V-NEXT: vl2r.v v10, (a0) # Unknown-size Folded Reload
>From 8b637f0ec6255b3054a26f2c7b4bf94cd01d7f3b Mon Sep 17 00:00:00 2001
From: Luke Lau <luke at igalia.com>
Date: Tue, 12 Sep 2023 14:35:22 +0100
Subject: [PATCH 5/5] [RISCV] Reduce LMUL when index is known when lowering
insert_vector_elt
Continuing on from #65997, if the index of insert_vector_elt is a constant then
we can work out what the minimum number of registers will be needed for the
slideup and choose a smaller type to operate on. This reduces the LMUL for not
just the slideup but also for the scalar insert.
---
llvm/lib/Target/RISCV/RISCVISelLowering.cpp | 77 +++--
.../RISCV/rvv/fixed-vectors-insert-i1.ll | 2 +-
.../CodeGen/RISCV/rvv/fixed-vectors-insert.ll | 14 +-
.../RISCV/rvv/fixed-vectors-masked-gather.ll | 292 +++++++++---------
llvm/test/CodeGen/RISCV/rvv/insertelt-fp.ll | 34 +-
llvm/test/CodeGen/RISCV/rvv/insertelt-i1.ll | 6 +-
.../CodeGen/RISCV/rvv/insertelt-int-rv32.ll | 54 ++--
.../CodeGen/RISCV/rvv/insertelt-int-rv64.ll | 46 +--
.../RISCV/rvv/vsetvli-insert-crossbb.ll | 5 +-
9 files changed, 278 insertions(+), 252 deletions(-)
diff --git a/llvm/lib/Target/RISCV/RISCVISelLowering.cpp b/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
index 3a81179d1cfc120..db743f1c67a6232 100644
--- a/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
+++ b/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
@@ -7345,6 +7345,32 @@ RISCVTargetLowering::lowerVectorFPExtendOrRoundLike(SDValue Op,
return Result;
}
+// Given a scalable vector type and an index into it, returns the type for the
+// smallest subvector that the index fits in. This can be used to reduce LMUL
+// for operations like vslidedown.
+//
+// E.g. With Zvl128b, index 3 in a nxv4i32 fits within the first nxv2i32.
+static std::optional<MVT>
+getSmallestVTForIndex(MVT VecVT, unsigned MaxIdx, SDLoc DL, SelectionDAG &DAG,
+ const RISCVSubtarget &Subtarget) {
+ assert(VecVT.isScalableVector());
+ const unsigned EltSize = VecVT.getScalarSizeInBits();
+ const unsigned VectorBitsMin = Subtarget.getRealMinVLen();
+ const unsigned MinVLMAX = VectorBitsMin / EltSize;
+ MVT SmallerVT;
+ if (MaxIdx < MinVLMAX)
+ SmallerVT = getLMUL1VT(VecVT);
+ else if (MaxIdx < MinVLMAX * 2)
+ SmallerVT = getLMUL1VT(VecVT).getDoubleNumVectorElementsVT();
+ else if (MaxIdx < MinVLMAX * 4)
+ SmallerVT = getLMUL1VT(VecVT)
+ .getDoubleNumVectorElementsVT()
+ .getDoubleNumVectorElementsVT();
+ if (!SmallerVT.isValid() || !VecVT.bitsGT(SmallerVT))
+ return std::nullopt;
+ return SmallerVT;
+}
+
// Custom-legalize INSERT_VECTOR_ELT so that the value is inserted into the
// first position of a vector, and that vector is slid up to the insert index.
// By limiting the active vector length to index+1 and merging with the
@@ -7375,6 +7401,19 @@ SDValue RISCVTargetLowering::lowerINSERT_VECTOR_ELT(SDValue Op,
Vec = convertToScalableVector(ContainerVT, Vec, DAG, Subtarget);
}
+ MVT OrigContainerVT = ContainerVT;
+ SDValue OrigVec = Vec;
+ // If we know the index we're going to insert at, we can shrink down Vec so
+ // we're performing the vslide1down on a smaller LMUL.
+ if (auto *CIdx = dyn_cast<ConstantSDNode>(Idx)) {
+ if (auto ShrunkVT = getSmallestVTForIndex(ContainerVT, CIdx->getZExtValue(),
+ DL, DAG, Subtarget)) {
+ ContainerVT = *ShrunkVT;
+ Vec = DAG.getNode(ISD::EXTRACT_SUBVECTOR, DL, ContainerVT, Vec,
+ DAG.getVectorIdxConstant(0, DL));
+ }
+ }
+
MVT XLenVT = Subtarget.getXLenVT();
bool IsLegalInsert = Subtarget.is64Bit() || Val.getValueType() != MVT::i64;
@@ -7399,6 +7438,10 @@ SDValue RISCVTargetLowering::lowerINSERT_VECTOR_ELT(SDValue Op,
VecVT.isFloatingPoint() ? RISCVISD::VFMV_S_F_VL : RISCVISD::VMV_S_X_VL;
if (isNullConstant(Idx)) {
Vec = DAG.getNode(Opc, DL, ContainerVT, Vec, Val, VL);
+
+ if (ContainerVT != OrigContainerVT)
+ Vec = DAG.getNode(ISD::INSERT_SUBVECTOR, DL, OrigContainerVT, OrigVec,
+ Vec, DAG.getVectorIdxConstant(0, DL));
if (!VecVT.isFixedLengthVector())
return Vec;
return convertFromScalableVector(VecVT, Vec, DAG, Subtarget);
@@ -7431,6 +7474,10 @@ SDValue RISCVTargetLowering::lowerINSERT_VECTOR_ELT(SDValue Op,
// Bitcast back to the right container type.
ValInVec = DAG.getBitcast(ContainerVT, ValInVec);
+ if (ContainerVT != OrigContainerVT)
+ ValInVec =
+ DAG.getNode(ISD::INSERT_SUBVECTOR, DL, OrigContainerVT, OrigVec,
+ ValInVec, DAG.getVectorIdxConstant(0, DL));
if (!VecVT.isFixedLengthVector())
return ValInVec;
return convertFromScalableVector(VecVT, ValInVec, DAG, Subtarget);
@@ -7461,37 +7508,15 @@ SDValue RISCVTargetLowering::lowerINSERT_VECTOR_ELT(SDValue Op,
Policy = RISCVII::TAIL_AGNOSTIC;
SDValue Slideup = getVSlideup(DAG, Subtarget, DL, ContainerVT, Vec, ValInVec,
Idx, Mask, InsertVL, Policy);
+
+ if (ContainerVT != OrigContainerVT)
+ Slideup = DAG.getNode(ISD::INSERT_SUBVECTOR, DL, OrigContainerVT, OrigVec,
+ Slideup, DAG.getVectorIdxConstant(0, DL));
if (!VecVT.isFixedLengthVector())
return Slideup;
return convertFromScalableVector(VecVT, Slideup, DAG, Subtarget);
}
-// Given a scalable vector type and an index into it, returns the type for the
-// smallest subvector that the index fits in. This can be used to reduce LMUL
-// for operations like vslidedown.
-//
-// E.g. With Zvl128b, index 3 in a nxv4i32 fits within the first nxv2i32.
-static std::optional<MVT>
-getSmallestVTForIndex(MVT VecVT, unsigned MaxIdx, SDLoc DL, SelectionDAG &DAG,
- const RISCVSubtarget &Subtarget) {
- assert(VecVT.isScalableVector());
- const unsigned EltSize = VecVT.getScalarSizeInBits();
- const unsigned VectorBitsMin = Subtarget.getRealMinVLen();
- const unsigned MinVLMAX = VectorBitsMin / EltSize;
- MVT SmallerVT;
- if (MaxIdx < MinVLMAX)
- SmallerVT = getLMUL1VT(VecVT);
- else if (MaxIdx < MinVLMAX * 2)
- SmallerVT = getLMUL1VT(VecVT).getDoubleNumVectorElementsVT();
- else if (MaxIdx < MinVLMAX * 4)
- SmallerVT = getLMUL1VT(VecVT)
- .getDoubleNumVectorElementsVT()
- .getDoubleNumVectorElementsVT();
- if (!SmallerVT.isValid() || !VecVT.bitsGT(SmallerVT))
- return std::nullopt;
- return SmallerVT;
-}
-
// Custom-lower EXTRACT_VECTOR_ELT operations to slide the vector down, then
// extract the first element: (extractelt (slidedown vec, idx), 0). For integer
// types this is done using VMV_X_S to allow us to glean information about the
diff --git a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-insert-i1.ll b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-insert-i1.ll
index b3cbad3d9e6b1d7..f7737784d4ca57e 100644
--- a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-insert-i1.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-insert-i1.ll
@@ -108,7 +108,7 @@ define <64 x i1> @insertelt_v64i1(<64 x i1> %x, i1 %elt) nounwind {
; CHECK-NEXT: vmv.v.i v8, 0
; CHECK-NEXT: vmerge.vim v8, v8, 1, v0
; CHECK-NEXT: vmv.s.x v12, a0
-; CHECK-NEXT: vsetivli zero, 2, e8, m4, tu, ma
+; CHECK-NEXT: vsetivli zero, 2, e8, m1, tu, ma
; CHECK-NEXT: vslideup.vi v8, v12, 1
; CHECK-NEXT: vsetvli zero, a1, e8, m4, ta, ma
; CHECK-NEXT: vand.vi v8, v8, 1
diff --git a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-insert.ll b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-insert.ll
index 373a96356a207e2..cbcca9d2696f4ba 100644
--- a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-insert.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-insert.ll
@@ -40,7 +40,7 @@ define <32 x i32> @insertelt_v32i32_0(<32 x i32> %a, i32 %y) {
; CHECK-LABEL: insertelt_v32i32_0:
; CHECK: # %bb.0:
; CHECK-NEXT: li a1, 32
-; CHECK-NEXT: vsetvli zero, a1, e32, m8, tu, ma
+; CHECK-NEXT: vsetvli zero, a1, e32, m1, tu, ma
; CHECK-NEXT: vmv.s.x v8, a0
; CHECK-NEXT: ret
%b = insertelement <32 x i32> %a, i32 %y, i32 0
@@ -54,7 +54,7 @@ define <32 x i32> @insertelt_v32i32_4(<32 x i32> %a, i32 %y) {
; CHECK-NEXT: li a1, 32
; CHECK-NEXT: vsetvli zero, a1, e32, m1, ta, ma
; CHECK-NEXT: vmv.s.x v16, a0
-; CHECK-NEXT: vsetivli zero, 5, e32, m8, tu, ma
+; CHECK-NEXT: vsetivli zero, 5, e32, m2, tu, ma
; CHECK-NEXT: vslideup.vi v8, v16, 4
; CHECK-NEXT: ret
%b = insertelement <32 x i32> %a, i32 %y, i32 4
@@ -92,7 +92,7 @@ define <64 x i32> @insertelt_v64i32_0(<64 x i32> %a, i32 %y) {
; CHECK-LABEL: insertelt_v64i32_0:
; CHECK: # %bb.0:
; CHECK-NEXT: li a1, 32
-; CHECK-NEXT: vsetvli zero, a1, e32, m8, tu, ma
+; CHECK-NEXT: vsetvli zero, a1, e32, m1, tu, ma
; CHECK-NEXT: vmv.s.x v8, a0
; CHECK-NEXT: ret
%b = insertelement <64 x i32> %a, i32 %y, i32 0
@@ -390,7 +390,7 @@ define <8 x i64> @insertelt_v8i64_0(<8 x i64> %a, ptr %x) {
; CHECK-LABEL: insertelt_v8i64_0:
; CHECK: # %bb.0:
; CHECK-NEXT: li a0, -1
-; CHECK-NEXT: vsetivli zero, 8, e64, m4, tu, ma
+; CHECK-NEXT: vsetivli zero, 8, e64, m1, tu, ma
; CHECK-NEXT: vmv.s.x v8, a0
; CHECK-NEXT: ret
%b = insertelement <8 x i64> %a, i64 -1, i32 0
@@ -468,7 +468,7 @@ define <8 x i64> @insertelt_c6_v8i64_0(<8 x i64> %a, ptr %x) {
; CHECK-LABEL: insertelt_c6_v8i64_0:
; CHECK: # %bb.0:
; CHECK-NEXT: li a0, 6
-; CHECK-NEXT: vsetivli zero, 8, e64, m4, tu, ma
+; CHECK-NEXT: vsetivli zero, 8, e64, m1, tu, ma
; CHECK-NEXT: vmv.s.x v8, a0
; CHECK-NEXT: ret
%b = insertelement <8 x i64> %a, i64 6, i32 0
@@ -550,9 +550,9 @@ define void @insertelt_c6_v8i64_0_add(ptr %x, ptr %y) {
; CHECK-NEXT: vsetivli zero, 8, e64, m4, ta, ma
; CHECK-NEXT: vle64.v v8, (a0)
; CHECK-NEXT: li a2, 6
-; CHECK-NEXT: vsetvli zero, zero, e64, m4, tu, ma
+; CHECK-NEXT: vsetivli zero, 8, e64, m1, tu, ma
; CHECK-NEXT: vmv.s.x v8, a2
-; CHECK-NEXT: vsetvli zero, zero, e64, m4, ta, ma
+; CHECK-NEXT: vsetivli zero, 8, e64, m4, ta, ma
; CHECK-NEXT: vle64.v v12, (a1)
; CHECK-NEXT: vadd.vv v8, v8, v12
; CHECK-NEXT: vse64.v v8, (a0)
diff --git a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-masked-gather.ll b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-masked-gather.ll
index f7352b4659e5a9b..5ae0884068f9bc3 100644
--- a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-masked-gather.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-masked-gather.ll
@@ -2426,14 +2426,14 @@ define <8 x i32> @mgather_v8i32(<8 x ptr> %ptrs, <8 x i1> %m, <8 x i32> %passthr
; RV64ZVE32F-NEXT: .LBB34_9: # %cond.load
; RV64ZVE32F-NEXT: ld a2, 0(a0)
; RV64ZVE32F-NEXT: lw a2, 0(a2)
-; RV64ZVE32F-NEXT: vsetivli zero, 8, e32, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 8, e32, m1, tu, ma
; RV64ZVE32F-NEXT: vmv.s.x v8, a2
; RV64ZVE32F-NEXT: andi a2, a1, 2
; RV64ZVE32F-NEXT: beqz a2, .LBB34_2
; RV64ZVE32F-NEXT: .LBB34_10: # %cond.load1
; RV64ZVE32F-NEXT: ld a2, 8(a0)
; RV64ZVE32F-NEXT: lw a2, 0(a2)
