[llvm] 1d26bbc - [RISCV] Return false from isShuffleMaskLegal except for splats.
Craig Topper via llvm-commits
llvm-commits at lists.llvm.org
Thu Mar 11 20:05:02 PST 2021
Author: Craig Topper
Date: 2021-03-11T20:02:49-08:00
New Revision: 1d26bbcf9bb3b9bf1d1efea9d0aeac667995db3f
URL: https://github.com/llvm/llvm-project/commit/1d26bbcf9bb3b9bf1d1efea9d0aeac667995db3f
DIFF: https://github.com/llvm/llvm-project/commit/1d26bbcf9bb3b9bf1d1efea9d0aeac667995db3f.diff
LOG: [RISCV] Return false from isShuffleMaskLegal except for splats.
We don't support any other shuffles currently.
This changes the bswap/bitreverse tests that check for this in
their expansion code. Previously we expanded a byte swapping
shuffle through memory. Now we're scalarizing and doing bit
operations on scalars to swap bytes.
In the future we can probably use vrgather.vx to do a byte swap
shuffle.
Added:
Modified:
llvm/lib/Target/RISCV/RISCVISelLowering.cpp
llvm/lib/Target/RISCV/RISCVISelLowering.h
llvm/test/CodeGen/RISCV/rvv/fixed-vectors-bitreverse.ll
llvm/test/CodeGen/RISCV/rvv/fixed-vectors-bswap.ll
Removed:
################################################################################
diff --git a/llvm/lib/Target/RISCV/RISCVISelLowering.cpp b/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
index 025f5f69c152..51000472c059 100644
--- a/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
+++ b/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
@@ -1068,6 +1068,14 @@ bool RISCVTargetLowering::shouldExpandBuildVectorWithShuffles(
return false;
}
+bool RISCVTargetLowering::isShuffleMaskLegal(ArrayRef<int> M, EVT VT) const {
+ // Only splats are currently supported.
+ if (ShuffleVectorSDNode::isSplatMask(M.data(), VT))
+ return true;
+
+ return false;
+}
+
static SDValue lowerBUILD_VECTOR(SDValue Op, SelectionDAG &DAG,
const RISCVSubtarget &Subtarget) {
MVT VT = Op.getSimpleValueType();
diff --git a/llvm/lib/Target/RISCV/RISCVISelLowering.h b/llvm/lib/Target/RISCV/RISCVISelLowering.h
index 0d02acad46dc..03a75c717cdf 100644
--- a/llvm/lib/Target/RISCV/RISCVISelLowering.h
+++ b/llvm/lib/Target/RISCV/RISCVISelLowering.h
@@ -250,6 +250,10 @@ class RISCVTargetLowering : public TargetLowering {
bool isFPImmLegal(const APFloat &Imm, EVT VT,
bool ForCodeSize) const override;
+ /// Return true if the given shuffle mask can be codegen'd directly, or if it
+ /// should be stack expanded.
+ bool isShuffleMaskLegal(ArrayRef<int> M, EVT VT) const override;
+
bool hasBitPreservingFPLogic(EVT VT) const override;
bool
shouldExpandBuildVectorWithShuffles(EVT VT,
diff --git a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-bitreverse.ll b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-bitreverse.ll
index 9198eba9362c..c3ea32110c28 100644
--- a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-bitreverse.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-bitreverse.ll
@@ -1,90 +1,157 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc -mtriple=riscv32 -mattr=+experimental-v -verify-machineinstrs -riscv-v-vector-bits-min=128 -riscv-v-fixed-length-vector-lmul-max=2 -verify-machineinstrs < %s | FileCheck %s --check-prefixes=CHECK,LMULMAX2-RV32
-; RUN: llc -mtriple=riscv64 -mattr=+experimental-v -verify-machineinstrs -riscv-v-vector-bits-min=128 -riscv-v-fixed-length-vector-lmul-max=2 -verify-machineinstrs < %s | FileCheck %s --check-prefixes=CHECK,LMULMAX2-RV64
-; RUN: llc -mtriple=riscv32 -mattr=+experimental-v -verify-machineinstrs -riscv-v-vector-bits-min=128 -riscv-v-fixed-length-vector-lmul-max=1 -verify-machineinstrs < %s | FileCheck %s --check-prefixes=CHECK,LMULMAX1
-; RUN: llc -mtriple=riscv64 -mattr=+experimental-v -verify-machineinstrs -riscv-v-vector-bits-min=128 -riscv-v-fixed-length-vector-lmul-max=1 -verify-machineinstrs < %s | FileCheck %s --check-prefixes=CHECK,LMULMAX1
+; RUN: llc -mtriple=riscv32 -mattr=+experimental-v -verify-machineinstrs -riscv-v-vector-bits-min=128 -riscv-v-fixed-length-vector-lmul-max=2 -verify-machineinstrs < %s | FileCheck %s --check-prefixes=LMULMAX2-RV32
+; RUN: llc -mtriple=riscv64 -mattr=+experimental-v -verify-machineinstrs -riscv-v-vector-bits-min=128 -riscv-v-fixed-length-vector-lmul-max=2 -verify-machineinstrs < %s | FileCheck %s --check-prefixes=LMULMAX2-RV64
+; RUN: llc -mtriple=riscv32 -mattr=+experimental-v -verify-machineinstrs -riscv-v-vector-bits-min=128 -riscv-v-fixed-length-vector-lmul-max=1 -verify-machineinstrs < %s | FileCheck %s --check-prefixes=LMULMAX1-RV32
+; RUN: llc -mtriple=riscv64 -mattr=+experimental-v -verify-machineinstrs -riscv-v-vector-bits-min=128 -riscv-v-fixed-length-vector-lmul-max=1 -verify-machineinstrs < %s | FileCheck %s --check-prefixes=LMULMAX1-RV64
define void @bitreverse_v8i16(<8 x i16>* %x, <8 x i16>* %y) {
-; CHECK-LABEL: bitreverse_v8i16:
-; CHECK: # %bb.0:
-; CHECK-NEXT: addi sp, sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
-; CHECK-NEXT: vsetivli a1, 8, e16,m1,ta,mu
-; CHECK-NEXT: vle16.v v25, (a0)
-; CHECK-NEXT: vsetvli zero, zero, e8,m1,ta,mu
-; CHECK-NEXT: vmv.x.s a1, v25
-; CHECK-NEXT: sb a1, 1(sp)
-; CHECK-NEXT: vsetivli a1, 1, e8,m1,ta,mu
-; CHECK-NEXT: vslidedown.vi v26, v25, 14
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 15(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 15
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 14(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 12
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 13(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 13
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 12(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 10
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 11(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 11
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 10(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 8
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 9(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 9
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 8(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 6
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 7(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 7
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 6(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 4
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 5(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 5
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 4(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 2
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 3(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 3
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 2(sp)
-; CHECK-NEXT: vslidedown.vi v25, v25, 1
-; CHECK-NEXT: vmv.x.s a1, v25
-; CHECK-NEXT: sb a1, 0(sp)
-; CHECK-NEXT: vsetivli a1, 16, e8,m1,ta,mu
-; CHECK-NEXT: vle8.v v25, (sp)
-; CHECK-NEXT: vand.vi v26, v25, 15
-; CHECK-NEXT: vsll.vi v26, v26, 4
-; CHECK-NEXT: vand.vi v25, v25, -16
-; CHECK-NEXT: vsrl.vi v25, v25, 4
-; CHECK-NEXT: vor.vv v25, v25, v26
-; CHECK-NEXT: addi a1, zero, 51
-; CHECK-NEXT: vand.vx v26, v25, a1
-; CHECK-NEXT: vsll.vi v26, v26, 2
-; CHECK-NEXT: addi a1, zero, 204
-; CHECK-NEXT: vand.vx v25, v25, a1
-; CHECK-NEXT: vsrl.vi v25, v25, 2
-; CHECK-NEXT: vor.vv v25, v25, v26
-; CHECK-NEXT: addi a1, zero, 85
-; CHECK-NEXT: vand.vx v26, v25, a1
-; CHECK-NEXT: vsll.vi v26, v26, 1
-; CHECK-NEXT: addi a1, zero, 170
-; CHECK-NEXT: vand.vx v25, v25, a1
-; CHECK-NEXT: vsrl.vi v25, v25, 1
-; CHECK-NEXT: vor.vv v25, v25, v26
-; CHECK-NEXT: vsetivli a1, 8, e16,m1,ta,mu
-; CHECK-NEXT: vse16.v v25, (a0)
-; CHECK-NEXT: addi sp, sp, 16
-; CHECK-NEXT: ret
+; LMULMAX2-RV32-LABEL: bitreverse_v8i16:
+; LMULMAX2-RV32: # %bb.0:
+; LMULMAX2-RV32-NEXT: vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vle16.v v25, (a0)
+; LMULMAX2-RV32-NEXT: vsrl.vi v26, v25, 8
+; LMULMAX2-RV32-NEXT: vsll.vi v25, v25, 8
+; LMULMAX2-RV32-NEXT: vor.vv v25, v25, v26
+; LMULMAX2-RV32-NEXT: lui a1, 1
+; LMULMAX2-RV32-NEXT: addi a1, a1, -241
+; LMULMAX2-RV32-NEXT: vand.vx v26, v25, a1
+; LMULMAX2-RV32-NEXT: vsll.vi v26, v26, 4
+; LMULMAX2-RV32-NEXT: lui a1, 15
+; LMULMAX2-RV32-NEXT: addi a1, a1, 240
+; LMULMAX2-RV32-NEXT: vand.vx v25, v25, a1
+; LMULMAX2-RV32-NEXT: vsrl.vi v25, v25, 4
+; LMULMAX2-RV32-NEXT: vor.vv v25, v25, v26
+; LMULMAX2-RV32-NEXT: lui a1, 3
+; LMULMAX2-RV32-NEXT: addi a1, a1, 819
+; LMULMAX2-RV32-NEXT: vand.vx v26, v25, a1
+; LMULMAX2-RV32-NEXT: vsll.vi v26, v26, 2
+; LMULMAX2-RV32-NEXT: lui a1, 13
+; LMULMAX2-RV32-NEXT: addi a1, a1, -820
+; LMULMAX2-RV32-NEXT: vand.vx v25, v25, a1
+; LMULMAX2-RV32-NEXT: vsrl.vi v25, v25, 2
+; LMULMAX2-RV32-NEXT: vor.vv v25, v25, v26
+; LMULMAX2-RV32-NEXT: lui a1, 5
+; LMULMAX2-RV32-NEXT: addi a1, a1, 1365
+; LMULMAX2-RV32-NEXT: vand.vx v26, v25, a1
+; LMULMAX2-RV32-NEXT: vsll.vi v26, v26, 1
+; LMULMAX2-RV32-NEXT: lui a1, 11
+; LMULMAX2-RV32-NEXT: addi a1, a1, -1366
+; LMULMAX2-RV32-NEXT: vand.vx v25, v25, a1
+; LMULMAX2-RV32-NEXT: vsrl.vi v25, v25, 1
+; LMULMAX2-RV32-NEXT: vor.vv v25, v25, v26
+; LMULMAX2-RV32-NEXT: vse16.v v25, (a0)
+; LMULMAX2-RV32-NEXT: ret
+;
+; LMULMAX2-RV64-LABEL: bitreverse_v8i16:
+; LMULMAX2-RV64: # %bb.0:
+; LMULMAX2-RV64-NEXT: vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX2-RV64-NEXT: vle16.v v25, (a0)
+; LMULMAX2-RV64-NEXT: vsrl.vi v26, v25, 8
+; LMULMAX2-RV64-NEXT: vsll.vi v25, v25, 8
+; LMULMAX2-RV64-NEXT: vor.vv v25, v25, v26
+; LMULMAX2-RV64-NEXT: lui a1, 1
+; LMULMAX2-RV64-NEXT: addiw a1, a1, -241
+; LMULMAX2-RV64-NEXT: vand.vx v26, v25, a1
+; LMULMAX2-RV64-NEXT: vsll.vi v26, v26, 4
+; LMULMAX2-RV64-NEXT: lui a1, 15
+; LMULMAX2-RV64-NEXT: addiw a1, a1, 240
+; LMULMAX2-RV64-NEXT: vand.vx v25, v25, a1
+; LMULMAX2-RV64-NEXT: vsrl.vi v25, v25, 4
+; LMULMAX2-RV64-NEXT: vor.vv v25, v25, v26
+; LMULMAX2-RV64-NEXT: lui a1, 3
+; LMULMAX2-RV64-NEXT: addiw a1, a1, 819
+; LMULMAX2-RV64-NEXT: vand.vx v26, v25, a1
+; LMULMAX2-RV64-NEXT: vsll.vi v26, v26, 2
+; LMULMAX2-RV64-NEXT: lui a1, 13
+; LMULMAX2-RV64-NEXT: addiw a1, a1, -820
+; LMULMAX2-RV64-NEXT: vand.vx v25, v25, a1
+; LMULMAX2-RV64-NEXT: vsrl.vi v25, v25, 2
+; LMULMAX2-RV64-NEXT: vor.vv v25, v25, v26
+; LMULMAX2-RV64-NEXT: lui a1, 5
+; LMULMAX2-RV64-NEXT: addiw a1, a1, 1365
+; LMULMAX2-RV64-NEXT: vand.vx v26, v25, a1
+; LMULMAX2-RV64-NEXT: vsll.vi v26, v26, 1
+; LMULMAX2-RV64-NEXT: lui a1, 11
+; LMULMAX2-RV64-NEXT: addiw a1, a1, -1366
+; LMULMAX2-RV64-NEXT: vand.vx v25, v25, a1
+; LMULMAX2-RV64-NEXT: vsrl.vi v25, v25, 1
+; LMULMAX2-RV64-NEXT: vor.vv v25, v25, v26
+; LMULMAX2-RV64-NEXT: vse16.v v25, (a0)
+; LMULMAX2-RV64-NEXT: ret
+;
+; LMULMAX1-RV32-LABEL: bitreverse_v8i16:
+; LMULMAX1-RV32: # %bb.0:
+; LMULMAX1-RV32-NEXT: vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vle16.v v25, (a0)
+; LMULMAX1-RV32-NEXT: vsrl.vi v26, v25, 8
+; LMULMAX1-RV32-NEXT: vsll.vi v25, v25, 8
+; LMULMAX1-RV32-NEXT: vor.vv v25, v25, v26
+; LMULMAX1-RV32-NEXT: lui a1, 1
+; LMULMAX1-RV32-NEXT: addi a1, a1, -241
+; LMULMAX1-RV32-NEXT: vand.vx v26, v25, a1
+; LMULMAX1-RV32-NEXT: vsll.vi v26, v26, 4
+; LMULMAX1-RV32-NEXT: lui a1, 15
+; LMULMAX1-RV32-NEXT: addi a1, a1, 240
+; LMULMAX1-RV32-NEXT: vand.vx v25, v25, a1
+; LMULMAX1-RV32-NEXT: vsrl.vi v25, v25, 4
+; LMULMAX1-RV32-NEXT: vor.vv v25, v25, v26
+; LMULMAX1-RV32-NEXT: lui a1, 3
+; LMULMAX1-RV32-NEXT: addi a1, a1, 819
+; LMULMAX1-RV32-NEXT: vand.vx v26, v25, a1
+; LMULMAX1-RV32-NEXT: vsll.vi v26, v26, 2
+; LMULMAX1-RV32-NEXT: lui a1, 13
+; LMULMAX1-RV32-NEXT: addi a1, a1, -820
+; LMULMAX1-RV32-NEXT: vand.vx v25, v25, a1
+; LMULMAX1-RV32-NEXT: vsrl.vi v25, v25, 2
+; LMULMAX1-RV32-NEXT: vor.vv v25, v25, v26
+; LMULMAX1-RV32-NEXT: lui a1, 5
+; LMULMAX1-RV32-NEXT: addi a1, a1, 1365
+; LMULMAX1-RV32-NEXT: vand.vx v26, v25, a1
+; LMULMAX1-RV32-NEXT: vsll.vi v26, v26, 1
+; LMULMAX1-RV32-NEXT: lui a1, 11
+; LMULMAX1-RV32-NEXT: addi a1, a1, -1366
+; LMULMAX1-RV32-NEXT: vand.vx v25, v25, a1
+; LMULMAX1-RV32-NEXT: vsrl.vi v25, v25, 1
+; LMULMAX1-RV32-NEXT: vor.vv v25, v25, v26
+; LMULMAX1-RV32-NEXT: vse16.v v25, (a0)
+; LMULMAX1-RV32-NEXT: ret
+;
+; LMULMAX1-RV64-LABEL: bitreverse_v8i16:
+; LMULMAX1-RV64: # %bb.0:
+; LMULMAX1-RV64-NEXT: vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX1-RV64-NEXT: vle16.v v25, (a0)
+; LMULMAX1-RV64-NEXT: vsrl.vi v26, v25, 8
+; LMULMAX1-RV64-NEXT: vsll.vi v25, v25, 8
+; LMULMAX1-RV64-NEXT: vor.vv v25, v25, v26
+; LMULMAX1-RV64-NEXT: lui a1, 1
+; LMULMAX1-RV64-NEXT: addiw a1, a1, -241
+; LMULMAX1-RV64-NEXT: vand.vx v26, v25, a1
+; LMULMAX1-RV64-NEXT: vsll.vi v26, v26, 4
+; LMULMAX1-RV64-NEXT: lui a1, 15
+; LMULMAX1-RV64-NEXT: addiw a1, a1, 240
+; LMULMAX1-RV64-NEXT: vand.vx v25, v25, a1
+; LMULMAX1-RV64-NEXT: vsrl.vi v25, v25, 4
+; LMULMAX1-RV64-NEXT: vor.vv v25, v25, v26
+; LMULMAX1-RV64-NEXT: lui a1, 3
+; LMULMAX1-RV64-NEXT: addiw a1, a1, 819
+; LMULMAX1-RV64-NEXT: vand.vx v26, v25, a1
+; LMULMAX1-RV64-NEXT: vsll.vi v26, v26, 2
+; LMULMAX1-RV64-NEXT: lui a1, 13
+; LMULMAX1-RV64-NEXT: addiw a1, a1, -820
+; LMULMAX1-RV64-NEXT: vand.vx v25, v25, a1
+; LMULMAX1-RV64-NEXT: vsrl.vi v25, v25, 2
+; LMULMAX1-RV64-NEXT: vor.vv v25, v25, v26
+; LMULMAX1-RV64-NEXT: lui a1, 5
+; LMULMAX1-RV64-NEXT: addiw a1, a1, 1365
+; LMULMAX1-RV64-NEXT: vand.vx v26, v25, a1
+; LMULMAX1-RV64-NEXT: vsll.vi v26, v26, 1
+; LMULMAX1-RV64-NEXT: lui a1, 11
+; LMULMAX1-RV64-NEXT: addiw a1, a1, -1366
+; LMULMAX1-RV64-NEXT: vand.vx v25, v25, a1
+; LMULMAX1-RV64-NEXT: vsrl.vi v25, v25, 1
+; LMULMAX1-RV64-NEXT: vor.vv v25, v25, v26
+; LMULMAX1-RV64-NEXT: vse16.v v25, (a0)
+; LMULMAX1-RV64-NEXT: ret
%a = load <8 x i16>, <8 x i16>* %x
%b = load <8 x i16>, <8 x i16>* %y
%c = call <8 x i16> @llvm.bitreverse.v8i16(<8 x i16> %a)
@@ -94,86 +161,201 @@ define void @bitreverse_v8i16(<8 x i16>* %x, <8 x i16>* %y) {
declare <8 x i16> @llvm.bitreverse.v8i16(<8 x i16>)
define void @bitreverse_v4i32(<4 x i32>* %x, <4 x i32>* %y) {
-; CHECK-LABEL: bitreverse_v4i32:
-; CHECK: # %bb.0:
-; CHECK-NEXT: addi sp, sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
-; CHECK-NEXT: vsetivli a1, 4, e32,m1,ta,mu
-; CHECK-NEXT: vle32.v v25, (a0)
-; CHECK-NEXT: vsetvli zero, zero, e8,m1,ta,mu
-; CHECK-NEXT: vmv.x.s a1, v25
-; CHECK-NEXT: sb a1, 3(sp)
-; CHECK-NEXT: vsetivli a1, 1, e8,m1,ta,mu
-; CHECK-NEXT: vslidedown.vi v26, v25, 12
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 15(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 13
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 14(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 14
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 13(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 15
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 12(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 8
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 11(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 9
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 10(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 10
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 9(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 11
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 8(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 4
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 7(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 5
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 6(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 6
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 5(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 7
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 4(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 1
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 2(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 2
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 1(sp)
-; CHECK-NEXT: vslidedown.vi v25, v25, 3
-; CHECK-NEXT: vmv.x.s a1, v25
-; CHECK-NEXT: sb a1, 0(sp)
-; CHECK-NEXT: vsetivli a1, 16, e8,m1,ta,mu
-; CHECK-NEXT: vle8.v v25, (sp)
-; CHECK-NEXT: vand.vi v26, v25, 15
-; CHECK-NEXT: vsll.vi v26, v26, 4
-; CHECK-NEXT: vand.vi v25, v25, -16
-; CHECK-NEXT: vsrl.vi v25, v25, 4
-; CHECK-NEXT: vor.vv v25, v25, v26
-; CHECK-NEXT: addi a1, zero, 51
-; CHECK-NEXT: vand.vx v26, v25, a1
-; CHECK-NEXT: vsll.vi v26, v26, 2
-; CHECK-NEXT: addi a1, zero, 204
-; CHECK-NEXT: vand.vx v25, v25, a1
-; CHECK-NEXT: vsrl.vi v25, v25, 2
-; CHECK-NEXT: vor.vv v25, v25, v26
-; CHECK-NEXT: addi a1, zero, 85
-; CHECK-NEXT: vand.vx v26, v25, a1
-; CHECK-NEXT: vsll.vi v26, v26, 1
-; CHECK-NEXT: addi a1, zero, 170
-; CHECK-NEXT: vand.vx v25, v25, a1