-; RV64ZVE32F-NEXT: vsetivli zero, 2, e32, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 2, e32, m1, tu, ma
; RV64ZVE32F-NEXT: vmv.s.x v10, a2
; RV64ZVE32F-NEXT: vslideup.vi v8, v10, 1
; RV64ZVE32F-NEXT: andi a2, a1, 4
@@ -2441,7 +2441,7 @@ define <8 x i32> @mgather_v8i32(<8 x ptr> %ptrs, <8 x i1> %m, <8 x i32> %passthr
; RV64ZVE32F-NEXT: .LBB34_11: # %cond.load4
; RV64ZVE32F-NEXT: ld a2, 16(a0)
; RV64ZVE32F-NEXT: lw a2, 0(a2)
-; RV64ZVE32F-NEXT: vsetivli zero, 3, e32, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 3, e32, m1, tu, ma
; RV64ZVE32F-NEXT: vmv.s.x v10, a2
; RV64ZVE32F-NEXT: vslideup.vi v8, v10, 2
; RV64ZVE32F-NEXT: andi a2, a1, 8
@@ -2449,7 +2449,7 @@ define <8 x i32> @mgather_v8i32(<8 x ptr> %ptrs, <8 x i1> %m, <8 x i32> %passthr
; RV64ZVE32F-NEXT: .LBB34_12: # %cond.load7
; RV64ZVE32F-NEXT: ld a2, 24(a0)
; RV64ZVE32F-NEXT: lw a2, 0(a2)
-; RV64ZVE32F-NEXT: vsetivli zero, 4, e32, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 4, e32, m1, tu, ma
; RV64ZVE32F-NEXT: vmv.s.x v10, a2
; RV64ZVE32F-NEXT: vslideup.vi v8, v10, 3
; RV64ZVE32F-NEXT: andi a2, a1, 16
@@ -2520,7 +2520,7 @@ define <8 x i32> @mgather_baseidx_v8i8_v8i32(ptr %base, <8 x i8> %idxs, <8 x i1>
; RV64ZVE32F-NEXT: slli a2, a2, 2
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: lw a2, 0(a2)
-; RV64ZVE32F-NEXT: vsetivli zero, 8, e32, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 8, e32, m1, tu, ma
; RV64ZVE32F-NEXT: vmv.s.x v10, a2
; RV64ZVE32F-NEXT: .LBB35_2: # %else
; RV64ZVE32F-NEXT: andi a2, a1, 2
@@ -2533,9 +2533,9 @@ define <8 x i32> @mgather_baseidx_v8i8_v8i32(ptr %base, <8 x i8> %idxs, <8 x i1>
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: lw a2, 0(a2)
; RV64ZVE32F-NEXT: vsetvli zero, zero, e32, m1, ta, ma
-; RV64ZVE32F-NEXT: vmv.s.x v12, a2
-; RV64ZVE32F-NEXT: vsetivli zero, 2, e32, m2, tu, ma
-; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 1
+; RV64ZVE32F-NEXT: vmv.s.x v9, a2
+; RV64ZVE32F-NEXT: vsetivli zero, 2, e32, m1, tu, ma
+; RV64ZVE32F-NEXT: vslideup.vi v10, v9, 1
; RV64ZVE32F-NEXT: .LBB35_4: # %else2
; RV64ZVE32F-NEXT: vsetivli zero, 4, e8, mf2, ta, ma
; RV64ZVE32F-NEXT: vslidedown.vi v9, v8, 4
@@ -2581,7 +2581,7 @@ define <8 x i32> @mgather_baseidx_v8i8_v8i32(ptr %base, <8 x i8> %idxs, <8 x i1>
; RV64ZVE32F-NEXT: lw a2, 0(a2)
; RV64ZVE32F-NEXT: vsetvli zero, zero, e32, m1, ta, ma
; RV64ZVE32F-NEXT: vmv.s.x v12, a2
-; RV64ZVE32F-NEXT: vsetivli zero, 3, e32, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 3, e32, m1, tu, ma
; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 2
; RV64ZVE32F-NEXT: andi a2, a1, 8
; RV64ZVE32F-NEXT: beqz a2, .LBB35_6
@@ -2593,9 +2593,9 @@ define <8 x i32> @mgather_baseidx_v8i8_v8i32(ptr %base, <8 x i8> %idxs, <8 x i1>
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: lw a2, 0(a2)
; RV64ZVE32F-NEXT: vsetvli zero, zero, e32, m1, ta, ma
-; RV64ZVE32F-NEXT: vmv.s.x v12, a2
-; RV64ZVE32F-NEXT: vsetivli zero, 4, e32, m2, tu, ma
-; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 3
+; RV64ZVE32F-NEXT: vmv.s.x v8, a2
+; RV64ZVE32F-NEXT: vsetivli zero, 4, e32, m1, tu, ma
+; RV64ZVE32F-NEXT: vslideup.vi v10, v8, 3
; RV64ZVE32F-NEXT: andi a2, a1, 16
; RV64ZVE32F-NEXT: beqz a2, .LBB35_7
; RV64ZVE32F-NEXT: .LBB35_14: # %cond.load10
@@ -2670,7 +2670,7 @@ define <8 x i32> @mgather_baseidx_sext_v8i8_v8i32(ptr %base, <8 x i8> %idxs, <8
; RV64ZVE32F-NEXT: slli a2, a2, 2
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: lw a2, 0(a2)
-; RV64ZVE32F-NEXT: vsetivli zero, 8, e32, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 8, e32, m1, tu, ma
; RV64ZVE32F-NEXT: vmv.s.x v10, a2
; RV64ZVE32F-NEXT: .LBB36_2: # %else
; RV64ZVE32F-NEXT: andi a2, a1, 2
@@ -2683,9 +2683,9 @@ define <8 x i32> @mgather_baseidx_sext_v8i8_v8i32(ptr %base, <8 x i8> %idxs, <8
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: lw a2, 0(a2)
; RV64ZVE32F-NEXT: vsetvli zero, zero, e32, m1, ta, ma
-; RV64ZVE32F-NEXT: vmv.s.x v12, a2
-; RV64ZVE32F-NEXT: vsetivli zero, 2, e32, m2, tu, ma
-; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 1
+; RV64ZVE32F-NEXT: vmv.s.x v9, a2
+; RV64ZVE32F-NEXT: vsetivli zero, 2, e32, m1, tu, ma
+; RV64ZVE32F-NEXT: vslideup.vi v10, v9, 1
; RV64ZVE32F-NEXT: .LBB36_4: # %else2
; RV64ZVE32F-NEXT: vsetivli zero, 4, e8, mf2, ta, ma
; RV64ZVE32F-NEXT: vslidedown.vi v9, v8, 4
@@ -2731,7 +2731,7 @@ define <8 x i32> @mgather_baseidx_sext_v8i8_v8i32(ptr %base, <8 x i8> %idxs, <8
; RV64ZVE32F-NEXT: lw a2, 0(a2)
; RV64ZVE32F-NEXT: vsetvli zero, zero, e32, m1, ta, ma
; RV64ZVE32F-NEXT: vmv.s.x v12, a2
-; RV64ZVE32F-NEXT: vsetivli zero, 3, e32, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 3, e32, m1, tu, ma
; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 2
; RV64ZVE32F-NEXT: andi a2, a1, 8
; RV64ZVE32F-NEXT: beqz a2, .LBB36_6
@@ -2743,9 +2743,9 @@ define <8 x i32> @mgather_baseidx_sext_v8i8_v8i32(ptr %base, <8 x i8> %idxs, <8
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: lw a2, 0(a2)
; RV64ZVE32F-NEXT: vsetvli zero, zero, e32, m1, ta, ma
-; RV64ZVE32F-NEXT: vmv.s.x v12, a2
-; RV64ZVE32F-NEXT: vsetivli zero, 4, e32, m2, tu, ma
-; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 3
+; RV64ZVE32F-NEXT: vmv.s.x v8, a2
+; RV64ZVE32F-NEXT: vsetivli zero, 4, e32, m1, tu, ma
+; RV64ZVE32F-NEXT: vslideup.vi v10, v8, 3
; RV64ZVE32F-NEXT: andi a2, a1, 16
; RV64ZVE32F-NEXT: beqz a2, .LBB36_7
; RV64ZVE32F-NEXT: .LBB36_14: # %cond.load10
@@ -2822,7 +2822,7 @@ define <8 x i32> @mgather_baseidx_zext_v8i8_v8i32(ptr %base, <8 x i8> %idxs, <8
; RV64ZVE32F-NEXT: slli a2, a2, 2
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: lw a2, 0(a2)
-; RV64ZVE32F-NEXT: vsetivli zero, 8, e32, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 8, e32, m1, tu, ma
; RV64ZVE32F-NEXT: vmv.s.x v10, a2
; RV64ZVE32F-NEXT: .LBB37_2: # %else
; RV64ZVE32F-NEXT: andi a2, a1, 2
@@ -2836,9 +2836,9 @@ define <8 x i32> @mgather_baseidx_zext_v8i8_v8i32(ptr %base, <8 x i8> %idxs, <8
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: lw a2, 0(a2)
; RV64ZVE32F-NEXT: vsetvli zero, zero, e32, m1, ta, ma
-; RV64ZVE32F-NEXT: vmv.s.x v12, a2
-; RV64ZVE32F-NEXT: vsetivli zero, 2, e32, m2, tu, ma
-; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 1
+; RV64ZVE32F-NEXT: vmv.s.x v9, a2
+; RV64ZVE32F-NEXT: vsetivli zero, 2, e32, m1, tu, ma
+; RV64ZVE32F-NEXT: vslideup.vi v10, v9, 1
; RV64ZVE32F-NEXT: .LBB37_4: # %else2
; RV64ZVE32F-NEXT: vsetivli zero, 4, e8, mf2, ta, ma
; RV64ZVE32F-NEXT: vslidedown.vi v9, v8, 4
@@ -2886,7 +2886,7 @@ define <8 x i32> @mgather_baseidx_zext_v8i8_v8i32(ptr %base, <8 x i8> %idxs, <8
; RV64ZVE32F-NEXT: lw a2, 0(a2)
; RV64ZVE32F-NEXT: vsetvli zero, zero, e32, m1, ta, ma
; RV64ZVE32F-NEXT: vmv.s.x v12, a2
-; RV64ZVE32F-NEXT: vsetivli zero, 3, e32, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 3, e32, m1, tu, ma
; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 2
; RV64ZVE32F-NEXT: andi a2, a1, 8
; RV64ZVE32F-NEXT: beqz a2, .LBB37_6
@@ -2900,7 +2900,7 @@ define <8 x i32> @mgather_baseidx_zext_v8i8_v8i32(ptr %base, <8 x i8> %idxs, <8
; RV64ZVE32F-NEXT: lw a2, 0(a2)
; RV64ZVE32F-NEXT: vsetvli zero, zero, e32, m1, ta, ma
; RV64ZVE32F-NEXT: vmv.s.x v8, a2
-; RV64ZVE32F-NEXT: vsetivli zero, 4, e32, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 4, e32, m1, tu, ma
; RV64ZVE32F-NEXT: vslideup.vi v10, v8, 3
; RV64ZVE32F-NEXT: andi a2, a1, 16
; RV64ZVE32F-NEXT: beqz a2, .LBB37_7
@@ -2981,7 +2981,7 @@ define <8 x i32> @mgather_baseidx_v8i16_v8i32(ptr %base, <8 x i16> %idxs, <8 x i
; RV64ZVE32F-NEXT: slli a2, a2, 2
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: lw a2, 0(a2)
-; RV64ZVE32F-NEXT: vsetivli zero, 8, e32, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 8, e32, m1, tu, ma
; RV64ZVE32F-NEXT: vmv.s.x v10, a2
; RV64ZVE32F-NEXT: .LBB38_2: # %else
; RV64ZVE32F-NEXT: andi a2, a1, 2
@@ -2994,9 +2994,9 @@ define <8 x i32> @mgather_baseidx_v8i16_v8i32(ptr %base, <8 x i16> %idxs, <8 x i
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: lw a2, 0(a2)
; RV64ZVE32F-NEXT: vsetvli zero, zero, e32, m1, ta, ma
-; RV64ZVE32F-NEXT: vmv.s.x v12, a2
-; RV64ZVE32F-NEXT: vsetivli zero, 2, e32, m2, tu, ma
-; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 1
+; RV64ZVE32F-NEXT: vmv.s.x v9, a2
+; RV64ZVE32F-NEXT: vsetivli zero, 2, e32, m1, tu, ma
+; RV64ZVE32F-NEXT: vslideup.vi v10, v9, 1
; RV64ZVE32F-NEXT: .LBB38_4: # %else2
; RV64ZVE32F-NEXT: vsetivli zero, 4, e16, m1, ta, ma
; RV64ZVE32F-NEXT: vslidedown.vi v9, v8, 4
@@ -3042,7 +3042,7 @@ define <8 x i32> @mgather_baseidx_v8i16_v8i32(ptr %base, <8 x i16> %idxs, <8 x i
; RV64ZVE32F-NEXT: lw a2, 0(a2)
; RV64ZVE32F-NEXT: vsetvli zero, zero, e32, m1, ta, ma
; RV64ZVE32F-NEXT: vmv.s.x v12, a2
-; RV64ZVE32F-NEXT: vsetivli zero, 3, e32, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 3, e32, m1, tu, ma
; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 2
; RV64ZVE32F-NEXT: andi a2, a1, 8
; RV64ZVE32F-NEXT: beqz a2, .LBB38_6
@@ -3054,9 +3054,9 @@ define <8 x i32> @mgather_baseidx_v8i16_v8i32(ptr %base, <8 x i16> %idxs, <8 x i
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: lw a2, 0(a2)
; RV64ZVE32F-NEXT: vsetvli zero, zero, e32, m1, ta, ma
-; RV64ZVE32F-NEXT: vmv.s.x v12, a2
-; RV64ZVE32F-NEXT: vsetivli zero, 4, e32, m2, tu, ma
-; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 3
+; RV64ZVE32F-NEXT: vmv.s.x v8, a2
+; RV64ZVE32F-NEXT: vsetivli zero, 4, e32, m1, tu, ma
+; RV64ZVE32F-NEXT: vslideup.vi v10, v8, 3
; RV64ZVE32F-NEXT: andi a2, a1, 16
; RV64ZVE32F-NEXT: beqz a2, .LBB38_7
; RV64ZVE32F-NEXT: .LBB38_14: # %cond.load10
@@ -3132,7 +3132,7 @@ define <8 x i32> @mgather_baseidx_sext_v8i16_v8i32(ptr %base, <8 x i16> %idxs, <
; RV64ZVE32F-NEXT: slli a2, a2, 2
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: lw a2, 0(a2)
-; RV64ZVE32F-NEXT: vsetivli zero, 8, e32, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 8, e32, m1, tu, ma
; RV64ZVE32F-NEXT: vmv.s.x v10, a2
; RV64ZVE32F-NEXT: .LBB39_2: # %else
; RV64ZVE32F-NEXT: andi a2, a1, 2
@@ -3145,9 +3145,9 @@ define <8 x i32> @mgather_baseidx_sext_v8i16_v8i32(ptr %base, <8 x i16> %idxs, <
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: lw a2, 0(a2)
; RV64ZVE32F-NEXT: vsetvli zero, zero, e32, m1, ta, ma
-; RV64ZVE32F-NEXT: vmv.s.x v12, a2
-; RV64ZVE32F-NEXT: vsetivli zero, 2, e32, m2, tu, ma
-; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 1
+; RV64ZVE32F-NEXT: vmv.s.x v9, a2
+; RV64ZVE32F-NEXT: vsetivli zero, 2, e32, m1, tu, ma
+; RV64ZVE32F-NEXT: vslideup.vi v10, v9, 1
; RV64ZVE32F-NEXT: .LBB39_4: # %else2
; RV64ZVE32F-NEXT: vsetivli zero, 4, e16, m1, ta, ma