-; CHECK-NEXT: vsrl.vi v25, v25, 1
-; CHECK-NEXT: vor.vv v25, v25, v26
-; CHECK-NEXT: vsetivli a1, 4, e32,m1,ta,mu
-; CHECK-NEXT: vse32.v v25, (a0)
-; CHECK-NEXT: addi sp, sp, 16
-; CHECK-NEXT: ret
+; LMULMAX2-RV32-LABEL: bitreverse_v4i32:
+; LMULMAX2-RV32: # %bb.0:
+; LMULMAX2-RV32-NEXT: vsetivli a1, 4, e32,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vle32.v v25, (a0)
+; LMULMAX2-RV32-NEXT: vsrl.vi v26, v25, 8
+; LMULMAX2-RV32-NEXT: lui a1, 16
+; LMULMAX2-RV32-NEXT: addi a1, a1, -256
+; LMULMAX2-RV32-NEXT: vand.vx v26, v26, a1
+; LMULMAX2-RV32-NEXT: vsrl.vi v27, v25, 24
+; LMULMAX2-RV32-NEXT: vor.vv v26, v26, v27
+; LMULMAX2-RV32-NEXT: vsll.vi v27, v25, 8
+; LMULMAX2-RV32-NEXT: lui a1, 4080
+; LMULMAX2-RV32-NEXT: vand.vx v27, v27, a1
+; LMULMAX2-RV32-NEXT: vsll.vi v25, v25, 24
+; LMULMAX2-RV32-NEXT: vor.vv v25, v25, v27
+; LMULMAX2-RV32-NEXT: vor.vv v25, v25, v26
+; LMULMAX2-RV32-NEXT: lui a1, 61681
+; LMULMAX2-RV32-NEXT: addi a1, a1, -241
+; LMULMAX2-RV32-NEXT: vand.vx v26, v25, a1
+; LMULMAX2-RV32-NEXT: vsll.vi v26, v26, 4
+; LMULMAX2-RV32-NEXT: lui a1, 986895
+; LMULMAX2-RV32-NEXT: addi a1, a1, 240
+; LMULMAX2-RV32-NEXT: vand.vx v25, v25, a1
+; LMULMAX2-RV32-NEXT: vsrl.vi v25, v25, 4
+; LMULMAX2-RV32-NEXT: vor.vv v25, v25, v26
+; LMULMAX2-RV32-NEXT: lui a1, 209715
+; LMULMAX2-RV32-NEXT: addi a1, a1, 819
+; LMULMAX2-RV32-NEXT: vand.vx v26, v25, a1
+; LMULMAX2-RV32-NEXT: vsll.vi v26, v26, 2
+; LMULMAX2-RV32-NEXT: lui a1, 838861
+; LMULMAX2-RV32-NEXT: addi a1, a1, -820
+; LMULMAX2-RV32-NEXT: vand.vx v25, v25, a1
+; LMULMAX2-RV32-NEXT: vsrl.vi v25, v25, 2
+; LMULMAX2-RV32-NEXT: vor.vv v25, v25, v26
+; LMULMAX2-RV32-NEXT: lui a1, 349525
+; LMULMAX2-RV32-NEXT: addi a1, a1, 1365
+; LMULMAX2-RV32-NEXT: vand.vx v26, v25, a1
+; LMULMAX2-RV32-NEXT: vsll.vi v26, v26, 1
+; LMULMAX2-RV32-NEXT: lui a1, 699051
+; LMULMAX2-RV32-NEXT: addi a1, a1, -1366
+; LMULMAX2-RV32-NEXT: vand.vx v25, v25, a1
+; LMULMAX2-RV32-NEXT: vsrl.vi v25, v25, 1
+; LMULMAX2-RV32-NEXT: vor.vv v25, v25, v26
+; LMULMAX2-RV32-NEXT: vse32.v v25, (a0)
+; LMULMAX2-RV32-NEXT: ret
+;
+; LMULMAX2-RV64-LABEL: bitreverse_v4i32:
+; LMULMAX2-RV64: # %bb.0:
+; LMULMAX2-RV64-NEXT: vsetivli a1, 4, e32,m1,ta,mu
+; LMULMAX2-RV64-NEXT: vle32.v v25, (a0)
+; LMULMAX2-RV64-NEXT: vsrl.vi v26, v25, 8
+; LMULMAX2-RV64-NEXT: lui a1, 16
+; LMULMAX2-RV64-NEXT: addiw a1, a1, -256
+; LMULMAX2-RV64-NEXT: vand.vx v26, v26, a1
+; LMULMAX2-RV64-NEXT: vsrl.vi v27, v25, 24
+; LMULMAX2-RV64-NEXT: vor.vv v26, v26, v27
+; LMULMAX2-RV64-NEXT: vsll.vi v27, v25, 8
+; LMULMAX2-RV64-NEXT: lui a1, 4080
+; LMULMAX2-RV64-NEXT: vand.vx v27, v27, a1
+; LMULMAX2-RV64-NEXT: vsll.vi v25, v25, 24
+; LMULMAX2-RV64-NEXT: vor.vv v25, v25, v27
+; LMULMAX2-RV64-NEXT: vor.vv v25, v25, v26
+; LMULMAX2-RV64-NEXT: lui a1, 61681
+; LMULMAX2-RV64-NEXT: addiw a1, a1, -241
+; LMULMAX2-RV64-NEXT: vand.vx v26, v25, a1
+; LMULMAX2-RV64-NEXT: vsll.vi v26, v26, 4
+; LMULMAX2-RV64-NEXT: lui a1, 241
+; LMULMAX2-RV64-NEXT: addiw a1, a1, -241
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, 240
+; LMULMAX2-RV64-NEXT: vand.vx v25, v25, a1
+; LMULMAX2-RV64-NEXT: vsrl.vi v25, v25, 4
+; LMULMAX2-RV64-NEXT: vor.vv v25, v25, v26
+; LMULMAX2-RV64-NEXT: lui a1, 209715
+; LMULMAX2-RV64-NEXT: addiw a1, a1, 819
+; LMULMAX2-RV64-NEXT: vand.vx v26, v25, a1
+; LMULMAX2-RV64-NEXT: vsll.vi v26, v26, 2
+; LMULMAX2-RV64-NEXT: lui a1, 205
+; LMULMAX2-RV64-NEXT: addiw a1, a1, -819
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, -820
+; LMULMAX2-RV64-NEXT: vand.vx v25, v25, a1
+; LMULMAX2-RV64-NEXT: vsrl.vi v25, v25, 2
+; LMULMAX2-RV64-NEXT: vor.vv v25, v25, v26
+; LMULMAX2-RV64-NEXT: lui a1, 349525
+; LMULMAX2-RV64-NEXT: addiw a1, a1, 1365
+; LMULMAX2-RV64-NEXT: vand.vx v26, v25, a1
+; LMULMAX2-RV64-NEXT: vsll.vi v26, v26, 1
+; LMULMAX2-RV64-NEXT: lui a1, 171
+; LMULMAX2-RV64-NEXT: addiw a1, a1, -1365
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, -1366
+; LMULMAX2-RV64-NEXT: vand.vx v25, v25, a1
+; LMULMAX2-RV64-NEXT: vsrl.vi v25, v25, 1
+; LMULMAX2-RV64-NEXT: vor.vv v25, v25, v26
+; LMULMAX2-RV64-NEXT: vse32.v v25, (a0)
+; LMULMAX2-RV64-NEXT: ret
+;
+; LMULMAX1-RV32-LABEL: bitreverse_v4i32:
+; LMULMAX1-RV32: # %bb.0:
+; LMULMAX1-RV32-NEXT: vsetivli a1, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vle32.v v25, (a0)
+; LMULMAX1-RV32-NEXT: vsrl.vi v26, v25, 8
+; LMULMAX1-RV32-NEXT: lui a1, 16
+; LMULMAX1-RV32-NEXT: addi a1, a1, -256
+; LMULMAX1-RV32-NEXT: vand.vx v26, v26, a1
+; LMULMAX1-RV32-NEXT: vsrl.vi v27, v25, 24
+; LMULMAX1-RV32-NEXT: vor.vv v26, v26, v27
+; LMULMAX1-RV32-NEXT: vsll.vi v27, v25, 8
+; LMULMAX1-RV32-NEXT: lui a1, 4080
+; LMULMAX1-RV32-NEXT: vand.vx v27, v27, a1
+; LMULMAX1-RV32-NEXT: vsll.vi v25, v25, 24
+; LMULMAX1-RV32-NEXT: vor.vv v25, v25, v27
+; LMULMAX1-RV32-NEXT: vor.vv v25, v25, v26
+; LMULMAX1-RV32-NEXT: lui a1, 61681
+; LMULMAX1-RV32-NEXT: addi a1, a1, -241
+; LMULMAX1-RV32-NEXT: vand.vx v26, v25, a1
+; LMULMAX1-RV32-NEXT: vsll.vi v26, v26, 4
+; LMULMAX1-RV32-NEXT: lui a1, 986895
+; LMULMAX1-RV32-NEXT: addi a1, a1, 240
+; LMULMAX1-RV32-NEXT: vand.vx v25, v25, a1
+; LMULMAX1-RV32-NEXT: vsrl.vi v25, v25, 4
+; LMULMAX1-RV32-NEXT: vor.vv v25, v25, v26
+; LMULMAX1-RV32-NEXT: lui a1, 209715
+; LMULMAX1-RV32-NEXT: addi a1, a1, 819
+; LMULMAX1-RV32-NEXT: vand.vx v26, v25, a1
+; LMULMAX1-RV32-NEXT: vsll.vi v26, v26, 2
+; LMULMAX1-RV32-NEXT: lui a1, 838861
+; LMULMAX1-RV32-NEXT: addi a1, a1, -820
+; LMULMAX1-RV32-NEXT: vand.vx v25, v25, a1
+; LMULMAX1-RV32-NEXT: vsrl.vi v25, v25, 2
+; LMULMAX1-RV32-NEXT: vor.vv v25, v25, v26
+; LMULMAX1-RV32-NEXT: lui a1, 349525
+; LMULMAX1-RV32-NEXT: addi a1, a1, 1365
+; LMULMAX1-RV32-NEXT: vand.vx v26, v25, a1
+; LMULMAX1-RV32-NEXT: vsll.vi v26, v26, 1
+; LMULMAX1-RV32-NEXT: lui a1, 699051
+; LMULMAX1-RV32-NEXT: addi a1, a1, -1366
+; LMULMAX1-RV32-NEXT: vand.vx v25, v25, a1
+; LMULMAX1-RV32-NEXT: vsrl.vi v25, v25, 1
+; LMULMAX1-RV32-NEXT: vor.vv v25, v25, v26
+; LMULMAX1-RV32-NEXT: vse32.v v25, (a0)
+; LMULMAX1-RV32-NEXT: ret
+;
+; LMULMAX1-RV64-LABEL: bitreverse_v4i32:
+; LMULMAX1-RV64: # %bb.0:
+; LMULMAX1-RV64-NEXT: vsetivli a1, 4, e32,m1,ta,mu
+; LMULMAX1-RV64-NEXT: vle32.v v25, (a0)
+; LMULMAX1-RV64-NEXT: vsrl.vi v26, v25, 8
+; LMULMAX1-RV64-NEXT: lui a1, 16
+; LMULMAX1-RV64-NEXT: addiw a1, a1, -256
+; LMULMAX1-RV64-NEXT: vand.vx v26, v26, a1
+; LMULMAX1-RV64-NEXT: vsrl.vi v27, v25, 24
+; LMULMAX1-RV64-NEXT: vor.vv v26, v26, v27
+; LMULMAX1-RV64-NEXT: vsll.vi v27, v25, 8
+; LMULMAX1-RV64-NEXT: lui a1, 4080
+; LMULMAX1-RV64-NEXT: vand.vx v27, v27, a1
+; LMULMAX1-RV64-NEXT: vsll.vi v25, v25, 24
+; LMULMAX1-RV64-NEXT: vor.vv v25, v25, v27
+; LMULMAX1-RV64-NEXT: vor.vv v25, v25, v26
+; LMULMAX1-RV64-NEXT: lui a1, 61681
+; LMULMAX1-RV64-NEXT: addiw a1, a1, -241
+; LMULMAX1-RV64-NEXT: vand.vx v26, v25, a1
+; LMULMAX1-RV64-NEXT: vsll.vi v26, v26, 4
+; LMULMAX1-RV64-NEXT: lui a1, 241
+; LMULMAX1-RV64-NEXT: addiw a1, a1, -241
+; LMULMAX1-RV64-NEXT: slli a1, a1, 12
+; LMULMAX1-RV64-NEXT: addi a1, a1, 240
+; LMULMAX1-RV64-NEXT: vand.vx v25, v25, a1
+; LMULMAX1-RV64-NEXT: vsrl.vi v25, v25, 4
+; LMULMAX1-RV64-NEXT: vor.vv v25, v25, v26
+; LMULMAX1-RV64-NEXT: lui a1, 209715
+; LMULMAX1-RV64-NEXT: addiw a1, a1, 819
+; LMULMAX1-RV64-NEXT: vand.vx v26, v25, a1
+; LMULMAX1-RV64-NEXT: vsll.vi v26, v26, 2
+; LMULMAX1-RV64-NEXT: lui a1, 205
+; LMULMAX1-RV64-NEXT: addiw a1, a1, -819
+; LMULMAX1-RV64-NEXT: slli a1, a1, 12
+; LMULMAX1-RV64-NEXT: addi a1, a1, -820
+; LMULMAX1-RV64-NEXT: vand.vx v25, v25, a1
+; LMULMAX1-RV64-NEXT: vsrl.vi v25, v25, 2
+; LMULMAX1-RV64-NEXT: vor.vv v25, v25, v26
+; LMULMAX1-RV64-NEXT: lui a1, 349525
+; LMULMAX1-RV64-NEXT: addiw a1, a1, 1365
+; LMULMAX1-RV64-NEXT: vand.vx v26, v25, a1
+; LMULMAX1-RV64-NEXT: vsll.vi v26, v26, 1
+; LMULMAX1-RV64-NEXT: lui a1, 171
+; LMULMAX1-RV64-NEXT: addiw a1, a1, -1365
+; LMULMAX1-RV64-NEXT: slli a1, a1, 12
+; LMULMAX1-RV64-NEXT: addi a1, a1, -1366
+; LMULMAX1-RV64-NEXT: vand.vx v25, v25, a1
+; LMULMAX1-RV64-NEXT: vsrl.vi v25, v25, 1
+; LMULMAX1-RV64-NEXT: vor.vv v25, v25, v26
+; LMULMAX1-RV64-NEXT: vse32.v v25, (a0)
+; LMULMAX1-RV64-NEXT: ret
%a = load <4 x i32>, <4 x i32>* %x
%b = load <4 x i32>, <4 x i32>* %y
%c = call <4 x i32> @llvm.bitreverse.v4i32(<4 x i32> %a)
@@ -183,86 +365,483 @@ define void @bitreverse_v4i32(<4 x i32>* %x, <4 x i32>* %y) {
declare <4 x i32> @llvm.bitreverse.v4i32(<4 x i32>)
define void @bitreverse_v2i64(<2 x i64>* %x, <2 x i64>* %y) {
-; CHECK-LABEL: bitreverse_v2i64:
-; CHECK: # %bb.0:
-; CHECK-NEXT: addi sp, sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
-; CHECK-NEXT: vsetivli a1, 2, e64,m1,ta,mu
-; CHECK-NEXT: vle64.v v25, (a0)
-; CHECK-NEXT: vsetvli zero, zero, e8,m1,ta,mu
-; CHECK-NEXT: vmv.x.s a1, v25
-; CHECK-NEXT: sb a1, 7(sp)
-; CHECK-NEXT: vsetivli a1, 1, e8,m1,ta,mu
-; CHECK-NEXT: vslidedown.vi v26, v25, 8
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 15(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 9
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 14(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 10
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 13(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 11
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 12(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 12
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 11(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 13
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 10(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 14
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 9(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 15
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 8(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 1
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 6(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 2
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 5(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 3
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 4(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 4
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 3(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 5
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 2(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 6
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 1(sp)
-; CHECK-NEXT: vslidedown.vi v25, v25, 7
-; CHECK-NEXT: vmv.x.s a1, v25
-; CHECK-NEXT: sb a1, 0(sp)
-; CHECK-NEXT: vsetivli a1, 16, e8,m1,ta,mu
-; CHECK-NEXT: vle8.v v25, (sp)
-; CHECK-NEXT: vand.vi v26, v25, 15
-; CHECK-NEXT: vsll.vi v26, v26, 4
-; CHECK-NEXT: vand.vi v25, v25, -16
-; CHECK-NEXT: vsrl.vi v25, v25, 4
-; CHECK-NEXT: vor.vv v25, v25, v26
-; CHECK-NEXT: addi a1, zero, 51
-; CHECK-NEXT: vand.vx v26, v25, a1
-; CHECK-NEXT: vsll.vi v26, v26, 2
-; CHECK-NEXT: addi a1, zero, 204
-; CHECK-NEXT: vand.vx v25, v25, a1
-; CHECK-NEXT: vsrl.vi v25, v25, 2
-; CHECK-NEXT: vor.vv v25, v25, v26
-; CHECK-NEXT: addi a1, zero, 85
-; CHECK-NEXT: vand.vx v26, v25, a1
-; CHECK-NEXT: vsll.vi v26, v26, 1
-; CHECK-NEXT: addi a1, zero, 170
-; CHECK-NEXT: vand.vx v25, v25, a1
-; CHECK-NEXT: vsrl.vi v25, v25, 1
-; CHECK-NEXT: vor.vv v25, v25, v26
-; CHECK-NEXT: vsetivli a1, 2, e64,m1,ta,mu
-; CHECK-NEXT: vse64.v v25, (a0)
-; CHECK-NEXT: addi sp, sp, 16
-; CHECK-NEXT: ret
+; LMULMAX2-RV32-LABEL: bitreverse_v2i64:
+; LMULMAX2-RV32: # %bb.0:
+; LMULMAX2-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vle64.v v25, (a0)
+; LMULMAX2-RV32-NEXT: lui a1, %hi(.LCPI2_0)
+; LMULMAX2-RV32-NEXT: addi a1, a1, %lo(.LCPI2_0)
+; LMULMAX2-RV32-NEXT: vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vle32.v v26, (a1)
+; LMULMAX2-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vsrl.vv v27, v25, v26
+; LMULMAX2-RV32-NEXT: lui a1, %hi(.LCPI2_1)
+; LMULMAX2-RV32-NEXT: addi a1, a1, %lo(.LCPI2_1)
+; LMULMAX2-RV32-NEXT: vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vle32.v v28, (a1)
+; LMULMAX2-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vsrl.vv v29, v25, v28
+; LMULMAX2-RV32-NEXT: lui a1, %hi(.LCPI2_2)
+; LMULMAX2-RV32-NEXT: addi a1, a1, %lo(.LCPI2_2)
+; LMULMAX2-RV32-NEXT: vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vle32.v v30, (a1)
+; LMULMAX2-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vand.vv v29, v29, v30
+; LMULMAX2-RV32-NEXT: vor.vv v27, v29, v27
+; LMULMAX2-RV32-NEXT: lui a1, %hi(.LCPI2_3)
+; LMULMAX2-RV32-NEXT: addi a1, a1, %lo(.LCPI2_3)
+; LMULMAX2-RV32-NEXT: vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vle32.v v29, (a1)
+; LMULMAX2-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vsrl.vv v30, v25, v29
+; LMULMAX2-RV32-NEXT: lui a1, %hi(.LCPI2_4)
+; LMULMAX2-RV32-NEXT: addi a1, a1, %lo(.LCPI2_4)
+; LMULMAX2-RV32-NEXT: vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vle32.v v31, (a1)
+; LMULMAX2-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vand.vv v30, v30, v31
+; LMULMAX2-RV32-NEXT: lui a1, %hi(.LCPI2_5)
+; LMULMAX2-RV32-NEXT: addi a1, a1, %lo(.LCPI2_5)
+; LMULMAX2-RV32-NEXT: vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vle32.v v31, (a1)
+; LMULMAX2-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vsrl.vv v8, v25, v31
+; LMULMAX2-RV32-NEXT: lui a1, %hi(.LCPI2_6)
+; LMULMAX2-RV32-NEXT: addi a1, a1, %lo(.LCPI2_6)
+; LMULMAX2-RV32-NEXT: vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vle32.v v9, (a1)
+; LMULMAX2-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vand.vv v8, v8, v9
+; LMULMAX2-RV32-NEXT: vor.vv v30, v8, v30
+; LMULMAX2-RV32-NEXT: vor.vv v27, v30, v27
+; LMULMAX2-RV32-NEXT: vsll.vv v30, v25, v31
+; LMULMAX2-RV32-NEXT: lui a1, %hi(.LCPI2_7)
+; LMULMAX2-RV32-NEXT: addi a1, a1, %lo(.LCPI2_7)
+; LMULMAX2-RV32-NEXT: vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vle32.v v31, (a1)
+; LMULMAX2-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vand.vv v30, v30, v31
+; LMULMAX2-RV32-NEXT: vsll.vv v29, v25, v29
+; LMULMAX2-RV32-NEXT: lui a1, %hi(.LCPI2_8)
+; LMULMAX2-RV32-NEXT: addi a1, a1, %lo(.LCPI2_8)
+; LMULMAX2-RV32-NEXT: vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vle32.v v31, (a1)
+; LMULMAX2-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vand.vv v29, v29, v31
+; LMULMAX2-RV32-NEXT: vor.vv v29, v29, v30
+; LMULMAX2-RV32-NEXT: vsll.vv v26, v25, v26
+; LMULMAX2-RV32-NEXT: vsll.vv v25, v25, v28
+; LMULMAX2-RV32-NEXT: lui a1, %hi(.LCPI2_9)
+; LMULMAX2-RV32-NEXT: addi a1, a1, %lo(.LCPI2_9)
+; LMULMAX2-RV32-NEXT: vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vle32.v v28, (a1)
+; LMULMAX2-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vand.vv v25, v25, v28
+; LMULMAX2-RV32-NEXT: vor.vv v25, v26, v25
+; LMULMAX2-RV32-NEXT: vor.vv v25, v25, v29
+; LMULMAX2-RV32-NEXT: vor.vv v25, v25, v27
+; LMULMAX2-RV32-NEXT: lui a1, 61681
+; LMULMAX2-RV32-NEXT: addi a1, a1, -241
+; LMULMAX2-RV32-NEXT: vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vmv.v.x v26, a1
+; LMULMAX2-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vand.vv v26, v25, v26
+; LMULMAX2-RV32-NEXT: lui a1, %hi(.LCPI2_10)
+; LMULMAX2-RV32-NEXT: addi a1, a1, %lo(.LCPI2_10)
+; LMULMAX2-RV32-NEXT: vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vle32.v v27, (a1)
+; LMULMAX2-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vsll.vv v26, v26, v27
+; LMULMAX2-RV32-NEXT: lui a1, 986895
+; LMULMAX2-RV32-NEXT: addi a1, a1, 240
+; LMULMAX2-RV32-NEXT: vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vmv.v.x v28, a1
+; LMULMAX2-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vand.vv v25, v25, v28
+; LMULMAX2-RV32-NEXT: vsrl.vv v25, v25, v27
+; LMULMAX2-RV32-NEXT: vor.vv v25, v25, v26
+; LMULMAX2-RV32-NEXT: lui a1, 209715
+; LMULMAX2-RV32-NEXT: addi a1, a1, 819
+; LMULMAX2-RV32-NEXT: vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vmv.v.x v26, a1
+; LMULMAX2-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vand.vv v26, v25, v26
+; LMULMAX2-RV32-NEXT: lui a1, %hi(.LCPI2_11)
+; LMULMAX2-RV32-NEXT: addi a1, a1, %lo(.LCPI2_11)
+; LMULMAX2-RV32-NEXT: vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vle32.v v27, (a1)
+; LMULMAX2-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vsll.vv v26, v26, v27
+; LMULMAX2-RV32-NEXT: lui a1, 838861
+; LMULMAX2-RV32-NEXT: addi a1, a1, -820
+; LMULMAX2-RV32-NEXT: vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vmv.v.x v28, a1
+; LMULMAX2-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vand.vv v25, v25, v28
+; LMULMAX2-RV32-NEXT: vsrl.vv v25, v25, v27
+; LMULMAX2-RV32-NEXT: vor.vv v25, v25, v26
+; LMULMAX2-RV32-NEXT: lui a1, 349525