; RV64ZVE32F-NEXT: vslidedown.vi v9, v8, 4
@@ -3193,7 +3193,7 @@ define <8 x i32> @mgather_baseidx_sext_v8i16_v8i32(ptr %base, <8 x i16> %idxs, <
; RV64ZVE32F-NEXT: lw a2, 0(a2)
; RV64ZVE32F-NEXT: vsetvli zero, zero, e32, m1, ta, ma
; RV64ZVE32F-NEXT: vmv.s.x v12, a2
-; RV64ZVE32F-NEXT: vsetivli zero, 3, e32, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 3, e32, m1, tu, ma
; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 2
; RV64ZVE32F-NEXT: andi a2, a1, 8
; RV64ZVE32F-NEXT: beqz a2, .LBB39_6
@@ -3205,9 +3205,9 @@ define <8 x i32> @mgather_baseidx_sext_v8i16_v8i32(ptr %base, <8 x i16> %idxs, <
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: lw a2, 0(a2)
; RV64ZVE32F-NEXT: vsetvli zero, zero, e32, m1, ta, ma
-; RV64ZVE32F-NEXT: vmv.s.x v12, a2
-; RV64ZVE32F-NEXT: vsetivli zero, 4, e32, m2, tu, ma
-; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 3
+; RV64ZVE32F-NEXT: vmv.s.x v8, a2
+; RV64ZVE32F-NEXT: vsetivli zero, 4, e32, m1, tu, ma
+; RV64ZVE32F-NEXT: vslideup.vi v10, v8, 3
; RV64ZVE32F-NEXT: andi a2, a1, 16
; RV64ZVE32F-NEXT: beqz a2, .LBB39_7
; RV64ZVE32F-NEXT: .LBB39_14: # %cond.load10
@@ -3287,7 +3287,7 @@ define <8 x i32> @mgather_baseidx_zext_v8i16_v8i32(ptr %base, <8 x i16> %idxs, <
; RV64ZVE32F-NEXT: slli a3, a3, 2
; RV64ZVE32F-NEXT: add a3, a0, a3
; RV64ZVE32F-NEXT: lw a3, 0(a3)
-; RV64ZVE32F-NEXT: vsetivli zero, 8, e32, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 8, e32, m1, tu, ma
; RV64ZVE32F-NEXT: vmv.s.x v10, a3
; RV64ZVE32F-NEXT: .LBB40_2: # %else
; RV64ZVE32F-NEXT: andi a3, a2, 2
@@ -3301,9 +3301,9 @@ define <8 x i32> @mgather_baseidx_zext_v8i16_v8i32(ptr %base, <8 x i16> %idxs, <
; RV64ZVE32F-NEXT: add a3, a0, a3
; RV64ZVE32F-NEXT: lw a3, 0(a3)
; RV64ZVE32F-NEXT: vsetvli zero, zero, e32, m1, ta, ma
-; RV64ZVE32F-NEXT: vmv.s.x v12, a3
-; RV64ZVE32F-NEXT: vsetivli zero, 2, e32, m2, tu, ma
-; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 1
+; RV64ZVE32F-NEXT: vmv.s.x v9, a3
+; RV64ZVE32F-NEXT: vsetivli zero, 2, e32, m1, tu, ma
+; RV64ZVE32F-NEXT: vslideup.vi v10, v9, 1
; RV64ZVE32F-NEXT: .LBB40_4: # %else2
; RV64ZVE32F-NEXT: vsetivli zero, 4, e16, m1, ta, ma
; RV64ZVE32F-NEXT: vslidedown.vi v9, v8, 4
@@ -3351,7 +3351,7 @@ define <8 x i32> @mgather_baseidx_zext_v8i16_v8i32(ptr %base, <8 x i16> %idxs, <
; RV64ZVE32F-NEXT: lw a3, 0(a3)
; RV64ZVE32F-NEXT: vsetvli zero, zero, e32, m1, ta, ma
; RV64ZVE32F-NEXT: vmv.s.x v12, a3
-; RV64ZVE32F-NEXT: vsetivli zero, 3, e32, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 3, e32, m1, tu, ma
; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 2
; RV64ZVE32F-NEXT: andi a3, a2, 8
; RV64ZVE32F-NEXT: beqz a3, .LBB40_6
@@ -3364,9 +3364,9 @@ define <8 x i32> @mgather_baseidx_zext_v8i16_v8i32(ptr %base, <8 x i16> %idxs, <
; RV64ZVE32F-NEXT: add a3, a0, a3
; RV64ZVE32F-NEXT: lw a3, 0(a3)
; RV64ZVE32F-NEXT: vsetvli zero, zero, e32, m1, ta, ma
-; RV64ZVE32F-NEXT: vmv.s.x v12, a3
-; RV64ZVE32F-NEXT: vsetivli zero, 4, e32, m2, tu, ma
-; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 3
+; RV64ZVE32F-NEXT: vmv.s.x v8, a3
+; RV64ZVE32F-NEXT: vsetivli zero, 4, e32, m1, tu, ma
+; RV64ZVE32F-NEXT: vslideup.vi v10, v8, 3
; RV64ZVE32F-NEXT: andi a3, a2, 16
; RV64ZVE32F-NEXT: beqz a3, .LBB40_7
; RV64ZVE32F-NEXT: .LBB40_14: # %cond.load10
@@ -3440,7 +3440,7 @@ define <8 x i32> @mgather_baseidx_v8i32(ptr %base, <8 x i32> %idxs, <8 x i1> %m,
; RV64ZVE32F-NEXT: andi a2, a1, 1
; RV64ZVE32F-NEXT: beqz a2, .LBB41_2
; RV64ZVE32F-NEXT: # %bb.1: # %cond.load
-; RV64ZVE32F-NEXT: vsetivli zero, 8, e32, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 8, e32, m1, tu, ma
; RV64ZVE32F-NEXT: vmv.x.s a2, v8
; RV64ZVE32F-NEXT: slli a2, a2, 2
; RV64ZVE32F-NEXT: add a2, a0, a2
@@ -3457,7 +3457,7 @@ define <8 x i32> @mgather_baseidx_v8i32(ptr %base, <8 x i32> %idxs, <8 x i1> %m,
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: lw a2, 0(a2)
; RV64ZVE32F-NEXT: vmv.s.x v12, a2
-; RV64ZVE32F-NEXT: vsetivli zero, 2, e32, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 2, e32, m1, tu, ma
; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 1
; RV64ZVE32F-NEXT: .LBB41_4: # %else2
; RV64ZVE32F-NEXT: vsetivli zero, 4, e32, m2, ta, ma
@@ -3501,9 +3501,9 @@ define <8 x i32> @mgather_baseidx_v8i32(ptr %base, <8 x i32> %idxs, <8 x i1> %m,
; RV64ZVE32F-NEXT: slli a2, a2, 2
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: lw a2, 0(a2)
-; RV64ZVE32F-NEXT: vmv.s.x v14, a2
-; RV64ZVE32F-NEXT: vsetivli zero, 3, e32, m2, tu, ma
-; RV64ZVE32F-NEXT: vslideup.vi v10, v14, 2
+; RV64ZVE32F-NEXT: vmv.s.x v9, a2
+; RV64ZVE32F-NEXT: vsetivli zero, 3, e32, m1, tu, ma
+; RV64ZVE32F-NEXT: vslideup.vi v10, v9, 2
; RV64ZVE32F-NEXT: andi a2, a1, 8
; RV64ZVE32F-NEXT: beqz a2, .LBB41_6
; RV64ZVE32F-NEXT: .LBB41_13: # %cond.load7
@@ -3514,7 +3514,7 @@ define <8 x i32> @mgather_baseidx_v8i32(ptr %base, <8 x i32> %idxs, <8 x i1> %m,
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: lw a2, 0(a2)
; RV64ZVE32F-NEXT: vmv.s.x v8, a2
-; RV64ZVE32F-NEXT: vsetivli zero, 4, e32, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 4, e32, m1, tu, ma
; RV64ZVE32F-NEXT: vslideup.vi v10, v8, 3
; RV64ZVE32F-NEXT: andi a2, a1, 16
; RV64ZVE32F-NEXT: beqz a2, .LBB41_7
@@ -8229,14 +8229,14 @@ define <8 x float> @mgather_v8f32(<8 x ptr> %ptrs, <8 x i1> %m, <8 x float> %pas
; RV64ZVE32F-NEXT: .LBB73_9: # %cond.load
; RV64ZVE32F-NEXT: ld a2, 0(a0)
; RV64ZVE32F-NEXT: flw fa5, 0(a2)
-; RV64ZVE32F-NEXT: vsetivli zero, 8, e32, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 8, e32, m1, tu, ma
; RV64ZVE32F-NEXT: vfmv.s.f v8, fa5
; RV64ZVE32F-NEXT: andi a2, a1, 2
; RV64ZVE32F-NEXT: beqz a2, .LBB73_2
; RV64ZVE32F-NEXT: .LBB73_10: # %cond.load1
; RV64ZVE32F-NEXT: ld a2, 8(a0)
; RV64ZVE32F-NEXT: flw fa5, 0(a2)
-; RV64ZVE32F-NEXT: vsetivli zero, 2, e32, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 2, e32, m1, tu, ma
; RV64ZVE32F-NEXT: vfmv.s.f v10, fa5
; RV64ZVE32F-NEXT: vslideup.vi v8, v10, 1
; RV64ZVE32F-NEXT: andi a2, a1, 4
@@ -8244,7 +8244,7 @@ define <8 x float> @mgather_v8f32(<8 x ptr> %ptrs, <8 x i1> %m, <8 x float> %pas
; RV64ZVE32F-NEXT: .LBB73_11: # %cond.load4
; RV64ZVE32F-NEXT: ld a2, 16(a0)
; RV64ZVE32F-NEXT: flw fa5, 0(a2)
-; RV64ZVE32F-NEXT: vsetivli zero, 3, e32, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 3, e32, m1, tu, ma
; RV64ZVE32F-NEXT: vfmv.s.f v10, fa5
; RV64ZVE32F-NEXT: vslideup.vi v8, v10, 2
; RV64ZVE32F-NEXT: andi a2, a1, 8
@@ -8252,7 +8252,7 @@ define <8 x float> @mgather_v8f32(<8 x ptr> %ptrs, <8 x i1> %m, <8 x float> %pas
; RV64ZVE32F-NEXT: .LBB73_12: # %cond.load7
; RV64ZVE32F-NEXT: ld a2, 24(a0)
; RV64ZVE32F-NEXT: flw fa5, 0(a2)
-; RV64ZVE32F-NEXT: vsetivli zero, 4, e32, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 4, e32, m1, tu, ma
; RV64ZVE32F-NEXT: vfmv.s.f v10, fa5
; RV64ZVE32F-NEXT: vslideup.vi v8, v10, 3
; RV64ZVE32F-NEXT: andi a2, a1, 16
@@ -8323,7 +8323,7 @@ define <8 x float> @mgather_baseidx_v8i8_v8f32(ptr %base, <8 x i8> %idxs, <8 x i
; RV64ZVE32F-NEXT: slli a2, a2, 2
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: flw fa5, 0(a2)
-; RV64ZVE32F-NEXT: vsetivli zero, 8, e32, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 8, e32, m1, tu, ma
; RV64ZVE32F-NEXT: vfmv.s.f v10, fa5
; RV64ZVE32F-NEXT: .LBB74_2: # %else
; RV64ZVE32F-NEXT: andi a2, a1, 2
@@ -8336,9 +8336,9 @@ define <8 x float> @mgather_baseidx_v8i8_v8f32(ptr %base, <8 x i8> %idxs, <8 x i
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: flw fa5, 0(a2)
; RV64ZVE32F-NEXT: vsetvli zero, zero, e32, m1, ta, ma
-; RV64ZVE32F-NEXT: vfmv.s.f v12, fa5
-; RV64ZVE32F-NEXT: vsetivli zero, 2, e32, m2, tu, ma
-; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 1
+; RV64ZVE32F-NEXT: vfmv.s.f v9, fa5
+; RV64ZVE32F-NEXT: vsetivli zero, 2, e32, m1, tu, ma
+; RV64ZVE32F-NEXT: vslideup.vi v10, v9, 1
; RV64ZVE32F-NEXT: .LBB74_4: # %else2
; RV64ZVE32F-NEXT: vsetivli zero, 4, e8, mf2, ta, ma
; RV64ZVE32F-NEXT: vslidedown.vi v9, v8, 4
@@ -8384,7 +8384,7 @@ define <8 x float> @mgather_baseidx_v8i8_v8f32(ptr %base, <8 x i8> %idxs, <8 x i
; RV64ZVE32F-NEXT: flw fa5, 0(a2)
; RV64ZVE32F-NEXT: vsetvli zero, zero, e32, m1, ta, ma
; RV64ZVE32F-NEXT: vfmv.s.f v12, fa5
-; RV64ZVE32F-NEXT: vsetivli zero, 3, e32, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 3, e32, m1, tu, ma
; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 2
; RV64ZVE32F-NEXT: andi a2, a1, 8
; RV64ZVE32F-NEXT: beqz a2, .LBB74_6
@@ -8396,9 +8396,9 @@ define <8 x float> @mgather_baseidx_v8i8_v8f32(ptr %base, <8 x i8> %idxs, <8 x i
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: flw fa5, 0(a2)
; RV64ZVE32F-NEXT: vsetvli zero, zero, e32, m1, ta, ma
-; RV64ZVE32F-NEXT: vfmv.s.f v12, fa5
-; RV64ZVE32F-NEXT: vsetivli zero, 4, e32, m2, tu, ma
-; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 3
+; RV64ZVE32F-NEXT: vfmv.s.f v8, fa5
+; RV64ZVE32F-NEXT: vsetivli zero, 4, e32, m1, tu, ma
+; RV64ZVE32F-NEXT: vslideup.vi v10, v8, 3
; RV64ZVE32F-NEXT: andi a2, a1, 16
; RV64ZVE32F-NEXT: beqz a2, .LBB74_7
; RV64ZVE32F-NEXT: .LBB74_14: # %cond.load10
@@ -8473,7 +8473,7 @@ define <8 x float> @mgather_baseidx_sext_v8i8_v8f32(ptr %base, <8 x i8> %idxs, <
; RV64ZVE32F-NEXT: slli a2, a2, 2
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: flw fa5, 0(a2)
-; RV64ZVE32F-NEXT: vsetivli zero, 8, e32, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 8, e32, m1, tu, ma
; RV64ZVE32F-NEXT: vfmv.s.f v10, fa5
; RV64ZVE32F-NEXT: .LBB75_2: # %else
; RV64ZVE32F-NEXT: andi a2, a1, 2
@@ -8486,9 +8486,9 @@ define <8 x float> @mgather_baseidx_sext_v8i8_v8f32(ptr %base, <8 x i8> %idxs, <
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: flw fa5, 0(a2)
; RV64ZVE32F-NEXT: vsetvli zero, zero, e32, m1, ta, ma
-; RV64ZVE32F-NEXT: vfmv.s.f v12, fa5
-; RV64ZVE32F-NEXT: vsetivli zero, 2, e32, m2, tu, ma
-; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 1
+; RV64ZVE32F-NEXT: vfmv.s.f v9, fa5
+; RV64ZVE32F-NEXT: vsetivli zero, 2, e32, m1, tu, ma
+; RV64ZVE32F-NEXT: vslideup.vi v10, v9, 1
; RV64ZVE32F-NEXT: .LBB75_4: # %else2
; RV64ZVE32F-NEXT: vsetivli zero, 4, e8, mf2, ta, ma
; RV64ZVE32F-NEXT: vslidedown.vi v9, v8, 4
@@ -8534,7 +8534,7 @@ define <8 x float> @mgather_baseidx_sext_v8i8_v8f32(ptr %base, <8 x i8> %idxs, <
; RV64ZVE32F-NEXT: flw fa5, 0(a2)
; RV64ZVE32F-NEXT: vsetvli zero, zero, e32, m1, ta, ma
; RV64ZVE32F-NEXT: vfmv.s.f v12, fa5
-; RV64ZVE32F-NEXT: vsetivli zero, 3, e32, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 3, e32, m1, tu, ma
; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 2
; RV64ZVE32F-NEXT: andi a2, a1, 8
; RV64ZVE32F-NEXT: beqz a2, .LBB75_6
@@ -8546,9 +8546,9 @@ define <8 x float> @mgather_baseidx_sext_v8i8_v8f32(ptr %base, <8 x i8> %idxs, <
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: flw fa5, 0(a2)
; RV64ZVE32F-NEXT: vsetvli zero, zero, e32, m1, ta, ma
-; RV64ZVE32F-NEXT: vfmv.s.f v12, fa5
-; RV64ZVE32F-NEXT: vsetivli zero, 4, e32, m2, tu, ma
-; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 3
+; RV64ZVE32F-NEXT: vfmv.s.f v8, fa5
+; RV64ZVE32F-NEXT: vsetivli zero, 4, e32, m1, tu, ma
+; RV64ZVE32F-NEXT: vslideup.vi v10, v8, 3
; RV64ZVE32F-NEXT: andi a2, a1, 16
; RV64ZVE32F-NEXT: beqz a2, .LBB75_7
; RV64ZVE32F-NEXT: .LBB75_14: # %cond.load10
@@ -8625,7 +8625,7 @@ define <8 x float> @mgather_baseidx_zext_v8i8_v8f32(ptr %base, <8 x i8> %idxs, <
; RV64ZVE32F-NEXT: slli a2, a2, 2
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: flw fa5, 0(a2)
-; RV64ZVE32F-NEXT: vsetivli zero, 8, e32, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 8, e32, m1, tu, ma
; RV64ZVE32F-NEXT: vfmv.s.f v10, fa5
; RV64ZVE32F-NEXT: .LBB76_2: # %else
; RV64ZVE32F-NEXT: andi a2, a1, 2
@@ -8639,9 +8639,9 @@ define <8 x float> @mgather_baseidx_zext_v8i8_v8f32(ptr %base, <8 x i8> %idxs, <
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: flw fa5, 0(a2)
; RV64ZVE32F-NEXT: vsetvli zero, zero, e32, m1, ta, ma
-; RV64ZVE32F-NEXT: vfmv.s.f v12, fa5
-; RV64ZVE32F-NEXT: vsetivli zero, 2, e32, m2, tu, ma
-; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 1
+; RV64ZVE32F-NEXT: vfmv.s.f v9, fa5
+; RV64ZVE32F-NEXT: vsetivli zero, 2, e32, m1, tu, ma
+; RV64ZVE32F-NEXT: vslideup.vi v10, v9, 1
; RV64ZVE32F-NEXT: .LBB76_4: # %else2
; RV64ZVE32F-NEXT: vsetivli zero, 4, e8, mf2, ta, ma
; RV64ZVE32F-NEXT: vslidedown.vi v9, v8, 4
@@ -8689,7 +8689,7 @@ define <8 x float> @mgather_baseidx_zext_v8i8_v8f32(ptr %base, <8 x i8> %idxs, <
; RV64ZVE32F-NEXT: flw fa5, 0(a2)
; RV64ZVE32F-NEXT: vsetvli zero, zero, e32, m1, ta, ma
; RV64ZVE32F-NEXT: vfmv.s.f v12, fa5
-; RV64ZVE32F-NEXT: vsetivli zero, 3, e32, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 3, e32, m1, tu, ma
; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 2
; RV64ZVE32F-NEXT: andi a2, a1, 8
; RV64ZVE32F-NEXT: beqz a2, .LBB76_6
@@ -8703,7 +8703,7 @@ define <8 x float> @mgather_baseidx_zext_v8i8_v8f32(ptr %base, <8 x i8> %idxs, <
; RV64ZVE32F-NEXT: flw fa5, 0(a2)
; RV64ZVE32F-NEXT: vsetvli zero, zero, e32, m1, ta, ma
; RV64ZVE32F-NEXT: vfmv.s.f v8, fa5
-; RV64ZVE32F-NEXT: vsetivli zero, 4, e32, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 4, e32, m1, tu, ma
; RV64ZVE32F-NEXT: vslideup.vi v10, v8, 3
; RV64ZVE32F-NEXT: andi a2, a1, 16
; RV64ZVE32F-NEXT: beqz a2, .LBB76_7
@@ -8784,7 +8784,7 @@ define <8 x float> @mgather_baseidx_v8i16_v8f32(ptr %base, <8 x i16> %idxs, <8 x
; RV64ZVE32F-NEXT: slli a2, a2, 2
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: flw fa5, 0(a2)
-; RV64ZVE32F-NEXT: vsetivli zero, 8, e32, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 8, e32, m1, tu, ma
; RV64ZVE32F-NEXT: vfmv.s.f v10, fa5
; RV64ZVE32F-NEXT: .LBB77_2: # %else
; RV64ZVE32F-NEXT: andi a2, a1, 2
@@ -8797,9 +8797,9 @@ define <8 x float> @mgather_baseidx_v8i16_v8f32(ptr %base, <8 x i16> %idxs, <8 x
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: flw fa5, 0(a2)
; RV64ZVE32F-NEXT: vsetvli zero, zero, e32, m1, ta, ma
-; RV64ZVE32F-NEXT: vfmv.s.f v12, fa5
-; RV64ZVE32F-NEXT: vsetivli zero, 2, e32, m2, tu, ma
-; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 1
+; RV64ZVE32F-NEXT: vfmv.s.f v9, fa5
+; RV64ZVE32F-NEXT: vsetivli zero, 2, e32, m1, tu, ma
+; RV64ZVE32F-NEXT: vslideup.vi v10, v9, 1
; RV64ZVE32F-NEXT: .LBB77_4: # %else2
; RV64ZVE32F-NEXT: vsetivli zero, 4, e16, m1, ta, ma
; RV64ZVE32F-NEXT: vslidedown.vi v9, v8, 4
@@ -8845,7 +8845,7 @@ define <8 x float> @mgather_baseidx_v8i16_v8f32(ptr %base, <8 x i16> %idxs, <8 x
; RV64ZVE32F-NEXT: flw fa5, 0(a2)
; RV64ZVE32F-NEXT: vsetvli zero, zero, e32, m1, ta, ma
; RV64ZVE32F-NEXT: vfmv.s.f v12, fa5
-; RV64ZVE32F-NEXT: vsetivli zero, 3, e32, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 3, e32, m1, tu, ma
; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 2
; RV64ZVE32F-NEXT: andi a2, a1, 8
; RV64ZVE32F-NEXT: beqz a2, .LBB77_6
@@ -8857,9 +8857,9 @@ define <8 x float> @mgather_baseidx_v8i16_v8f32(ptr %base, <8 x i16> %idxs, <8 x
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: flw fa5, 0(a2)
; RV64ZVE32F-NEXT: vsetvli zero, zero, e32, m1, ta, ma
-; RV64ZVE32F-NEXT: vfmv.s.f v12, fa5
-; RV64ZVE32F-NEXT: vsetivli zero, 4, e32, m2, tu, ma
-; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 3
+; RV64ZVE32F-NEXT: vfmv.s.f v8, fa5
+; RV64ZVE32F-NEXT: vsetivli zero, 4, e32, m1, tu, ma
+; RV64ZVE32F-NEXT: vslideup.vi v10, v8, 3
; RV64ZVE32F-NEXT: andi a2, a1, 16
; RV64ZVE32F-NEXT: beqz a2, .LBB77_7
; RV64ZVE32F-NEXT: .LBB77_14: # %cond.load10
@@ -8935,7 +8935,7 @@ define <8 x float> @mgather_baseidx_sext_v8i16_v8f32(ptr %base, <8 x i16> %idxs,
; RV64ZVE32F-NEXT: slli a2, a2, 2
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: flw fa5, 0(a2)
-; RV64ZVE32F-NEXT: vsetivli zero, 8, e32, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 8, e32, m1, tu, ma
; RV64ZVE32F-NEXT: vfmv.s.f v10, fa5
; RV64ZVE32F-NEXT: .LBB78_2: # %else
; RV64ZVE32F-NEXT: andi a2, a1, 2
@@ -8948,9 +8948,9 @@ define <8 x float> @mgather_baseidx_sext_v8i16_v8f32(ptr %base, <8 x i16> %idxs,
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: flw fa5, 0(a2)
; RV64ZVE32F-NEXT: vsetvli zero, zero, e32, m1, ta, ma
-; RV64ZVE32F-NEXT: vfmv.s.f v12, fa5
-; RV64ZVE32F-NEXT: vsetivli zero, 2, e32, m2, tu, ma
-; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 1
+; RV64ZVE32F-NEXT: vfmv.s.f v9, fa5
+; RV64ZVE32F-NEXT: vsetivli zero, 2, e32, m1, tu, ma
+; RV64ZVE32F-NEXT: vslideup.vi v10, v9, 1
; RV64ZVE32F-NEXT: .LBB78_4: # %else2
; RV64ZVE32F-NEXT: vsetivli zero, 4, e16, m1, ta, ma
; RV64ZVE32F-NEXT: vslidedown.vi v9, v8, 4
@@ -8996,7 +8996,7 @@ define <8 x float> @mgather_baseidx_sext_v8i16_v8f32(ptr %base, <8 x i16> %idxs,
; RV64ZVE32F-NEXT: flw fa5, 0(a2)
; RV64ZVE32F-NEXT: vsetvli zero, zero, e32, m1, ta, ma
; RV64ZVE32F-NEXT: vfmv.s.f v12, fa5
-; RV64ZVE32F-NEXT: vsetivli zero, 3, e32, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 3, e32, m1, tu, ma
; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 2
; RV64ZVE32F-NEXT: andi a2, a1, 8
; RV64ZVE32F-NEXT: beqz a2, .LBB78_6
@@ -9008,9 +9008,9 @@ define <8 x float> @mgather_baseidx_sext_v8i16_v8f32(ptr %base, <8 x i16> %idxs,
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: flw fa5, 0(a2)
; RV64ZVE32F-NEXT: vsetvli zero, zero, e32, m1, ta, ma
-; RV64ZVE32F-NEXT: vfmv.s.f v12, fa5
-; RV64ZVE32F-NEXT: vsetivli zero, 4, e32, m2, tu, ma
-; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 3
+; RV64ZVE32F-NEXT: vfmv.s.f v8, fa5
+; RV64ZVE32F-NEXT: vsetivli zero, 4, e32, m1, tu, ma
+; RV64ZVE32F-NEXT: vslideup.vi v10, v8, 3
; RV64ZVE32F-NEXT: andi a2, a1, 16
; RV64ZVE32F-NEXT: beqz a2, .LBB78_7
; RV64ZVE32F-NEXT: .LBB78_14: # %cond.load10
@@ -9090,7 +9090,7 @@ define <8 x float> @mgather_baseidx_zext_v8i16_v8f32(ptr %base, <8 x i16> %idxs,
; RV64ZVE32F-NEXT: slli a3, a3, 2
; RV64ZVE32F-NEXT: add a3, a0, a3
; RV64ZVE32F-NEXT: flw fa5, 0(a3)
-; RV64ZVE32F-NEXT: vsetivli zero, 8, e32, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 8, e32, m1, tu, ma
; RV64ZVE32F-NEXT: vfmv.s.f v10, fa5
; RV64ZVE32F-NEXT: .LBB79_2: # %else
; RV64ZVE32F-NEXT: andi a3, a2, 2
@@ -9104,9 +9104,9 @@ define <8 x float> @mgather_baseidx_zext_v8i16_v8f32(ptr %base, <8 x i16> %idxs,
; RV64ZVE32F-NEXT: add a3, a0, a3
; RV64ZVE32F-NEXT: flw fa5, 0(a3)
; RV64ZVE32F-NEXT: vsetvli zero, zero, e32, m1, ta, ma
-; RV64ZVE32F-NEXT: vfmv.s.f v12, fa5
-; RV64ZVE32F-NEXT: vsetivli zero, 2, e32, m2, tu, ma
-; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 1
+; RV64ZVE32F-NEXT: vfmv.s.f v9, fa5
+; RV64ZVE32F-NEXT: vsetivli zero, 2, e32, m1, tu, ma
+; RV64ZVE32F-NEXT: vslideup.vi v10, v9, 1
; RV64ZVE32F-NEXT: .LBB79_4: # %else2
; RV64ZVE32F-NEXT: vsetivli zero, 4, e16, m1, ta, ma
; RV64ZVE32F-NEXT: vslidedown.vi v9, v8, 4
@@ -9154,7 +9154,7 @@ define <8 x float> @mgather_baseidx_zext_v8i16_v8f32(ptr %base, <8 x i16> %idxs,
; RV64ZVE32F-NEXT: flw fa5, 0(a3)
; RV64ZVE32F-NEXT: vsetvli zero, zero, e32, m1, ta, ma
; RV64ZVE32F-NEXT: vfmv.s.f v12, fa5
-; RV64ZVE32F-NEXT: vsetivli zero, 3, e32, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 3, e32, m1, tu, ma
; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 2
; RV64ZVE32F-NEXT: andi a3, a2, 8
; RV64ZVE32F-NEXT: beqz a3, .LBB79_6
@@ -9167,9 +9167,9 @@ define <8 x float> @mgather_baseidx_zext_v8i16_v8f32(ptr %base, <8 x i16> %idxs,
; RV64ZVE32F-NEXT: add a3, a0, a3
; RV64ZVE32F-NEXT: flw fa5, 0(a3)
; RV64ZVE32F-NEXT: vsetvli zero, zero, e32, m1, ta, ma
-; RV64ZVE32F-NEXT: vfmv.s.f v12, fa5
-; RV64ZVE32F-NEXT: vsetivli zero, 4, e32, m2, tu, ma
-; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 3
+; RV64ZVE32F-NEXT: vfmv.s.f v8, fa5
+; RV64ZVE32F-NEXT: vsetivli zero, 4, e32, m1, tu, ma
+; RV64ZVE32F-NEXT: vslideup.vi v10, v8, 3
; RV64ZVE32F-NEXT: andi a3, a2, 16
; RV64ZVE32F-NEXT: beqz a3, .LBB79_7
; RV64ZVE32F-NEXT: .LBB79_14: # %cond.load10
@@ -9243,7 +9243,7 @@ define <8 x float> @mgather_baseidx_v8f32(ptr %base, <8 x i32> %idxs, <8 x i1> %
; RV64ZVE32F-NEXT: andi a2, a1, 1
; RV64ZVE32F-NEXT: beqz a2, .LBB80_2
; RV64ZVE32F-NEXT: # %bb.1: # %cond.load
-; RV64ZVE32F-NEXT: vsetivli zero, 8, e32, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 8, e32, m1, tu, ma
; RV64ZVE32F-NEXT: vmv.x.s a2, v8
; RV64ZVE32F-NEXT: slli a2, a2, 2
; RV64ZVE32F-NEXT: add a2, a0, a2
@@ -9260,7 +9260,7 @@ define <8 x float> @mgather_baseidx_v8f32(ptr %base, <8 x i32> %idxs, <8 x i1> %
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: flw fa5, 0(a2)
; RV64ZVE32F-NEXT: vfmv.s.f v12, fa5
-; RV64ZVE32F-NEXT: vsetivli zero, 2, e32, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 2, e32, m1, tu, ma
; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 1
; RV64ZVE32F-NEXT: .LBB80_4: # %else2
; RV64ZVE32F-NEXT: vsetivli zero, 4, e32, m2, ta, ma
@@ -9304,9 +9304,9 @@ define <8 x float> @mgather_baseidx_v8f32(ptr %base, <8 x i32> %idxs, <8 x i1> %
; RV64ZVE32F-NEXT: slli a2, a2, 2
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: flw fa5, 0(a2)
-; RV64ZVE32F-NEXT: vfmv.s.f v14, fa5
-; RV64ZVE32F-NEXT: vsetivli zero, 3, e32, m2, tu, ma
-; RV64ZVE32F-NEXT: vslideup.vi v10, v14, 2
+; RV64ZVE32F-NEXT: vfmv.s.f v9, fa5