+; LMULMAX2-RV32-NEXT: addi a1, a1, 1365
+; LMULMAX2-RV32-NEXT: vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vmv.v.x v26, a1
+; LMULMAX2-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vand.vv v26, v25, v26
+; LMULMAX2-RV32-NEXT: lui a1, %hi(.LCPI2_12)
+; LMULMAX2-RV32-NEXT: addi a1, a1, %lo(.LCPI2_12)
+; LMULMAX2-RV32-NEXT: vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vle32.v v27, (a1)
+; LMULMAX2-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vsll.vv v26, v26, v27
+; LMULMAX2-RV32-NEXT: lui a1, 699051
+; LMULMAX2-RV32-NEXT: addi a1, a1, -1366
+; LMULMAX2-RV32-NEXT: vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vmv.v.x v28, a1
+; LMULMAX2-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vand.vv v25, v25, v28
+; LMULMAX2-RV32-NEXT: vsrl.vv v25, v25, v27
+; LMULMAX2-RV32-NEXT: vor.vv v25, v25, v26
+; LMULMAX2-RV32-NEXT: vse64.v v25, (a0)
+; LMULMAX2-RV32-NEXT: ret
+;
+; LMULMAX2-RV64-LABEL: bitreverse_v2i64:
+; LMULMAX2-RV64: # %bb.0:
+; LMULMAX2-RV64-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX2-RV64-NEXT: vle64.v v25, (a0)
+; LMULMAX2-RV64-NEXT: addi a1, zero, 56
+; LMULMAX2-RV64-NEXT: vsrl.vx v26, v25, a1
+; LMULMAX2-RV64-NEXT: addi a2, zero, 40
+; LMULMAX2-RV64-NEXT: vsrl.vx v27, v25, a2
+; LMULMAX2-RV64-NEXT: lui a3, 16
+; LMULMAX2-RV64-NEXT: addiw a3, a3, -256
+; LMULMAX2-RV64-NEXT: vand.vx v27, v27, a3
+; LMULMAX2-RV64-NEXT: vor.vv v26, v27, v26
+; LMULMAX2-RV64-NEXT: vsrl.vi v27, v25, 24
+; LMULMAX2-RV64-NEXT: lui a3, 4080
+; LMULMAX2-RV64-NEXT: vand.vx v27, v27, a3
+; LMULMAX2-RV64-NEXT: vsrl.vi v28, v25, 8
+; LMULMAX2-RV64-NEXT: addi a3, zero, 255
+; LMULMAX2-RV64-NEXT: slli a4, a3, 24
+; LMULMAX2-RV64-NEXT: vand.vx v28, v28, a4
+; LMULMAX2-RV64-NEXT: vor.vv v27, v28, v27
+; LMULMAX2-RV64-NEXT: vor.vv v26, v27, v26
+; LMULMAX2-RV64-NEXT: vsll.vi v27, v25, 8
+; LMULMAX2-RV64-NEXT: slli a4, a3, 32
+; LMULMAX2-RV64-NEXT: vand.vx v27, v27, a4
+; LMULMAX2-RV64-NEXT: vsll.vi v28, v25, 24
+; LMULMAX2-RV64-NEXT: slli a4, a3, 40
+; LMULMAX2-RV64-NEXT: vand.vx v28, v28, a4
+; LMULMAX2-RV64-NEXT: vor.vv v27, v28, v27
+; LMULMAX2-RV64-NEXT: vsll.vx v28, v25, a1
+; LMULMAX2-RV64-NEXT: vsll.vx v25, v25, a2
+; LMULMAX2-RV64-NEXT: slli a1, a3, 48
+; LMULMAX2-RV64-NEXT: vand.vx v25, v25, a1
+; LMULMAX2-RV64-NEXT: vor.vv v25, v28, v25
+; LMULMAX2-RV64-NEXT: vor.vv v25, v25, v27
+; LMULMAX2-RV64-NEXT: vor.vv v25, v25, v26
+; LMULMAX2-RV64-NEXT: lui a1, 3855
+; LMULMAX2-RV64-NEXT: addiw a1, a1, 241
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, -241
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, 241
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, -241
+; LMULMAX2-RV64-NEXT: vand.vx v26, v25, a1
+; LMULMAX2-RV64-NEXT: vsll.vi v26, v26, 4
+; LMULMAX2-RV64-NEXT: lui a1, 1044721
+; LMULMAX2-RV64-NEXT: addiw a1, a1, -241
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, 241
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, -241
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, 240
+; LMULMAX2-RV64-NEXT: vand.vx v25, v25, a1
+; LMULMAX2-RV64-NEXT: vsrl.vi v25, v25, 4
+; LMULMAX2-RV64-NEXT: vor.vv v25, v25, v26
+; LMULMAX2-RV64-NEXT: lui a1, 13107
+; LMULMAX2-RV64-NEXT: addiw a1, a1, 819
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, 819
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, 819
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, 819
+; LMULMAX2-RV64-NEXT: vand.vx v26, v25, a1
+; LMULMAX2-RV64-NEXT: vsll.vi v26, v26, 2
+; LMULMAX2-RV64-NEXT: lui a1, 1035469
+; LMULMAX2-RV64-NEXT: addiw a1, a1, -819
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, -819
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, -819
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, -820
+; LMULMAX2-RV64-NEXT: vand.vx v25, v25, a1
+; LMULMAX2-RV64-NEXT: vsrl.vi v25, v25, 2
+; LMULMAX2-RV64-NEXT: vor.vv v25, v25, v26
+; LMULMAX2-RV64-NEXT: lui a1, 21845
+; LMULMAX2-RV64-NEXT: addiw a1, a1, 1365
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, 1365
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, 1365
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, 1365
+; LMULMAX2-RV64-NEXT: vand.vx v26, v25, a1
+; LMULMAX2-RV64-NEXT: vsll.vi v26, v26, 1
+; LMULMAX2-RV64-NEXT: lui a1, 1026731
+; LMULMAX2-RV64-NEXT: addiw a1, a1, -1365
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, -1365
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, -1365
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, -1366
+; LMULMAX2-RV64-NEXT: vand.vx v25, v25, a1
+; LMULMAX2-RV64-NEXT: vsrl.vi v25, v25, 1
+; LMULMAX2-RV64-NEXT: vor.vv v25, v25, v26
+; LMULMAX2-RV64-NEXT: vse64.v v25, (a0)
+; LMULMAX2-RV64-NEXT: ret
+;
+; LMULMAX1-RV32-LABEL: bitreverse_v2i64:
+; LMULMAX1-RV32: # %bb.0:
+; LMULMAX1-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vle64.v v25, (a0)
+; LMULMAX1-RV32-NEXT: lui a1, %hi(.LCPI2_0)
+; LMULMAX1-RV32-NEXT: addi a1, a1, %lo(.LCPI2_0)
+; LMULMAX1-RV32-NEXT: vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vle32.v v26, (a1)
+; LMULMAX1-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vsrl.vv v27, v25, v26
+; LMULMAX1-RV32-NEXT: lui a1, %hi(.LCPI2_1)
+; LMULMAX1-RV32-NEXT: addi a1, a1, %lo(.LCPI2_1)
+; LMULMAX1-RV32-NEXT: vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vle32.v v28, (a1)
+; LMULMAX1-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vsrl.vv v29, v25, v28
+; LMULMAX1-RV32-NEXT: lui a1, %hi(.LCPI2_2)
+; LMULMAX1-RV32-NEXT: addi a1, a1, %lo(.LCPI2_2)
+; LMULMAX1-RV32-NEXT: vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vle32.v v30, (a1)
+; LMULMAX1-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vand.vv v29, v29, v30
+; LMULMAX1-RV32-NEXT: vor.vv v27, v29, v27
+; LMULMAX1-RV32-NEXT: lui a1, %hi(.LCPI2_3)
+; LMULMAX1-RV32-NEXT: addi a1, a1, %lo(.LCPI2_3)
+; LMULMAX1-RV32-NEXT: vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vle32.v v29, (a1)
+; LMULMAX1-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vsrl.vv v30, v25, v29
+; LMULMAX1-RV32-NEXT: lui a1, %hi(.LCPI2_4)
+; LMULMAX1-RV32-NEXT: addi a1, a1, %lo(.LCPI2_4)
+; LMULMAX1-RV32-NEXT: vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vle32.v v31, (a1)
+; LMULMAX1-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vand.vv v30, v30, v31
+; LMULMAX1-RV32-NEXT: lui a1, %hi(.LCPI2_5)
+; LMULMAX1-RV32-NEXT: addi a1, a1, %lo(.LCPI2_5)
+; LMULMAX1-RV32-NEXT: vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vle32.v v31, (a1)
+; LMULMAX1-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vsrl.vv v8, v25, v31
+; LMULMAX1-RV32-NEXT: lui a1, %hi(.LCPI2_6)
+; LMULMAX1-RV32-NEXT: addi a1, a1, %lo(.LCPI2_6)
+; LMULMAX1-RV32-NEXT: vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vle32.v v9, (a1)
+; LMULMAX1-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vand.vv v8, v8, v9
+; LMULMAX1-RV32-NEXT: vor.vv v30, v8, v30
+; LMULMAX1-RV32-NEXT: vor.vv v27, v30, v27
+; LMULMAX1-RV32-NEXT: vsll.vv v30, v25, v31
+; LMULMAX1-RV32-NEXT: lui a1, %hi(.LCPI2_7)
+; LMULMAX1-RV32-NEXT: addi a1, a1, %lo(.LCPI2_7)
+; LMULMAX1-RV32-NEXT: vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vle32.v v31, (a1)
+; LMULMAX1-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vand.vv v30, v30, v31
+; LMULMAX1-RV32-NEXT: vsll.vv v29, v25, v29
+; LMULMAX1-RV32-NEXT: lui a1, %hi(.LCPI2_8)
+; LMULMAX1-RV32-NEXT: addi a1, a1, %lo(.LCPI2_8)
+; LMULMAX1-RV32-NEXT: vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vle32.v v31, (a1)
+; LMULMAX1-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vand.vv v29, v29, v31
+; LMULMAX1-RV32-NEXT: vor.vv v29, v29, v30
+; LMULMAX1-RV32-NEXT: vsll.vv v26, v25, v26
+; LMULMAX1-RV32-NEXT: vsll.vv v25, v25, v28
+; LMULMAX1-RV32-NEXT: lui a1, %hi(.LCPI2_9)
+; LMULMAX1-RV32-NEXT: addi a1, a1, %lo(.LCPI2_9)
+; LMULMAX1-RV32-NEXT: vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vle32.v v28, (a1)
+; LMULMAX1-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vand.vv v25, v25, v28
+; LMULMAX1-RV32-NEXT: vor.vv v25, v26, v25
+; LMULMAX1-RV32-NEXT: vor.vv v25, v25, v29
+; LMULMAX1-RV32-NEXT: vor.vv v25, v25, v27
+; LMULMAX1-RV32-NEXT: lui a1, 61681
+; LMULMAX1-RV32-NEXT: addi a1, a1, -241
+; LMULMAX1-RV32-NEXT: vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vmv.v.x v26, a1