+; RV64ZVE32F-NEXT: vsetivli zero, 3, e32, m1, tu, ma
+; RV64ZVE32F-NEXT: vslideup.vi v10, v9, 2
; RV64ZVE32F-NEXT: andi a2, a1, 8
; RV64ZVE32F-NEXT: beqz a2, .LBB80_6
; RV64ZVE32F-NEXT: .LBB80_13: # %cond.load7
@@ -9317,7 +9317,7 @@ define <8 x float> @mgather_baseidx_v8f32(ptr %base, <8 x i32> %idxs, <8 x i1> %
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: flw fa5, 0(a2)
; RV64ZVE32F-NEXT: vfmv.s.f v8, fa5
-; RV64ZVE32F-NEXT: vsetivli zero, 4, e32, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 4, e32, m1, tu, ma
; RV64ZVE32F-NEXT: vslideup.vi v10, v8, 3
; RV64ZVE32F-NEXT: andi a2, a1, 16
; RV64ZVE32F-NEXT: beqz a2, .LBB80_7
@@ -12382,7 +12382,7 @@ define <32 x i8> @mgather_baseidx_v32i8(ptr %base, <32 x i8> %idxs, <32 x i1> %m
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: lbu a2, 0(a2)
; RV64ZVE32F-NEXT: li a3, 32
-; RV64ZVE32F-NEXT: vsetvli zero, a3, e8, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetvli zero, a3, e8, m1, tu, ma
; RV64ZVE32F-NEXT: vmv.s.x v10, a2
; RV64ZVE32F-NEXT: .LBB98_2: # %else
; RV64ZVE32F-NEXT: andi a2, a1, 2
@@ -12395,7 +12395,7 @@ define <32 x i8> @mgather_baseidx_v32i8(ptr %base, <32 x i8> %idxs, <32 x i1> %m
; RV64ZVE32F-NEXT: lbu a2, 0(a2)
; RV64ZVE32F-NEXT: li a3, 32
; RV64ZVE32F-NEXT: vmv.s.x v12, a2
-; RV64ZVE32F-NEXT: vsetivli zero, 2, e8, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 2, e8, m1, tu, ma
; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 1
; RV64ZVE32F-NEXT: .LBB98_4: # %else2
; RV64ZVE32F-NEXT: vsetivli zero, 4, e8, mf2, ta, ma
@@ -12417,9 +12417,9 @@ define <32 x i8> @mgather_baseidx_v32i8(ptr %base, <32 x i8> %idxs, <32 x i1> %m
; RV64ZVE32F-NEXT: lbu a2, 0(a2)
; RV64ZVE32F-NEXT: li a3, 32
; RV64ZVE32F-NEXT: vsetvli zero, a3, e8, m1, ta, ma
-; RV64ZVE32F-NEXT: vmv.s.x v14, a2
-; RV64ZVE32F-NEXT: vsetivli zero, 5, e8, m2, tu, ma
-; RV64ZVE32F-NEXT: vslideup.vi v10, v14, 4
+; RV64ZVE32F-NEXT: vmv.s.x v12, a2
+; RV64ZVE32F-NEXT: vsetivli zero, 5, e8, m1, tu, ma
+; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 4
; RV64ZVE32F-NEXT: .LBB98_8: # %else11
; RV64ZVE32F-NEXT: vsetivli zero, 8, e8, m1, ta, ma
; RV64ZVE32F-NEXT: andi a2, a1, 32
@@ -12433,7 +12433,7 @@ define <32 x i8> @mgather_baseidx_v32i8(ptr %base, <32 x i8> %idxs, <32 x i1> %m
; RV64ZVE32F-NEXT: lbu a2, 0(a2)
; RV64ZVE32F-NEXT: li a3, 32
; RV64ZVE32F-NEXT: vmv.s.x v14, a2
-; RV64ZVE32F-NEXT: vsetivli zero, 6, e8, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 6, e8, m1, tu, ma
; RV64ZVE32F-NEXT: vslideup.vi v10, v14, 5
; RV64ZVE32F-NEXT: .LBB98_10: # %else14
; RV64ZVE32F-NEXT: vsetivli zero, 2, e8, mf4, ta, ma
@@ -12456,9 +12456,9 @@ define <32 x i8> @mgather_baseidx_v32i8(ptr %base, <32 x i8> %idxs, <32 x i1> %m
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: lbu a2, 0(a2)
; RV64ZVE32F-NEXT: li a3, 32
-; RV64ZVE32F-NEXT: vmv.s.x v14, a2
-; RV64ZVE32F-NEXT: vsetivli zero, 10, e8, m2, tu, ma
-; RV64ZVE32F-NEXT: vslideup.vi v10, v14, 9
+; RV64ZVE32F-NEXT: vmv.s.x v13, a2
+; RV64ZVE32F-NEXT: vsetivli zero, 10, e8, m1, tu, ma
+; RV64ZVE32F-NEXT: vslideup.vi v10, v13, 9
; RV64ZVE32F-NEXT: .LBB98_15: # %else26
; RV64ZVE32F-NEXT: vsetivli zero, 4, e8, mf2, ta, ma
; RV64ZVE32F-NEXT: vslidedown.vi v13, v12, 4
@@ -12472,7 +12472,7 @@ define <32 x i8> @mgather_baseidx_v32i8(ptr %base, <32 x i8> %idxs, <32 x i1> %m
; RV64ZVE32F-NEXT: lbu a2, 0(a2)
; RV64ZVE32F-NEXT: li a3, 32
; RV64ZVE32F-NEXT: vmv.s.x v14, a2
-; RV64ZVE32F-NEXT: vsetivli zero, 11, e8, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 11, e8, m1, tu, ma
; RV64ZVE32F-NEXT: vslideup.vi v10, v14, 10
; RV64ZVE32F-NEXT: .LBB98_17: # %else29
; RV64ZVE32F-NEXT: slli a2, a1, 52
@@ -12484,9 +12484,9 @@ define <32 x i8> @mgather_baseidx_v32i8(ptr %base, <32 x i8> %idxs, <32 x i1> %m
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: lbu a2, 0(a2)
; RV64ZVE32F-NEXT: li a3, 32
-; RV64ZVE32F-NEXT: vmv.s.x v14, a2
-; RV64ZVE32F-NEXT: vsetivli zero, 12, e8, m2, tu, ma
-; RV64ZVE32F-NEXT: vslideup.vi v10, v14, 11
+; RV64ZVE32F-NEXT: vmv.s.x v12, a2
+; RV64ZVE32F-NEXT: vsetivli zero, 12, e8, m1, tu, ma
+; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 11
; RV64ZVE32F-NEXT: .LBB98_19: # %else32
; RV64ZVE32F-NEXT: vsetivli zero, 16, e8, m2, ta, ma
; RV64ZVE32F-NEXT: slli a2, a1, 51
@@ -12497,9 +12497,9 @@ define <32 x i8> @mgather_baseidx_v32i8(ptr %base, <32 x i8> %idxs, <32 x i1> %m
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: lbu a2, 0(a2)
; RV64ZVE32F-NEXT: li a3, 32
-; RV64ZVE32F-NEXT: vmv.s.x v14, a2
-; RV64ZVE32F-NEXT: vsetivli zero, 13, e8, m2, tu, ma
-; RV64ZVE32F-NEXT: vslideup.vi v10, v14, 12
+; RV64ZVE32F-NEXT: vmv.s.x v9, a2
+; RV64ZVE32F-NEXT: vsetivli zero, 13, e8, m1, tu, ma
+; RV64ZVE32F-NEXT: vslideup.vi v10, v9, 12
; RV64ZVE32F-NEXT: .LBB98_21: # %else35
; RV64ZVE32F-NEXT: slli a2, a1, 50
; RV64ZVE32F-NEXT: bgez a2, .LBB98_23
@@ -12510,9 +12510,9 @@ define <32 x i8> @mgather_baseidx_v32i8(ptr %base, <32 x i8> %idxs, <32 x i1> %m
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: lbu a2, 0(a2)
; RV64ZVE32F-NEXT: li a3, 32
-; RV64ZVE32F-NEXT: vmv.s.x v14, a2
-; RV64ZVE32F-NEXT: vsetivli zero, 14, e8, m2, tu, ma
-; RV64ZVE32F-NEXT: vslideup.vi v10, v14, 13
+; RV64ZVE32F-NEXT: vmv.s.x v9, a2
+; RV64ZVE32F-NEXT: vsetivli zero, 14, e8, m1, tu, ma
+; RV64ZVE32F-NEXT: vslideup.vi v10, v9, 13
; RV64ZVE32F-NEXT: .LBB98_23: # %else38
; RV64ZVE32F-NEXT: vsetivli zero, 2, e8, mf4, ta, ma
; RV64ZVE32F-NEXT: slli a2, a1, 49
@@ -12661,7 +12661,7 @@ define <32 x i8> @mgather_baseidx_v32i8(ptr %base, <32 x i8> %idxs, <32 x i1> %m
; RV64ZVE32F-NEXT: lbu a2, 0(a2)
; RV64ZVE32F-NEXT: li a3, 32
; RV64ZVE32F-NEXT: vmv.s.x v14, a2
-; RV64ZVE32F-NEXT: vsetivli zero, 3, e8, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 3, e8, m1, tu, ma
; RV64ZVE32F-NEXT: vslideup.vi v10, v14, 2
; RV64ZVE32F-NEXT: andi a2, a1, 8
; RV64ZVE32F-NEXT: beqz a2, .LBB98_6
@@ -12672,9 +12672,9 @@ define <32 x i8> @mgather_baseidx_v32i8(ptr %base, <32 x i8> %idxs, <32 x i1> %m
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: lbu a2, 0(a2)
; RV64ZVE32F-NEXT: li a3, 32
-; RV64ZVE32F-NEXT: vmv.s.x v14, a2
-; RV64ZVE32F-NEXT: vsetivli zero, 4, e8, m2, tu, ma
-; RV64ZVE32F-NEXT: vslideup.vi v10, v14, 3
+; RV64ZVE32F-NEXT: vmv.s.x v12, a2
+; RV64ZVE32F-NEXT: vsetivli zero, 4, e8, m1, tu, ma
+; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 3
; RV64ZVE32F-NEXT: andi a2, a1, 16
; RV64ZVE32F-NEXT: bnez a2, .LBB98_7
; RV64ZVE32F-NEXT: j .LBB98_8
@@ -12684,7 +12684,7 @@ define <32 x i8> @mgather_baseidx_v32i8(ptr %base, <32 x i8> %idxs, <32 x i1> %m
; RV64ZVE32F-NEXT: lbu a2, 0(a2)
; RV64ZVE32F-NEXT: li a3, 32
; RV64ZVE32F-NEXT: vmv.s.x v14, a2
-; RV64ZVE32F-NEXT: vsetivli zero, 7, e8, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 7, e8, m1, tu, ma
; RV64ZVE32F-NEXT: vslideup.vi v10, v14, 6
; RV64ZVE32F-NEXT: andi a2, a1, 128
; RV64ZVE32F-NEXT: beqz a2, .LBB98_12
@@ -12695,9 +12695,9 @@ define <32 x i8> @mgather_baseidx_v32i8(ptr %base, <32 x i8> %idxs, <32 x i1> %m
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: lbu a2, 0(a2)
; RV64ZVE32F-NEXT: li a3, 32
-; RV64ZVE32F-NEXT: vmv.s.x v14, a2
-; RV64ZVE32F-NEXT: vsetivli zero, 8, e8, m2, tu, ma
-; RV64ZVE32F-NEXT: vslideup.vi v10, v14, 7
+; RV64ZVE32F-NEXT: vmv.s.x v13, a2
+; RV64ZVE32F-NEXT: vsetivli zero, 8, e8, m1, tu, ma
+; RV64ZVE32F-NEXT: vslideup.vi v10, v13, 7
; RV64ZVE32F-NEXT: andi a2, a1, 256
; RV64ZVE32F-NEXT: beqz a2, .LBB98_13
; RV64ZVE32F-NEXT: .LBB98_53: # %cond.load22
@@ -12707,9 +12707,9 @@ define <32 x i8> @mgather_baseidx_v32i8(ptr %base, <32 x i8> %idxs, <32 x i1> %m
; RV64ZVE32F-NEXT: lbu a2, 0(a2)
; RV64ZVE32F-NEXT: li a3, 32
; RV64ZVE32F-NEXT: vsetvli zero, a3, e8, m1, ta, ma
-; RV64ZVE32F-NEXT: vmv.s.x v14, a2
-; RV64ZVE32F-NEXT: vsetivli zero, 9, e8, m2, tu, ma
-; RV64ZVE32F-NEXT: vslideup.vi v10, v14, 8
+; RV64ZVE32F-NEXT: vmv.s.x v13, a2
+; RV64ZVE32F-NEXT: vsetivli zero, 9, e8, m1, tu, ma
+; RV64ZVE32F-NEXT: vslideup.vi v10, v13, 8
; RV64ZVE32F-NEXT: andi a2, a1, 512
; RV64ZVE32F-NEXT: bnez a2, .LBB98_14
; RV64ZVE32F-NEXT: j .LBB98_15
@@ -12719,7 +12719,7 @@ define <32 x i8> @mgather_baseidx_v32i8(ptr %base, <32 x i8> %idxs, <32 x i1> %m
; RV64ZVE32F-NEXT: lbu a2, 0(a2)
; RV64ZVE32F-NEXT: li a3, 32
; RV64ZVE32F-NEXT: vmv.s.x v12, a2
-; RV64ZVE32F-NEXT: vsetivli zero, 15, e8, m2, tu, ma
+; RV64ZVE32F-NEXT: vsetivli zero, 15, e8, m1, tu, ma
; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 14
; RV64ZVE32F-NEXT: slli a2, a1, 48
; RV64ZVE32F-NEXT: bgez a2, .LBB98_25
@@ -12730,9 +12730,9 @@ define <32 x i8> @mgather_baseidx_v32i8(ptr %base, <32 x i8> %idxs, <32 x i1> %m
; RV64ZVE32F-NEXT: add a2, a0, a2
; RV64ZVE32F-NEXT: lbu a2, 0(a2)
; RV64ZVE32F-NEXT: li a3, 32
-; RV64ZVE32F-NEXT: vmv.s.x v12, a2
-; RV64ZVE32F-NEXT: vsetivli zero, 16, e8, m2, tu, ma
-; RV64ZVE32F-NEXT: vslideup.vi v10, v12, 15
+; RV64ZVE32F-NEXT: vmv.s.x v9, a2
+; RV64ZVE32F-NEXT: vsetivli zero, 16, e8, m1, tu, ma
+; RV64ZVE32F-NEXT: vslideup.vi v10, v9, 15
; RV64ZVE32F-NEXT: slli a2, a1, 47
; RV64ZVE32F-NEXT: bgez a2, .LBB98_26
; RV64ZVE32F-NEXT: .LBB98_56: # %cond.load46
diff --git a/llvm/test/CodeGen/RISCV/rvv/insertelt-fp.ll b/llvm/test/CodeGen/RISCV/rvv/insertelt-fp.ll
index 4bd9f7befa52a1c..060e99691cb1381 100644
--- a/llvm/test/CodeGen/RISCV/rvv/insertelt-fp.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/insertelt-fp.ll
@@ -109,7 +109,7 @@ define <vscale x 4 x half> @insertelt_nxv4f16_idx(<vscale x 4 x half> %v, half %
define <vscale x 8 x half> @insertelt_nxv8f16_0(<vscale x 8 x half> %v, half %elt) {
; CHECK-LABEL: insertelt_nxv8f16_0:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a0, zero, e16, m2, tu, ma
+; CHECK-NEXT: vsetvli a0, zero, e16, m1, tu, ma
; CHECK-NEXT: vfmv.s.f v8, fa0
; CHECK-NEXT: ret
%r = insertelement <vscale x 8 x half> %v, half %elt, i32 0
@@ -119,7 +119,7 @@ define <vscale x 8 x half> @insertelt_nxv8f16_0(<vscale x 8 x half> %v, half %el
define <vscale x 8 x half> @insertelt_nxv8f16_imm(<vscale x 8 x half> %v, half %elt) {
; CHECK-LABEL: insertelt_nxv8f16_imm:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetivli zero, 4, e16, m2, tu, ma
+; CHECK-NEXT: vsetivli zero, 4, e16, m1, tu, ma
; CHECK-NEXT: vfmv.s.f v10, fa0
; CHECK-NEXT: vslideup.vi v8, v10, 3
; CHECK-NEXT: ret
@@ -143,7 +143,7 @@ define <vscale x 8 x half> @insertelt_nxv8f16_idx(<vscale x 8 x half> %v, half %
define <vscale x 16 x half> @insertelt_nxv16f16_0(<vscale x 16 x half> %v, half %elt) {
; CHECK-LABEL: insertelt_nxv16f16_0:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a0, zero, e16, m4, tu, ma
+; CHECK-NEXT: vsetvli a0, zero, e16, m1, tu, ma
; CHECK-NEXT: vfmv.s.f v8, fa0
; CHECK-NEXT: ret
%r = insertelement <vscale x 16 x half> %v, half %elt, i32 0
@@ -153,7 +153,7 @@ define <vscale x 16 x half> @insertelt_nxv16f16_0(<vscale x 16 x half> %v, half
define <vscale x 16 x half> @insertelt_nxv16f16_imm(<vscale x 16 x half> %v, half %elt) {
; CHECK-LABEL: insertelt_nxv16f16_imm:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetivli zero, 4, e16, m4, tu, ma
+; CHECK-NEXT: vsetivli zero, 4, e16, m1, tu, ma
; CHECK-NEXT: vfmv.s.f v12, fa0
; CHECK-NEXT: vslideup.vi v8, v12, 3
; CHECK-NEXT: ret
@@ -177,7 +177,7 @@ define <vscale x 16 x half> @insertelt_nxv16f16_idx(<vscale x 16 x half> %v, hal
define <vscale x 32 x half> @insertelt_nxv32f16_0(<vscale x 32 x half> %v, half %elt) {
; CHECK-LABEL: insertelt_nxv32f16_0:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a0, zero, e16, m8, tu, ma
+; CHECK-NEXT: vsetvli a0, zero, e16, m1, tu, ma
; CHECK-NEXT: vfmv.s.f v8, fa0
; CHECK-NEXT: ret
%r = insertelement <vscale x 32 x half> %v, half %elt, i32 0
@@ -187,7 +187,7 @@ define <vscale x 32 x half> @insertelt_nxv32f16_0(<vscale x 32 x half> %v, half
define <vscale x 32 x half> @insertelt_nxv32f16_imm(<vscale x 32 x half> %v, half %elt) {
; CHECK-LABEL: insertelt_nxv32f16_imm:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetivli zero, 4, e16, m8, tu, ma
+; CHECK-NEXT: vsetivli zero, 4, e16, m1, tu, ma
; CHECK-NEXT: vfmv.s.f v16, fa0
; CHECK-NEXT: vslideup.vi v8, v16, 3
; CHECK-NEXT: ret
@@ -279,7 +279,7 @@ define <vscale x 2 x float> @insertelt_nxv2f32_idx(<vscale x 2 x float> %v, floa
define <vscale x 4 x float> @insertelt_nxv4f32_0(<vscale x 4 x float> %v, float %elt) {
; CHECK-LABEL: insertelt_nxv4f32_0:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a0, zero, e32, m2, tu, ma
+; CHECK-NEXT: vsetvli a0, zero, e32, m1, tu, ma
; CHECK-NEXT: vfmv.s.f v8, fa0
; CHECK-NEXT: ret
%r = insertelement <vscale x 4 x float> %v, float %elt, i32 0
@@ -289,7 +289,7 @@ define <vscale x 4 x float> @insertelt_nxv4f32_0(<vscale x 4 x float> %v, float
define <vscale x 4 x float> @insertelt_nxv4f32_imm(<vscale x 4 x float> %v, float %elt) {
; CHECK-LABEL: insertelt_nxv4f32_imm:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetivli zero, 4, e32, m2, tu, ma
+; CHECK-NEXT: vsetivli zero, 4, e32, m1, tu, ma
; CHECK-NEXT: vfmv.s.f v10, fa0
; CHECK-NEXT: vslideup.vi v8, v10, 3
; CHECK-NEXT: ret
@@ -313,7 +313,7 @@ define <vscale x 4 x float> @insertelt_nxv4f32_idx(<vscale x 4 x float> %v, floa
define <vscale x 8 x float> @insertelt_nxv8f32_0(<vscale x 8 x float> %v, float %elt) {
; CHECK-LABEL: insertelt_nxv8f32_0:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a0, zero, e32, m4, tu, ma
+; CHECK-NEXT: vsetvli a0, zero, e32, m1, tu, ma
; CHECK-NEXT: vfmv.s.f v8, fa0
; CHECK-NEXT: ret
%r = insertelement <vscale x 8 x float> %v, float %elt, i32 0
@@ -323,7 +323,7 @@ define <vscale x 8 x float> @insertelt_nxv8f32_0(<vscale x 8 x float> %v, float
define <vscale x 8 x float> @insertelt_nxv8f32_imm(<vscale x 8 x float> %v, float %elt) {
; CHECK-LABEL: insertelt_nxv8f32_imm:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetivli zero, 4, e32, m4, tu, ma
+; CHECK-NEXT: vsetivli zero, 4, e32, m1, tu, ma
; CHECK-NEXT: vfmv.s.f v12, fa0
; CHECK-NEXT: vslideup.vi v8, v12, 3
; CHECK-NEXT: ret
@@ -347,7 +347,7 @@ define <vscale x 8 x float> @insertelt_nxv8f32_idx(<vscale x 8 x float> %v, floa
define <vscale x 16 x float> @insertelt_nxv16f32_0(<vscale x 16 x float> %v, float %elt) {
; CHECK-LABEL: insertelt_nxv16f32_0:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a0, zero, e32, m8, tu, ma
+; CHECK-NEXT: vsetvli a0, zero, e32, m1, tu, ma
; CHECK-NEXT: vfmv.s.f v8, fa0
; CHECK-NEXT: ret
%r = insertelement <vscale x 16 x float> %v, float %elt, i32 0
@@ -357,7 +357,7 @@ define <vscale x 16 x float> @insertelt_nxv16f32_0(<vscale x 16 x float> %v, flo
define <vscale x 16 x float> @insertelt_nxv16f32_imm(<vscale x 16 x float> %v, float %elt) {
; CHECK-LABEL: insertelt_nxv16f32_imm:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetivli zero, 4, e32, m8, tu, ma
+; CHECK-NEXT: vsetivli zero, 4, e32, m1, tu, ma
; CHECK-NEXT: vfmv.s.f v16, fa0
; CHECK-NEXT: vslideup.vi v8, v16, 3
; CHECK-NEXT: ret
@@ -415,7 +415,7 @@ define <vscale x 1 x double> @insertelt_nxv1f64_idx(<vscale x 1 x double> %v, do
define <vscale x 2 x double> @insertelt_nxv2f64_0(<vscale x 2 x double> %v, double %elt) {
; CHECK-LABEL: insertelt_nxv2f64_0:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a0, zero, e64, m2, tu, ma
+; CHECK-NEXT: vsetvli a0, zero, e64, m1, tu, ma
; CHECK-NEXT: vfmv.s.f v8, fa0
; CHECK-NEXT: ret
%r = insertelement <vscale x 2 x double> %v, double %elt, i32 0
@@ -449,7 +449,7 @@ define <vscale x 2 x double> @insertelt_nxv2f64_idx(<vscale x 2 x double> %v, do
define <vscale x 4 x double> @insertelt_nxv4f64_0(<vscale x 4 x double> %v, double %elt) {
; CHECK-LABEL: insertelt_nxv4f64_0:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a0, zero, e64, m4, tu, ma
+; CHECK-NEXT: vsetvli a0, zero, e64, m1, tu, ma
; CHECK-NEXT: vfmv.s.f v8, fa0
; CHECK-NEXT: ret
%r = insertelement <vscale x 4 x double> %v, double %elt, i32 0
@@ -459,7 +459,7 @@ define <vscale x 4 x double> @insertelt_nxv4f64_0(<vscale x 4 x double> %v, doub
define <vscale x 4 x double> @insertelt_nxv4f64_imm(<vscale x 4 x double> %v, double %elt) {
; CHECK-LABEL: insertelt_nxv4f64_imm:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetivli zero, 4, e64, m4, tu, ma
+; CHECK-NEXT: vsetivli zero, 4, e64, m2, tu, ma
; CHECK-NEXT: vfmv.s.f v12, fa0
; CHECK-NEXT: vslideup.vi v8, v12, 3
; CHECK-NEXT: ret
@@ -483,7 +483,7 @@ define <vscale x 4 x double> @insertelt_nxv4f64_idx(<vscale x 4 x double> %v, do
define <vscale x 8 x double> @insertelt_nxv8f64_0(<vscale x 8 x double> %v, double %elt) {
; CHECK-LABEL: insertelt_nxv8f64_0:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a0, zero, e64, m8, tu, ma
+; CHECK-NEXT: vsetvli a0, zero, e64, m1, tu, ma
; CHECK-NEXT: vfmv.s.f v8, fa0
; CHECK-NEXT: ret
%r = insertelement <vscale x 8 x double> %v, double %elt, i32 0
@@ -493,7 +493,7 @@ define <vscale x 8 x double> @insertelt_nxv8f64_0(<vscale x 8 x double> %v, doub
define <vscale x 8 x double> @insertelt_nxv8f64_imm(<vscale x 8 x double> %v, double %elt) {
; CHECK-LABEL: insertelt_nxv8f64_imm:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetivli zero, 4, e64, m8, tu, ma
+; CHECK-NEXT: vsetivli zero, 4, e64, m2, tu, ma
; CHECK-NEXT: vfmv.s.f v16, fa0
; CHECK-NEXT: vslideup.vi v8, v16, 3
; CHECK-NEXT: ret
diff --git a/llvm/test/CodeGen/RISCV/rvv/insertelt-i1.ll b/llvm/test/CodeGen/RISCV/rvv/insertelt-i1.ll
index a7bd15f2a7b330a..911072d9571ff81 100644
--- a/llvm/test/CodeGen/RISCV/rvv/insertelt-i1.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/insertelt-i1.ll
@@ -149,7 +149,7 @@ define <vscale x 16 x i1> @insertelt_nxv16i1(<vscale x 16 x i1> %x, i1 %elt) {
; CHECK-NEXT: vmv.v.i v8, 0
; CHECK-NEXT: vmerge.vim v8, v8, 1, v0
; CHECK-NEXT: vmv.s.x v10, a0
-; CHECK-NEXT: vsetivli zero, 3, e8, m2, tu, ma
+; CHECK-NEXT: vsetivli zero, 3, e8, m1, tu, ma
; CHECK-NEXT: vslideup.vi v8, v10, 2
; CHECK-NEXT: vsetvli a0, zero, e8, m2, ta, ma
; CHECK-NEXT: vand.vi v8, v8, 1
@@ -184,7 +184,7 @@ define <vscale x 32 x i1> @insertelt_nxv32i1(<vscale x 32 x i1> %x, i1 %elt) {
; CHECK-NEXT: vmv.v.i v8, 0
; CHECK-NEXT: vmerge.vim v8, v8, 1, v0
; CHECK-NEXT: vmv.s.x v12, a0
-; CHECK-NEXT: vsetivli zero, 3, e8, m4, tu, ma
+; CHECK-NEXT: vsetivli zero, 3, e8, m1, tu, ma
; CHECK-NEXT: vslideup.vi v8, v12, 2
; CHECK-NEXT: vsetvli a0, zero, e8, m4, ta, ma
; CHECK-NEXT: vand.vi v8, v8, 1
@@ -219,7 +219,7 @@ define <vscale x 64 x i1> @insertelt_nxv64i1(<vscale x 64 x i1> %x, i1 %elt) {
; CHECK-NEXT: vmv.v.i v8, 0
; CHECK-NEXT: vmerge.vim v8, v8, 1, v0
; CHECK-NEXT: vmv.s.x v16, a0
-; CHECK-NEXT: vsetivli zero, 3, e8, m8, tu, ma
+; CHECK-NEXT: vsetivli zero, 3, e8, m1, tu, ma
; CHECK-NEXT: vslideup.vi v8, v16, 2
; CHECK-NEXT: vsetvli a0, zero, e8, m8, ta, ma
; CHECK-NEXT: vand.vi v8, v8, 1
diff --git a/llvm/test/CodeGen/RISCV/rvv/insertelt-int-rv32.ll b/llvm/test/CodeGen/RISCV/rvv/insertelt-int-rv32.ll
index 39f94eab2aa6606..883906772874e72 100644
--- a/llvm/test/CodeGen/RISCV/rvv/insertelt-int-rv32.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/insertelt-int-rv32.ll
@@ -141,7 +141,7 @@ define <vscale x 8 x i8> @insertelt_nxv8i8_idx(<vscale x 8 x i8> %v, i8 signext
define <vscale x 16 x i8> @insertelt_nxv16i8_0(<vscale x 16 x i8> %v, i8 signext %elt) {
; CHECK-LABEL: insertelt_nxv16i8_0:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a1, zero, e8, m2, tu, ma
+; CHECK-NEXT: vsetvli a1, zero, e8, m1, tu, ma
; CHECK-NEXT: vmv.s.x v8, a0
; CHECK-NEXT: ret
%r = insertelement <vscale x 16 x i8> %v, i8 %elt, i32 0
@@ -151,7 +151,7 @@ define <vscale x 16 x i8> @insertelt_nxv16i8_0(<vscale x 16 x i8> %v, i8 signext
define <vscale x 16 x i8> @insertelt_nxv16i8_imm(<vscale x 16 x i8> %v, i8 signext %elt) {
; CHECK-LABEL: insertelt_nxv16i8_imm:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetivli zero, 4, e8, m2, tu, ma
+; CHECK-NEXT: vsetivli zero, 4, e8, m1, tu, ma
; CHECK-NEXT: vmv.s.x v10, a0
; CHECK-NEXT: vslideup.vi v8, v10, 3
; CHECK-NEXT: ret
@@ -175,7 +175,7 @@ define <vscale x 16 x i8> @insertelt_nxv16i8_idx(<vscale x 16 x i8> %v, i8 signe
define <vscale x 32 x i8> @insertelt_nxv32i8_0(<vscale x 32 x i8> %v, i8 signext %elt) {
; CHECK-LABEL: insertelt_nxv32i8_0:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a1, zero, e8, m4, tu, ma
+; CHECK-NEXT: vsetvli a1, zero, e8, m1, tu, ma
; CHECK-NEXT: vmv.s.x v8, a0
; CHECK-NEXT: ret
%r = insertelement <vscale x 32 x i8> %v, i8 %elt, i32 0
@@ -185,7 +185,7 @@ define <vscale x 32 x i8> @insertelt_nxv32i8_0(<vscale x 32 x i8> %v, i8 signext
define <vscale x 32 x i8> @insertelt_nxv32i8_imm(<vscale x 32 x i8> %v, i8 signext %elt) {
; CHECK-LABEL: insertelt_nxv32i8_imm:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetivli zero, 4, e8, m4, tu, ma
+; CHECK-NEXT: vsetivli zero, 4, e8, m1, tu, ma
; CHECK-NEXT: vmv.s.x v12, a0
; CHECK-NEXT: vslideup.vi v8, v12, 3
; CHECK-NEXT: ret
@@ -209,7 +209,7 @@ define <vscale x 32 x i8> @insertelt_nxv32i8_idx(<vscale x 32 x i8> %v, i8 signe
define <vscale x 64 x i8> @insertelt_nxv64i8_0(<vscale x 64 x i8> %v, i8 signext %elt) {
; CHECK-LABEL: insertelt_nxv64i8_0:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a1, zero, e8, m8, tu, ma