+; LMULMAX1-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vand.vv v26, v25, v26
+; LMULMAX1-RV32-NEXT: lui a1, %hi(.LCPI2_10)
+; LMULMAX1-RV32-NEXT: addi a1, a1, %lo(.LCPI2_10)
+; LMULMAX1-RV32-NEXT: vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vle32.v v27, (a1)
+; LMULMAX1-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vsll.vv v26, v26, v27
+; LMULMAX1-RV32-NEXT: lui a1, 986895
+; LMULMAX1-RV32-NEXT: addi a1, a1, 240
+; LMULMAX1-RV32-NEXT: vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vmv.v.x v28, a1
+; LMULMAX1-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vand.vv v25, v25, v28
+; LMULMAX1-RV32-NEXT: vsrl.vv v25, v25, v27
+; LMULMAX1-RV32-NEXT: vor.vv v25, v25, v26
+; LMULMAX1-RV32-NEXT: lui a1, 209715
+; LMULMAX1-RV32-NEXT: addi a1, a1, 819
+; LMULMAX1-RV32-NEXT: vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vmv.v.x v26, a1
+; LMULMAX1-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vand.vv v26, v25, v26
+; LMULMAX1-RV32-NEXT: lui a1, %hi(.LCPI2_11)
+; LMULMAX1-RV32-NEXT: addi a1, a1, %lo(.LCPI2_11)
+; LMULMAX1-RV32-NEXT: vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vle32.v v27, (a1)
+; LMULMAX1-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vsll.vv v26, v26, v27
+; LMULMAX1-RV32-NEXT: lui a1, 838861
+; LMULMAX1-RV32-NEXT: addi a1, a1, -820
+; LMULMAX1-RV32-NEXT: vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vmv.v.x v28, a1
+; LMULMAX1-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vand.vv v25, v25, v28
+; LMULMAX1-RV32-NEXT: vsrl.vv v25, v25, v27
+; LMULMAX1-RV32-NEXT: vor.vv v25, v25, v26
+; LMULMAX1-RV32-NEXT: lui a1, 349525
+; LMULMAX1-RV32-NEXT: addi a1, a1, 1365
+; LMULMAX1-RV32-NEXT: vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vmv.v.x v26, a1
+; LMULMAX1-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vand.vv v26, v25, v26
+; LMULMAX1-RV32-NEXT: lui a1, %hi(.LCPI2_12)
+; LMULMAX1-RV32-NEXT: addi a1, a1, %lo(.LCPI2_12)
+; LMULMAX1-RV32-NEXT: vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vle32.v v27, (a1)
+; LMULMAX1-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vsll.vv v26, v26, v27
+; LMULMAX1-RV32-NEXT: lui a1, 699051
+; LMULMAX1-RV32-NEXT: addi a1, a1, -1366
+; LMULMAX1-RV32-NEXT: vsetivli a2, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vmv.v.x v28, a1
+; LMULMAX1-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vand.vv v25, v25, v28
+; LMULMAX1-RV32-NEXT: vsrl.vv v25, v25, v27
+; LMULMAX1-RV32-NEXT: vor.vv v25, v25, v26
+; LMULMAX1-RV32-NEXT: vse64.v v25, (a0)
+; LMULMAX1-RV32-NEXT: ret
+;
+; LMULMAX1-RV64-LABEL: bitreverse_v2i64:
+; LMULMAX1-RV64: # %bb.0:
+; LMULMAX1-RV64-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX1-RV64-NEXT: vle64.v v25, (a0)
+; LMULMAX1-RV64-NEXT: addi a1, zero, 56
+; LMULMAX1-RV64-NEXT: vsrl.vx v26, v25, a1
+; LMULMAX1-RV64-NEXT: addi a2, zero, 40
+; LMULMAX1-RV64-NEXT: vsrl.vx v27, v25, a2
+; LMULMAX1-RV64-NEXT: lui a3, 16
+; LMULMAX1-RV64-NEXT: addiw a3, a3, -256
+; LMULMAX1-RV64-NEXT: vand.vx v27, v27, a3
+; LMULMAX1-RV64-NEXT: vor.vv v26, v27, v26
+; LMULMAX1-RV64-NEXT: vsrl.vi v27, v25, 24
+; LMULMAX1-RV64-NEXT: lui a3, 4080
+; LMULMAX1-RV64-NEXT: vand.vx v27, v27, a3
+; LMULMAX1-RV64-NEXT: vsrl.vi v28, v25, 8
+; LMULMAX1-RV64-NEXT: addi a3, zero, 255
+; LMULMAX1-RV64-NEXT: slli a4, a3, 24
+; LMULMAX1-RV64-NEXT: vand.vx v28, v28, a4
+; LMULMAX1-RV64-NEXT: vor.vv v27, v28, v27
+; LMULMAX1-RV64-NEXT: vor.vv v26, v27, v26
+; LMULMAX1-RV64-NEXT: vsll.vi v27, v25, 8
+; LMULMAX1-RV64-NEXT: slli a4, a3, 32
+; LMULMAX1-RV64-NEXT: vand.vx v27, v27, a4
+; LMULMAX1-RV64-NEXT: vsll.vi v28, v25, 24
+; LMULMAX1-RV64-NEXT: slli a4, a3, 40
+; LMULMAX1-RV64-NEXT: vand.vx v28, v28, a4
+; LMULMAX1-RV64-NEXT: vor.vv v27, v28, v27
+; LMULMAX1-RV64-NEXT: vsll.vx v28, v25, a1
+; LMULMAX1-RV64-NEXT: vsll.vx v25, v25, a2
+; LMULMAX1-RV64-NEXT: slli a1, a3, 48
+; LMULMAX1-RV64-NEXT: vand.vx v25, v25, a1
+; LMULMAX1-RV64-NEXT: vor.vv v25, v28, v25
+; LMULMAX1-RV64-NEXT: vor.vv v25, v25, v27
+; LMULMAX1-RV64-NEXT: vor.vv v25, v25, v26
+; LMULMAX1-RV64-NEXT: lui a1, 3855
+; LMULMAX1-RV64-NEXT: addiw a1, a1, 241
+; LMULMAX1-RV64-NEXT: slli a1, a1, 12
+; LMULMAX1-RV64-NEXT: addi a1, a1, -241
+; LMULMAX1-RV64-NEXT: slli a1, a1, 12
+; LMULMAX1-RV64-NEXT: addi a1, a1, 241
+; LMULMAX1-RV64-NEXT: slli a1, a1, 12
+; LMULMAX1-RV64-NEXT: addi a1, a1, -241
+; LMULMAX1-RV64-NEXT: vand.vx v26, v25, a1
+; LMULMAX1-RV64-NEXT: vsll.vi v26, v26, 4
+; LMULMAX1-RV64-NEXT: lui a1, 1044721
+; LMULMAX1-RV64-NEXT: addiw a1, a1, -241
+; LMULMAX1-RV64-NEXT: slli a1, a1, 12
+; LMULMAX1-RV64-NEXT: addi a1, a1, 241
+; LMULMAX1-RV64-NEXT: slli a1, a1, 12
+; LMULMAX1-RV64-NEXT: addi a1, a1, -241
+; LMULMAX1-RV64-NEXT: slli a1, a1, 12
+; LMULMAX1-RV64-NEXT: addi a1, a1, 240
+; LMULMAX1-RV64-NEXT: vand.vx v25, v25, a1
+; LMULMAX1-RV64-NEXT: vsrl.vi v25, v25, 4
+; LMULMAX1-RV64-NEXT: vor.vv v25, v25, v26
+; LMULMAX1-RV64-NEXT: lui a1, 13107
+; LMULMAX1-RV64-NEXT: addiw a1, a1, 819
+; LMULMAX1-RV64-NEXT: slli a1, a1, 12
+; LMULMAX1-RV64-NEXT: addi a1, a1, 819
+; LMULMAX1-RV64-NEXT: slli a1, a1, 12
+; LMULMAX1-RV64-NEXT: addi a1, a1, 819
+; LMULMAX1-RV64-NEXT: slli a1, a1, 12
+; LMULMAX1-RV64-NEXT: addi a1, a1, 819
+; LMULMAX1-RV64-NEXT: vand.vx v26, v25, a1
+; LMULMAX1-RV64-NEXT: vsll.vi v26, v26, 2
+; LMULMAX1-RV64-NEXT: lui a1, 1035469
+; LMULMAX1-RV64-NEXT: addiw a1, a1, -819
+; LMULMAX1-RV64-NEXT: slli a1, a1, 12
+; LMULMAX1-RV64-NEXT: addi a1, a1, -819
+; LMULMAX1-RV64-NEXT: slli a1, a1, 12
+; LMULMAX1-RV64-NEXT: addi a1, a1, -819
+; LMULMAX1-RV64-NEXT: slli a1, a1, 12
+; LMULMAX1-RV64-NEXT: addi a1, a1, -820
+; LMULMAX1-RV64-NEXT: vand.vx v25, v25, a1
+; LMULMAX1-RV64-NEXT: vsrl.vi v25, v25, 2
+; LMULMAX1-RV64-NEXT: vor.vv v25, v25, v26
+; LMULMAX1-RV64-NEXT: lui a1, 21845
+; LMULMAX1-RV64-NEXT: addiw a1, a1, 1365
+; LMULMAX1-RV64-NEXT: slli a1, a1, 12
+; LMULMAX1-RV64-NEXT: addi a1, a1, 1365
+; LMULMAX1-RV64-NEXT: slli a1, a1, 12
+; LMULMAX1-RV64-NEXT: addi a1, a1, 1365
+; LMULMAX1-RV64-NEXT: slli a1, a1, 12
+; LMULMAX1-RV64-NEXT: addi a1, a1, 1365
+; LMULMAX1-RV64-NEXT: vand.vx v26, v25, a1
+; LMULMAX1-RV64-NEXT: vsll.vi v26, v26, 1
+; LMULMAX1-RV64-NEXT: lui a1, 1026731
+; LMULMAX1-RV64-NEXT: addiw a1, a1, -1365
+; LMULMAX1-RV64-NEXT: slli a1, a1, 12
+; LMULMAX1-RV64-NEXT: addi a1, a1, -1365
+; LMULMAX1-RV64-NEXT: slli a1, a1, 12
+; LMULMAX1-RV64-NEXT: addi a1, a1, -1365
+; LMULMAX1-RV64-NEXT: slli a1, a1, 12
+; LMULMAX1-RV64-NEXT: addi a1, a1, -1366
+; LMULMAX1-RV64-NEXT: vand.vx v25, v25, a1
+; LMULMAX1-RV64-NEXT: vsrl.vi v25, v25, 1
+; LMULMAX1-RV64-NEXT: vor.vv v25, v25, v26
+; LMULMAX1-RV64-NEXT: vse64.v v25, (a0)
+; LMULMAX1-RV64-NEXT: ret
%a = load <2 x i64>, <2 x i64>* %x
%b = load <2 x i64>, <2 x i64>* %y
%c = call <2 x i64> @llvm.bitreverse.v2i64(<2 x i64> %a)
@@ -274,431 +853,193 @@ declare <2 x i64> @llvm.bitreverse.v2i64(<2 x i64>)
define void @bitreverse_v16i16(<16 x i16>* %x, <16 x i16>* %y) {
; LMULMAX2-RV32-LABEL: bitreverse_v16i16:
; LMULMAX2-RV32: # %bb.0:
-; LMULMAX2-RV32-NEXT: addi sp, sp, -64
-; LMULMAX2-RV32-NEXT: .cfi_def_cfa_offset 64
-; LMULMAX2-RV32-NEXT: sw ra, 60(sp) # 4-byte Folded Spill
-; LMULMAX2-RV32-NEXT: sw s0, 56(sp) # 4-byte Folded Spill
-; LMULMAX2-RV32-NEXT: .cfi_offset ra, -4
-; LMULMAX2-RV32-NEXT: .cfi_offset s0, -8
-; LMULMAX2-RV32-NEXT: addi s0, sp, 64
-; LMULMAX2-RV32-NEXT: .cfi_def_cfa s0, 0
-; LMULMAX2-RV32-NEXT: andi sp, sp, -32
; LMULMAX2-RV32-NEXT: vsetivli a1, 16, e16,m2,ta,mu
; LMULMAX2-RV32-NEXT: vle16.v v26, (a0)
-; LMULMAX2-RV32-NEXT: vsetvli zero, zero, e8,m2,ta,mu
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v26
-; LMULMAX2-RV32-NEXT: sb a1, 1(sp)
-; LMULMAX2-RV32-NEXT: vsetivli a1, 1, e8,m2,ta,mu
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 30
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 31(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 31
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 30(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 28
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 29(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 29
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 28(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 26
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 27(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 27