+; CHECK-NEXT: vsetvli a1, zero, e8, m1, tu, ma
; CHECK-NEXT: vmv.s.x v8, a0
; CHECK-NEXT: ret
%r = insertelement <vscale x 64 x i8> %v, i8 %elt, i32 0
@@ -219,7 +219,7 @@ define <vscale x 64 x i8> @insertelt_nxv64i8_0(<vscale x 64 x i8> %v, i8 signext
define <vscale x 64 x i8> @insertelt_nxv64i8_imm(<vscale x 64 x i8> %v, i8 signext %elt) {
; CHECK-LABEL: insertelt_nxv64i8_imm:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetivli zero, 4, e8, m8, tu, ma
+; CHECK-NEXT: vsetivli zero, 4, e8, m1, tu, ma
; CHECK-NEXT: vmv.s.x v16, a0
; CHECK-NEXT: vslideup.vi v8, v16, 3
; CHECK-NEXT: ret
@@ -345,7 +345,7 @@ define <vscale x 4 x i16> @insertelt_nxv4i16_idx(<vscale x 4 x i16> %v, i16 sign
define <vscale x 8 x i16> @insertelt_nxv8i16_0(<vscale x 8 x i16> %v, i16 signext %elt) {
; CHECK-LABEL: insertelt_nxv8i16_0:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a1, zero, e16, m2, tu, ma
+; CHECK-NEXT: vsetvli a1, zero, e16, m1, tu, ma
; CHECK-NEXT: vmv.s.x v8, a0
; CHECK-NEXT: ret
%r = insertelement <vscale x 8 x i16> %v, i16 %elt, i32 0
@@ -355,7 +355,7 @@ define <vscale x 8 x i16> @insertelt_nxv8i16_0(<vscale x 8 x i16> %v, i16 signex
define <vscale x 8 x i16> @insertelt_nxv8i16_imm(<vscale x 8 x i16> %v, i16 signext %elt) {
; CHECK-LABEL: insertelt_nxv8i16_imm:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetivli zero, 4, e16, m2, tu, ma
+; CHECK-NEXT: vsetivli zero, 4, e16, m1, tu, ma
; CHECK-NEXT: vmv.s.x v10, a0
; CHECK-NEXT: vslideup.vi v8, v10, 3
; CHECK-NEXT: ret
@@ -379,7 +379,7 @@ define <vscale x 8 x i16> @insertelt_nxv8i16_idx(<vscale x 8 x i16> %v, i16 sign
define <vscale x 16 x i16> @insertelt_nxv16i16_0(<vscale x 16 x i16> %v, i16 signext %elt) {
; CHECK-LABEL: insertelt_nxv16i16_0:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a1, zero, e16, m4, tu, ma
+; CHECK-NEXT: vsetvli a1, zero, e16, m1, tu, ma
; CHECK-NEXT: vmv.s.x v8, a0
; CHECK-NEXT: ret
%r = insertelement <vscale x 16 x i16> %v, i16 %elt, i32 0
@@ -389,7 +389,7 @@ define <vscale x 16 x i16> @insertelt_nxv16i16_0(<vscale x 16 x i16> %v, i16 sig
define <vscale x 16 x i16> @insertelt_nxv16i16_imm(<vscale x 16 x i16> %v, i16 signext %elt) {
; CHECK-LABEL: insertelt_nxv16i16_imm:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetivli zero, 4, e16, m4, tu, ma
+; CHECK-NEXT: vsetivli zero, 4, e16, m1, tu, ma
; CHECK-NEXT: vmv.s.x v12, a0
; CHECK-NEXT: vslideup.vi v8, v12, 3
; CHECK-NEXT: ret
@@ -413,7 +413,7 @@ define <vscale x 16 x i16> @insertelt_nxv16i16_idx(<vscale x 16 x i16> %v, i16 s
define <vscale x 32 x i16> @insertelt_nxv32i16_0(<vscale x 32 x i16> %v, i16 signext %elt) {
; CHECK-LABEL: insertelt_nxv32i16_0:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a1, zero, e16, m8, tu, ma
+; CHECK-NEXT: vsetvli a1, zero, e16, m1, tu, ma
; CHECK-NEXT: vmv.s.x v8, a0
; CHECK-NEXT: ret
%r = insertelement <vscale x 32 x i16> %v, i16 %elt, i32 0
@@ -423,7 +423,7 @@ define <vscale x 32 x i16> @insertelt_nxv32i16_0(<vscale x 32 x i16> %v, i16 sig
define <vscale x 32 x i16> @insertelt_nxv32i16_imm(<vscale x 32 x i16> %v, i16 signext %elt) {
; CHECK-LABEL: insertelt_nxv32i16_imm:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetivli zero, 4, e16, m8, tu, ma
+; CHECK-NEXT: vsetivli zero, 4, e16, m1, tu, ma
; CHECK-NEXT: vmv.s.x v16, a0
; CHECK-NEXT: vslideup.vi v8, v16, 3
; CHECK-NEXT: ret
@@ -515,7 +515,7 @@ define <vscale x 2 x i32> @insertelt_nxv2i32_idx(<vscale x 2 x i32> %v, i32 %elt
define <vscale x 4 x i32> @insertelt_nxv4i32_0(<vscale x 4 x i32> %v, i32 %elt) {
; CHECK-LABEL: insertelt_nxv4i32_0:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a1, zero, e32, m2, tu, ma
+; CHECK-NEXT: vsetvli a1, zero, e32, m1, tu, ma
; CHECK-NEXT: vmv.s.x v8, a0
; CHECK-NEXT: ret
%r = insertelement <vscale x 4 x i32> %v, i32 %elt, i32 0
@@ -525,7 +525,7 @@ define <vscale x 4 x i32> @insertelt_nxv4i32_0(<vscale x 4 x i32> %v, i32 %elt)
define <vscale x 4 x i32> @insertelt_nxv4i32_imm(<vscale x 4 x i32> %v, i32 %elt) {
; CHECK-LABEL: insertelt_nxv4i32_imm:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetivli zero, 4, e32, m2, tu, ma
+; CHECK-NEXT: vsetivli zero, 4, e32, m1, tu, ma
; CHECK-NEXT: vmv.s.x v10, a0
; CHECK-NEXT: vslideup.vi v8, v10, 3
; CHECK-NEXT: ret
@@ -549,7 +549,7 @@ define <vscale x 4 x i32> @insertelt_nxv4i32_idx(<vscale x 4 x i32> %v, i32 %elt
define <vscale x 8 x i32> @insertelt_nxv8i32_0(<vscale x 8 x i32> %v, i32 %elt) {
; CHECK-LABEL: insertelt_nxv8i32_0:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a1, zero, e32, m4, tu, ma
+; CHECK-NEXT: vsetvli a1, zero, e32, m1, tu, ma
; CHECK-NEXT: vmv.s.x v8, a0
; CHECK-NEXT: ret
%r = insertelement <vscale x 8 x i32> %v, i32 %elt, i32 0
@@ -559,7 +559,7 @@ define <vscale x 8 x i32> @insertelt_nxv8i32_0(<vscale x 8 x i32> %v, i32 %elt)
define <vscale x 8 x i32> @insertelt_nxv8i32_imm(<vscale x 8 x i32> %v, i32 %elt) {
; CHECK-LABEL: insertelt_nxv8i32_imm:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetivli zero, 4, e32, m4, tu, ma
+; CHECK-NEXT: vsetivli zero, 4, e32, m1, tu, ma
; CHECK-NEXT: vmv.s.x v12, a0
; CHECK-NEXT: vslideup.vi v8, v12, 3
; CHECK-NEXT: ret
@@ -583,7 +583,7 @@ define <vscale x 8 x i32> @insertelt_nxv8i32_idx(<vscale x 8 x i32> %v, i32 %elt
define <vscale x 16 x i32> @insertelt_nxv16i32_0(<vscale x 16 x i32> %v, i32 %elt) {
; CHECK-LABEL: insertelt_nxv16i32_0:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a1, zero, e32, m8, tu, ma
+; CHECK-NEXT: vsetvli a1, zero, e32, m1, tu, ma
; CHECK-NEXT: vmv.s.x v8, a0
; CHECK-NEXT: ret
%r = insertelement <vscale x 16 x i32> %v, i32 %elt, i32 0
@@ -593,7 +593,7 @@ define <vscale x 16 x i32> @insertelt_nxv16i32_0(<vscale x 16 x i32> %v, i32 %el
define <vscale x 16 x i32> @insertelt_nxv16i32_imm(<vscale x 16 x i32> %v, i32 %elt) {
; CHECK-LABEL: insertelt_nxv16i32_imm:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetivli zero, 4, e32, m8, tu, ma
+; CHECK-NEXT: vsetivli zero, 4, e32, m1, tu, ma
; CHECK-NEXT: vmv.s.x v16, a0
; CHECK-NEXT: vslideup.vi v8, v16, 3
; CHECK-NEXT: ret
@@ -655,7 +655,7 @@ define <vscale x 1 x i64> @insertelt_nxv1i64_idx(<vscale x 1 x i64> %v, i64 %elt
define <vscale x 2 x i64> @insertelt_nxv2i64_0(<vscale x 2 x i64> %v, i64 %elt) {
; CHECK-LABEL: insertelt_nxv2i64_0:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetivli zero, 2, e32, m2, tu, ma
+; CHECK-NEXT: vsetivli zero, 2, e32, m1, tu, ma
; CHECK-NEXT: vslide1down.vx v8, v8, a0
; CHECK-NEXT: vslide1down.vx v8, v8, a1
; CHECK-NEXT: ret
@@ -693,7 +693,7 @@ define <vscale x 2 x i64> @insertelt_nxv2i64_idx(<vscale x 2 x i64> %v, i64 %elt
define <vscale x 4 x i64> @insertelt_nxv4i64_0(<vscale x 4 x i64> %v, i64 %elt) {
; CHECK-LABEL: insertelt_nxv4i64_0:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetivli zero, 2, e32, m4, tu, ma
+; CHECK-NEXT: vsetivli zero, 2, e32, m1, tu, ma
; CHECK-NEXT: vslide1down.vx v8, v8, a0
; CHECK-NEXT: vslide1down.vx v8, v8, a1
; CHECK-NEXT: ret
@@ -704,10 +704,10 @@ define <vscale x 4 x i64> @insertelt_nxv4i64_0(<vscale x 4 x i64> %v, i64 %elt)
define <vscale x 4 x i64> @insertelt_nxv4i64_imm(<vscale x 4 x i64> %v, i64 %elt) {
; CHECK-LABEL: insertelt_nxv4i64_imm:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetivli zero, 2, e32, m4, ta, ma
+; CHECK-NEXT: vsetivli zero, 2, e32, m2, ta, ma
; CHECK-NEXT: vslide1down.vx v12, v8, a0
; CHECK-NEXT: vslide1down.vx v12, v12, a1
-; CHECK-NEXT: vsetivli zero, 4, e64, m4, tu, ma
+; CHECK-NEXT: vsetivli zero, 4, e64, m2, tu, ma
; CHECK-NEXT: vslideup.vi v8, v12, 3
; CHECK-NEXT: ret
%r = insertelement <vscale x 4 x i64> %v, i64 %elt, i32 3
@@ -731,7 +731,7 @@ define <vscale x 4 x i64> @insertelt_nxv4i64_idx(<vscale x 4 x i64> %v, i64 %elt
define <vscale x 8 x i64> @insertelt_nxv8i64_0(<vscale x 8 x i64> %v, i64 %elt) {
; CHECK-LABEL: insertelt_nxv8i64_0:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetivli zero, 2, e32, m8, tu, ma
+; CHECK-NEXT: vsetivli zero, 2, e32, m1, tu, ma
; CHECK-NEXT: vslide1down.vx v8, v8, a0
; CHECK-NEXT: vslide1down.vx v8, v8, a1
; CHECK-NEXT: ret
@@ -742,10 +742,10 @@ define <vscale x 8 x i64> @insertelt_nxv8i64_0(<vscale x 8 x i64> %v, i64 %elt)
define <vscale x 8 x i64> @insertelt_nxv8i64_imm(<vscale x 8 x i64> %v, i64 %elt) {
; CHECK-LABEL: insertelt_nxv8i64_imm:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetivli zero, 2, e32, m8, ta, ma
+; CHECK-NEXT: vsetivli zero, 2, e32, m2, ta, ma
; CHECK-NEXT: vslide1down.vx v16, v8, a0
; CHECK-NEXT: vslide1down.vx v16, v16, a1
-; CHECK-NEXT: vsetivli zero, 4, e64, m8, tu, ma
+; CHECK-NEXT: vsetivli zero, 4, e64, m2, tu, ma
; CHECK-NEXT: vslideup.vi v8, v16, 3
; CHECK-NEXT: ret
%r = insertelement <vscale x 8 x i64> %v, i64 %elt, i32 3
@@ -771,7 +771,7 @@ define <vscale x 2 x i64> @insertelt_nxv2i64_0_c10(<vscale x 2 x i64> %v) {
; CHECK-LABEL: insertelt_nxv2i64_0_c10:
; CHECK: # %bb.0:
; CHECK-NEXT: li a0, 10
-; CHECK-NEXT: vsetvli a1, zero, e64, m2, tu, ma
+; CHECK-NEXT: vsetvli a1, zero, e64, m1, tu, ma
; CHECK-NEXT: vmv.s.x v8, a0
; CHECK-NEXT: ret
%r = insertelement <vscale x 2 x i64> %v, i64 10, i32 0
@@ -807,7 +807,7 @@ define <vscale x 2 x i64> @insertelt_nxv2i64_0_cn1(<vscale x 2 x i64> %v) {
; CHECK-LABEL: insertelt_nxv2i64_0_cn1:
; CHECK: # %bb.0:
; CHECK-NEXT: li a0, -1
-; CHECK-NEXT: vsetvli a1, zero, e64, m2, tu, ma
+; CHECK-NEXT: vsetvli a1, zero, e64, m1, tu, ma
; CHECK-NEXT: vmv.s.x v8, a0
; CHECK-NEXT: ret
%r = insertelement <vscale x 2 x i64> %v, i64 -1, i32 0
diff --git a/llvm/test/CodeGen/RISCV/rvv/insertelt-int-rv64.ll b/llvm/test/CodeGen/RISCV/rvv/insertelt-int-rv64.ll
index 1dd00197bbbb044..40767eefdd814cb 100644
--- a/llvm/test/CodeGen/RISCV/rvv/insertelt-int-rv64.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/insertelt-int-rv64.ll
@@ -141,7 +141,7 @@ define <vscale x 8 x i8> @insertelt_nxv8i8_idx(<vscale x 8 x i8> %v, i8 signext
define <vscale x 16 x i8> @insertelt_nxv16i8_0(<vscale x 16 x i8> %v, i8 signext %elt) {
; CHECK-LABEL: insertelt_nxv16i8_0:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a1, zero, e8, m2, tu, ma
+; CHECK-NEXT: vsetvli a1, zero, e8, m1, tu, ma
; CHECK-NEXT: vmv.s.x v8, a0
; CHECK-NEXT: ret
%r = insertelement <vscale x 16 x i8> %v, i8 %elt, i32 0
@@ -151,7 +151,7 @@ define <vscale x 16 x i8> @insertelt_nxv16i8_0(<vscale x 16 x i8> %v, i8 signext
define <vscale x 16 x i8> @insertelt_nxv16i8_imm(<vscale x 16 x i8> %v, i8 signext %elt) {
; CHECK-LABEL: insertelt_nxv16i8_imm:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetivli zero, 4, e8, m2, tu, ma
+; CHECK-NEXT: vsetivli zero, 4, e8, m1, tu, ma