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 26(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 24
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 25(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 25
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 24(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 22
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 23(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 23
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 22(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 20
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 21(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 21
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 20(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 18
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 19(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 19
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 18(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 16
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 17(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 17
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 16(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 14
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 15(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 15
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 14(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 12
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 13(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 13
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 12(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 10
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 11(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 11
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 10(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 8
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 9(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 9
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 8(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 6
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 7(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 7
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 6(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 4
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 5(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 5
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 4(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 2
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 3(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 3
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 2(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v26, v26, 1
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v26
-; LMULMAX2-RV32-NEXT: sb a1, 0(sp)
-; LMULMAX2-RV32-NEXT: addi a1, zero, 32
-; LMULMAX2-RV32-NEXT: vsetvli a1, a1, e8,m2,ta,mu
-; LMULMAX2-RV32-NEXT: vle8.v v26, (sp)
-; LMULMAX2-RV32-NEXT: vand.vi v28, v26, 15
+; LMULMAX2-RV32-NEXT: vsrl.vi v28, v26, 8
+; LMULMAX2-RV32-NEXT: vsll.vi v26, v26, 8
+; LMULMAX2-RV32-NEXT: vor.vv v26, v26, v28
+; LMULMAX2-RV32-NEXT: lui a1, 1
+; LMULMAX2-RV32-NEXT: addi a1, a1, -241
+; LMULMAX2-RV32-NEXT: vand.vx v28, v26, a1
; LMULMAX2-RV32-NEXT: vsll.vi v28, v28, 4
-; LMULMAX2-RV32-NEXT: vand.vi v26, v26, -16
+; LMULMAX2-RV32-NEXT: lui a1, 15
+; LMULMAX2-RV32-NEXT: addi a1, a1, 240
+; LMULMAX2-RV32-NEXT: vand.vx v26, v26, a1
; LMULMAX2-RV32-NEXT: vsrl.vi v26, v26, 4
; LMULMAX2-RV32-NEXT: vor.vv v26, v26, v28
-; LMULMAX2-RV32-NEXT: addi a1, zero, 51
+; LMULMAX2-RV32-NEXT: lui a1, 3
+; LMULMAX2-RV32-NEXT: addi a1, a1, 819
; LMULMAX2-RV32-NEXT: vand.vx v28, v26, a1
; LMULMAX2-RV32-NEXT: vsll.vi v28, v28, 2
-; LMULMAX2-RV32-NEXT: addi a1, zero, 204
+; LMULMAX2-RV32-NEXT: lui a1, 13
+; LMULMAX2-RV32-NEXT: addi a1, a1, -820
; LMULMAX2-RV32-NEXT: vand.vx v26, v26, a1
; LMULMAX2-RV32-NEXT: vsrl.vi v26, v26, 2
; LMULMAX2-RV32-NEXT: vor.vv v26, v26, v28
-; LMULMAX2-RV32-NEXT: addi a1, zero, 85
+; LMULMAX2-RV32-NEXT: lui a1, 5
+; LMULMAX2-RV32-NEXT: addi a1, a1, 1365
; LMULMAX2-RV32-NEXT: vand.vx v28, v26, a1
; LMULMAX2-RV32-NEXT: vsll.vi v28, v28, 1
-; LMULMAX2-RV32-NEXT: addi a1, zero, 170
+; LMULMAX2-RV32-NEXT: lui a1, 11
+; LMULMAX2-RV32-NEXT: addi a1, a1, -1366
; LMULMAX2-RV32-NEXT: vand.vx v26, v26, a1
; LMULMAX2-RV32-NEXT: vsrl.vi v26, v26, 1
; LMULMAX2-RV32-NEXT: vor.vv v26, v26, v28
-; LMULMAX2-RV32-NEXT: vsetivli a1, 16, e16,m2,ta,mu
; LMULMAX2-RV32-NEXT: vse16.v v26, (a0)
-; LMULMAX2-RV32-NEXT: addi sp, s0, -64
-; LMULMAX2-RV32-NEXT: lw s0, 56(sp) # 4-byte Folded Reload
-; LMULMAX2-RV32-NEXT: lw ra, 60(sp) # 4-byte Folded Reload
-; LMULMAX2-RV32-NEXT: addi sp, sp, 64
; LMULMAX2-RV32-NEXT: ret
;
; LMULMAX2-RV64-LABEL: bitreverse_v16i16:
; LMULMAX2-RV64: # %bb.0:
-; LMULMAX2-RV64-NEXT: addi sp, sp, -64
-; LMULMAX2-RV64-NEXT: .cfi_def_cfa_offset 64
-; LMULMAX2-RV64-NEXT: sd ra, 56(sp) # 8-byte Folded Spill
-; LMULMAX2-RV64-NEXT: sd s0, 48(sp) # 8-byte Folded Spill
-; LMULMAX2-RV64-NEXT: .cfi_offset ra, -8
-; LMULMAX2-RV64-NEXT: .cfi_offset s0, -16
-; LMULMAX2-RV64-NEXT: addi s0, sp, 64
-; LMULMAX2-RV64-NEXT: .cfi_def_cfa s0, 0
-; LMULMAX2-RV64-NEXT: andi sp, sp, -32
; LMULMAX2-RV64-NEXT: vsetivli a1, 16, e16,m2,ta,mu
; LMULMAX2-RV64-NEXT: vle16.v v26, (a0)
-; LMULMAX2-RV64-NEXT: vsetvli zero, zero, e8,m2,ta,mu
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v26
-; LMULMAX2-RV64-NEXT: sb a1, 1(sp)
-; LMULMAX2-RV64-NEXT: vsetivli a1, 1, e8,m2,ta,mu
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 30
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 31(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 31
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 30(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 28
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 29(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 29
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 28(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 26
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 27(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 27
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 26(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 24
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 25(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 25
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 24(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 22
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 23(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 23
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 22(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 20
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 21(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 21
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 20(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 18
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 19(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 19
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 18(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 16
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 17(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 17
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 16(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 14
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 15(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 15
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 14(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 12
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 13(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 13
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 12(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 10
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 11(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 11
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 10(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 8