; CHECK-NEXT: vmv.s.x v10, a0
; CHECK-NEXT: vslideup.vi v8, v10, 3
; CHECK-NEXT: ret
@@ -175,7 +175,7 @@ define <vscale x 16 x i8> @insertelt_nxv16i8_idx(<vscale x 16 x i8> %v, i8 signe
define <vscale x 32 x i8> @insertelt_nxv32i8_0(<vscale x 32 x i8> %v, i8 signext %elt) {
; CHECK-LABEL: insertelt_nxv32i8_0:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a1, zero, e8, m4, tu, ma
+; CHECK-NEXT: vsetvli a1, zero, e8, m1, tu, ma
; CHECK-NEXT: vmv.s.x v8, a0
; CHECK-NEXT: ret
%r = insertelement <vscale x 32 x i8> %v, i8 %elt, i32 0
@@ -185,7 +185,7 @@ define <vscale x 32 x i8> @insertelt_nxv32i8_0(<vscale x 32 x i8> %v, i8 signext
define <vscale x 32 x i8> @insertelt_nxv32i8_imm(<vscale x 32 x i8> %v, i8 signext %elt) {
; CHECK-LABEL: insertelt_nxv32i8_imm:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetivli zero, 4, e8, m4, tu, ma
+; CHECK-NEXT: vsetivli zero, 4, e8, m1, tu, ma
; CHECK-NEXT: vmv.s.x v12, a0
; CHECK-NEXT: vslideup.vi v8, v12, 3
; CHECK-NEXT: ret
@@ -209,7 +209,7 @@ define <vscale x 32 x i8> @insertelt_nxv32i8_idx(<vscale x 32 x i8> %v, i8 signe
define <vscale x 64 x i8> @insertelt_nxv64i8_0(<vscale x 64 x i8> %v, i8 signext %elt) {
; CHECK-LABEL: insertelt_nxv64i8_0:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a1, zero, e8, m8, tu, ma
+; CHECK-NEXT: vsetvli a1, zero, e8, m1, tu, ma
; CHECK-NEXT: vmv.s.x v8, a0
; CHECK-NEXT: ret
%r = insertelement <vscale x 64 x i8> %v, i8 %elt, i32 0
@@ -219,7 +219,7 @@ define <vscale x 64 x i8> @insertelt_nxv64i8_0(<vscale x 64 x i8> %v, i8 signext
define <vscale x 64 x i8> @insertelt_nxv64i8_imm(<vscale x 64 x i8> %v, i8 signext %elt) {
; CHECK-LABEL: insertelt_nxv64i8_imm:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetivli zero, 4, e8, m8, tu, ma
+; CHECK-NEXT: vsetivli zero, 4, e8, m1, tu, ma
; CHECK-NEXT: vmv.s.x v16, a0
; CHECK-NEXT: vslideup.vi v8, v16, 3
; CHECK-NEXT: ret
@@ -345,7 +345,7 @@ define <vscale x 4 x i16> @insertelt_nxv4i16_idx(<vscale x 4 x i16> %v, i16 sign
define <vscale x 8 x i16> @insertelt_nxv8i16_0(<vscale x 8 x i16> %v, i16 signext %elt) {
; CHECK-LABEL: insertelt_nxv8i16_0:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a1, zero, e16, m2, tu, ma
+; CHECK-NEXT: vsetvli a1, zero, e16, m1, tu, ma
; CHECK-NEXT: vmv.s.x v8, a0
; CHECK-NEXT: ret
%r = insertelement <vscale x 8 x i16> %v, i16 %elt, i32 0
@@ -355,7 +355,7 @@ define <vscale x 8 x i16> @insertelt_nxv8i16_0(<vscale x 8 x i16> %v, i16 signex
define <vscale x 8 x i16> @insertelt_nxv8i16_imm(<vscale x 8 x i16> %v, i16 signext %elt) {
; CHECK-LABEL: insertelt_nxv8i16_imm:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetivli zero, 4, e16, m2, tu, ma
+; CHECK-NEXT: vsetivli zero, 4, e16, m1, tu, ma
; CHECK-NEXT: vmv.s.x v10, a0
; CHECK-NEXT: vslideup.vi v8, v10, 3
; CHECK-NEXT: ret
@@ -379,7 +379,7 @@ define <vscale x 8 x i16> @insertelt_nxv8i16_idx(<vscale x 8 x i16> %v, i16 sign
define <vscale x 16 x i16> @insertelt_nxv16i16_0(<vscale x 16 x i16> %v, i16 signext %elt) {
; CHECK-LABEL: insertelt_nxv16i16_0:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a1, zero, e16, m4, tu, ma
+; CHECK-NEXT: vsetvli a1, zero, e16, m1, tu, ma
; CHECK-NEXT: vmv.s.x v8, a0
; CHECK-NEXT: ret
%r = insertelement <vscale x 16 x i16> %v, i16 %elt, i32 0
@@ -389,7 +389,7 @@ define <vscale x 16 x i16> @insertelt_nxv16i16_0(<vscale x 16 x i16> %v, i16 sig
define <vscale x 16 x i16> @insertelt_nxv16i16_imm(<vscale x 16 x i16> %v, i16 signext %elt) {
; CHECK-LABEL: insertelt_nxv16i16_imm:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetivli zero, 4, e16, m4, tu, ma
+; CHECK-NEXT: vsetivli zero, 4, e16, m1, tu, ma
; CHECK-NEXT: vmv.s.x v12, a0
; CHECK-NEXT: vslideup.vi v8, v12, 3
; CHECK-NEXT: ret
@@ -413,7 +413,7 @@ define <vscale x 16 x i16> @insertelt_nxv16i16_idx(<vscale x 16 x i16> %v, i16 s
define <vscale x 32 x i16> @insertelt_nxv32i16_0(<vscale x 32 x i16> %v, i16 signext %elt) {
; CHECK-LABEL: insertelt_nxv32i16_0:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a1, zero, e16, m8, tu, ma
+; CHECK-NEXT: vsetvli a1, zero, e16, m1, tu, ma
; CHECK-NEXT: vmv.s.x v8, a0
; CHECK-NEXT: ret
%r = insertelement <vscale x 32 x i16> %v, i16 %elt, i32 0
@@ -423,7 +423,7 @@ define <vscale x 32 x i16> @insertelt_nxv32i16_0(<vscale x 32 x i16> %v, i16 sig
define <vscale x 32 x i16> @insertelt_nxv32i16_imm(<vscale x 32 x i16> %v, i16 signext %elt) {
; CHECK-LABEL: insertelt_nxv32i16_imm:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetivli zero, 4, e16, m8, tu, ma
+; CHECK-NEXT: vsetivli zero, 4, e16, m1, tu, ma
; CHECK-NEXT: vmv.s.x v16, a0
; CHECK-NEXT: vslideup.vi v8, v16, 3
; CHECK-NEXT: ret
@@ -515,7 +515,7 @@ define <vscale x 2 x i32> @insertelt_nxv2i32_idx(<vscale x 2 x i32> %v, i32 sign
define <vscale x 4 x i32> @insertelt_nxv4i32_0(<vscale x 4 x i32> %v, i32 signext %elt) {
; CHECK-LABEL: insertelt_nxv4i32_0:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a1, zero, e32, m2, tu, ma
+; CHECK-NEXT: vsetvli a1, zero, e32, m1, tu, ma
; CHECK-NEXT: vmv.s.x v8, a0
; CHECK-NEXT: ret
%r = insertelement <vscale x 4 x i32> %v, i32 %elt, i32 0
@@ -525,7 +525,7 @@ define <vscale x 4 x i32> @insertelt_nxv4i32_0(<vscale x 4 x i32> %v, i32 signex
define <vscale x 4 x i32> @insertelt_nxv4i32_imm(<vscale x 4 x i32> %v, i32 signext %elt) {
; CHECK-LABEL: insertelt_nxv4i32_imm:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetivli zero, 4, e32, m2, tu, ma
+; CHECK-NEXT: vsetivli zero, 4, e32, m1, tu, ma
; CHECK-NEXT: vmv.s.x v10, a0
; CHECK-NEXT: vslideup.vi v8, v10, 3
; CHECK-NEXT: ret
@@ -549,7 +549,7 @@ define <vscale x 4 x i32> @insertelt_nxv4i32_idx(<vscale x 4 x i32> %v, i32 sign
define <vscale x 8 x i32> @insertelt_nxv8i32_0(<vscale x 8 x i32> %v, i32 signext %elt) {
; CHECK-LABEL: insertelt_nxv8i32_0:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a1, zero, e32, m4, tu, ma
+; CHECK-NEXT: vsetvli a1, zero, e32, m1, tu, ma
; CHECK-NEXT: vmv.s.x v8, a0
; CHECK-NEXT: ret
%r = insertelement <vscale x 8 x i32> %v, i32 %elt, i32 0
@@ -559,7 +559,7 @@ define <vscale x 8 x i32> @insertelt_nxv8i32_0(<vscale x 8 x i32> %v, i32 signex
define <vscale x 8 x i32> @insertelt_nxv8i32_imm(<vscale x 8 x i32> %v, i32 signext %elt) {
; CHECK-LABEL: insertelt_nxv8i32_imm:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetivli zero, 4, e32, m4, tu, ma
+; CHECK-NEXT: vsetivli zero, 4, e32, m1, tu, ma
; CHECK-NEXT: vmv.s.x v12, a0
; CHECK-NEXT: vslideup.vi v8, v12, 3
; CHECK-NEXT: ret
@@ -583,7 +583,7 @@ define <vscale x 8 x i32> @insertelt_nxv8i32_idx(<vscale x 8 x i32> %v, i32 sign
define <vscale x 16 x i32> @insertelt_nxv16i32_0(<vscale x 16 x i32> %v, i32 signext %elt) {
; CHECK-LABEL: insertelt_nxv16i32_0:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a1, zero, e32, m8, tu, ma
+; CHECK-NEXT: vsetvli a1, zero, e32, m1, tu, ma
; CHECK-NEXT: vmv.s.x v8, a0
; CHECK-NEXT: ret
%r = insertelement <vscale x 16 x i32> %v, i32 %elt, i32 0
@@ -593,7 +593,7 @@ define <vscale x 16 x i32> @insertelt_nxv16i32_0(<vscale x 16 x i32> %v, i32 sig
define <vscale x 16 x i32> @insertelt_nxv16i32_imm(<vscale x 16 x i32> %v, i32 signext %elt) {
; CHECK-LABEL: insertelt_nxv16i32_imm:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetivli zero, 4, e32, m8, tu, ma
+; CHECK-NEXT: vsetivli zero, 4, e32, m1, tu, ma
; CHECK-NEXT: vmv.s.x v16, a0
; CHECK-NEXT: vslideup.vi v8, v16, 3
; CHECK-NEXT: ret
@@ -653,7 +653,7 @@ define <vscale x 1 x i64> @insertelt_nxv1i64_idx(<vscale x 1 x i64> %v, i64 %elt
define <vscale x 2 x i64> @insertelt_nxv2i64_0(<vscale x 2 x i64> %v, i64 %elt) {
; CHECK-LABEL: insertelt_nxv2i64_0:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a1, zero, e64, m2, tu, ma
+; CHECK-NEXT: vsetvli a1, zero, e64, m1, tu, ma
; CHECK-NEXT: vmv.s.x v8, a0
; CHECK-NEXT: ret
%r = insertelement <vscale x 2 x i64> %v, i64 %elt, i32 0
@@ -689,7 +689,7 @@ define <vscale x 2 x i64> @insertelt_nxv2i64_idx(<vscale x 2 x i64> %v, i64 %elt
define <vscale x 4 x i64> @insertelt_nxv4i64_0(<vscale x 4 x i64> %v, i64 %elt) {
; CHECK-LABEL: insertelt_nxv4i64_0:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a1, zero, e64, m4, tu, ma
+; CHECK-NEXT: vsetvli a1, zero, e64, m1, tu, ma
; CHECK-NEXT: vmv.s.x v8, a0
; CHECK-NEXT: ret
%r = insertelement <vscale x 4 x i64> %v, i64 %elt, i32 0
@@ -699,7 +699,7 @@ define <vscale x 4 x i64> @insertelt_nxv4i64_0(<vscale x 4 x i64> %v, i64 %elt)
define <vscale x 4 x i64> @insertelt_nxv4i64_imm(<vscale x 4 x i64> %v, i64 %elt) {
; CHECK-LABEL: insertelt_nxv4i64_imm:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetivli zero, 4, e64, m4, tu, ma
+; CHECK-NEXT: vsetivli zero, 4, e64, m2, tu, ma
; CHECK-NEXT: vmv.s.x v12, a0
; CHECK-NEXT: vslideup.vi v8, v12, 3
; CHECK-NEXT: ret
@@ -725,7 +725,7 @@ define <vscale x 4 x i64> @insertelt_nxv4i64_idx(<vscale x 4 x i64> %v, i64 %elt
define <vscale x 8 x i64> @insertelt_nxv8i64_0(<vscale x 8 x i64> %v, i64 %elt) {
; CHECK-LABEL: insertelt_nxv8i64_0:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetvli a1, zero, e64, m8, tu, ma
+; CHECK-NEXT: vsetvli a1, zero, e64, m1, tu, ma
; CHECK-NEXT: vmv.s.x v8, a0
; CHECK-NEXT: ret
%r = insertelement <vscale x 8 x i64> %v, i64 %elt, i32 0
@@ -735,7 +735,7 @@ define <vscale x 8 x i64> @insertelt_nxv8i64_0(<vscale x 8 x i64> %v, i64 %elt)
define <vscale x 8 x i64> @insertelt_nxv8i64_imm(<vscale x 8 x i64> %v, i64 %elt) {
; CHECK-LABEL: insertelt_nxv8i64_imm:
; CHECK: # %bb.0:
-; CHECK-NEXT: vsetivli zero, 4, e64, m8, tu, ma
+; CHECK-NEXT: vsetivli zero, 4, e64, m2, tu, ma
; CHECK-NEXT: vmv.s.x v16, a0
; CHECK-NEXT: vslideup.vi v8, v16, 3
; CHECK-NEXT: ret
diff --git a/llvm/test/CodeGen/RISCV/rvv/vsetvli-insert-crossbb.ll b/llvm/test/CodeGen/RISCV/rvv/vsetvli-insert-crossbb.ll
index 73f651225da64d3..44a396ee29a8ab3 100644
--- a/llvm/test/CodeGen/RISCV/rvv/vsetvli-insert-crossbb.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/vsetvli-insert-crossbb.ll
@@ -810,9 +810,10 @@ for.end: ; preds = %for.body
define <vscale x 4 x i32> @cross_block_mutate(<vscale x 4 x i32> %a, <vscale x 4 x i32> %b,
; CHECK-LABEL: cross_block_mutate:
; CHECK: # %bb.0: # %entry
-; CHECK-NEXT: vsetivli a0, 6, e32, m2, tu, ma
+; CHECK-NEXT: vsetivli a0, 6, e16, m1, ta, ma
+; CHECK-NEXT: vsetvli a1, zero, e32, m1, tu, ma
; CHECK-NEXT: vmv.s.x v8, a0
-; CHECK-NEXT: vsetvli zero, zero, e32, m2, ta, ma
+; CHECK-NEXT: vsetivli zero, 6, e32, m2, ta, ma
; CHECK-NEXT: vadd.vv v8, v8, v10, v0.t
; CHECK-NEXT: ret
<vscale x 4 x i1> %mask) {
More information about the llvm-commits
mailing list