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 9(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 9
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 8(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 6
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 7(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 7
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 6(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 4
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 5(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 5
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 4(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 2
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 3(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 3
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 2(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v26, v26, 1
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v26
-; LMULMAX2-RV64-NEXT: sb a1, 0(sp)
-; LMULMAX2-RV64-NEXT: addi a1, zero, 32
-; LMULMAX2-RV64-NEXT: vsetvli a1, a1, e8,m2,ta,mu
-; LMULMAX2-RV64-NEXT: vle8.v v26, (sp)
-; LMULMAX2-RV64-NEXT: vand.vi v28, v26, 15
+; LMULMAX2-RV64-NEXT: vsrl.vi v28, v26, 8
+; LMULMAX2-RV64-NEXT: vsll.vi v26, v26, 8
+; LMULMAX2-RV64-NEXT: vor.vv v26, v26, v28
+; LMULMAX2-RV64-NEXT: lui a1, 1
+; LMULMAX2-RV64-NEXT: addiw a1, a1, -241
+; LMULMAX2-RV64-NEXT: vand.vx v28, v26, a1
; LMULMAX2-RV64-NEXT: vsll.vi v28, v28, 4
-; LMULMAX2-RV64-NEXT: vand.vi v26, v26, -16
+; LMULMAX2-RV64-NEXT: lui a1, 15
+; LMULMAX2-RV64-NEXT: addiw a1, a1, 240
+; LMULMAX2-RV64-NEXT: vand.vx v26, v26, a1
; LMULMAX2-RV64-NEXT: vsrl.vi v26, v26, 4
; LMULMAX2-RV64-NEXT: vor.vv v26, v26, v28
-; LMULMAX2-RV64-NEXT: addi a1, zero, 51
+; LMULMAX2-RV64-NEXT: lui a1, 3
+; LMULMAX2-RV64-NEXT: addiw a1, a1, 819
; LMULMAX2-RV64-NEXT: vand.vx v28, v26, a1
; LMULMAX2-RV64-NEXT: vsll.vi v28, v28, 2
-; LMULMAX2-RV64-NEXT: addi a1, zero, 204
+; LMULMAX2-RV64-NEXT: lui a1, 13
+; LMULMAX2-RV64-NEXT: addiw a1, a1, -820
; LMULMAX2-RV64-NEXT: vand.vx v26, v26, a1
; LMULMAX2-RV64-NEXT: vsrl.vi v26, v26, 2
; LMULMAX2-RV64-NEXT: vor.vv v26, v26, v28
-; LMULMAX2-RV64-NEXT: addi a1, zero, 85
+; LMULMAX2-RV64-NEXT: lui a1, 5
+; LMULMAX2-RV64-NEXT: addiw a1, a1, 1365
; LMULMAX2-RV64-NEXT: vand.vx v28, v26, a1
; LMULMAX2-RV64-NEXT: vsll.vi v28, v28, 1
-; LMULMAX2-RV64-NEXT: addi a1, zero, 170
+; LMULMAX2-RV64-NEXT: lui a1, 11
+; LMULMAX2-RV64-NEXT: addiw a1, a1, -1366
; LMULMAX2-RV64-NEXT: vand.vx v26, v26, a1
; LMULMAX2-RV64-NEXT: vsrl.vi v26, v26, 1
; LMULMAX2-RV64-NEXT: vor.vv v26, v26, v28
-; LMULMAX2-RV64-NEXT: vsetivli a1, 16, e16,m2,ta,mu
; LMULMAX2-RV64-NEXT: vse16.v v26, (a0)
-; LMULMAX2-RV64-NEXT: addi sp, s0, -64
-; LMULMAX2-RV64-NEXT: ld s0, 48(sp) # 8-byte Folded Reload
-; LMULMAX2-RV64-NEXT: ld ra, 56(sp) # 8-byte Folded Reload
-; LMULMAX2-RV64-NEXT: addi sp, sp, 64
; LMULMAX2-RV64-NEXT: ret
;
-; LMULMAX1-LABEL: bitreverse_v16i16:
-; LMULMAX1: # %bb.0:
-; LMULMAX1-NEXT: addi sp, sp, -32
-; LMULMAX1-NEXT: .cfi_def_cfa_offset 32
-; LMULMAX1-NEXT: vsetivli a1, 8, e16,m1,ta,mu
-; LMULMAX1-NEXT: addi a1, a0, 16
-; LMULMAX1-NEXT: vle16.v v26, (a1)
-; LMULMAX1-NEXT: vle16.v v25, (a0)
-; LMULMAX1-NEXT: vsetvli zero, zero, e8,m1,ta,mu
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 1(sp)
-; LMULMAX1-NEXT: vsetivli a2, 1, e8,m1,ta,mu
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 14
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 15(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 15
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 14(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 12
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 13(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 13
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 12(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 10
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 11(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 11
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 10(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 8
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 9(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 9
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 8(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 6
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 7(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 7
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 6(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 4
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 5(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 5
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 4(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 2
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 3(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 3
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 2(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v26, 1
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 0(sp)
-; LMULMAX1-NEXT: vmv.x.s a2, v25
-; LMULMAX1-NEXT: sb a2, 17(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 14
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 31(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 15
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 30(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 12
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 29(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 13
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 28(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 10
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 27(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 11
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 26(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 8
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 25(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 9
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 24(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 6
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 23(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 7
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 22(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 4
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 21(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 5
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 20(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 2
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 19(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 3
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 18(sp)
-; LMULMAX1-NEXT: vslidedown.vi v25, v25, 1
-; LMULMAX1-NEXT: vmv.x.s a2, v25
-; LMULMAX1-NEXT: sb a2, 16(sp)
-; LMULMAX1-NEXT: vsetivli a2, 16, e8,m1,ta,mu
-; LMULMAX1-NEXT: vle8.v v25, (sp)
-; LMULMAX1-NEXT: vand.vi v26, v25, 15
-; LMULMAX1-NEXT: vsll.vi v26, v26, 4
-; LMULMAX1-NEXT: vand.vi v25, v25, -16
-; LMULMAX1-NEXT: vsrl.vi v25, v25, 4
-; LMULMAX1-NEXT: vor.vv v25, v25, v26
-; LMULMAX1-NEXT: addi a2, zero, 51
-; LMULMAX1-NEXT: vand.vx v26, v25, a2
-; LMULMAX1-NEXT: vsll.vi v26, v26, 2
-; LMULMAX1-NEXT: addi a3, zero, 204
-; LMULMAX1-NEXT: vand.vx v25, v25, a3
-; LMULMAX1-NEXT: vsrl.vi v25, v25, 2
-; LMULMAX1-NEXT: vor.vv v25, v25, v26
-; LMULMAX1-NEXT: addi a4, zero, 85
-; LMULMAX1-NEXT: vand.vx v26, v25, a4
-; LMULMAX1-NEXT: vsll.vi v26, v26, 1
-; LMULMAX1-NEXT: addi a5, zero, 170
-; LMULMAX1-NEXT: addi a6, sp, 16
-; LMULMAX1-NEXT: vle8.v v27, (a6)
-; LMULMAX1-NEXT: vand.vx v25, v25, a5
-; LMULMAX1-NEXT: vsrl.vi v25, v25, 1
-; LMULMAX1-NEXT: vor.vv v25, v25, v26
-; LMULMAX1-NEXT: vand.vi v26, v27, 15
-; LMULMAX1-NEXT: vsll.vi v26, v26, 4
-; LMULMAX1-NEXT: vand.vi v27, v27, -16
-; LMULMAX1-NEXT: vsrl.vi v27, v27, 4
-; LMULMAX1-NEXT: vor.vv v26, v27, v26
-; LMULMAX1-NEXT: vand.vx v27, v26, a2
-; LMULMAX1-NEXT: vsll.vi v27, v27, 2
-; LMULMAX1-NEXT: vand.vx v26, v26, a3
-; LMULMAX1-NEXT: vsrl.vi v26, v26, 2
-; LMULMAX1-NEXT: vor.vv v26, v26, v27
-; LMULMAX1-NEXT: vand.vx v27, v26, a4
-; LMULMAX1-NEXT: vsll.vi v27, v27, 1
-; LMULMAX1-NEXT: vand.vx v26, v26, a5
-; LMULMAX1-NEXT: vsrl.vi v26, v26, 1
-; LMULMAX1-NEXT: vor.vv v26, v26, v27
-; LMULMAX1-NEXT: vsetivli a2, 8, e16,m1,ta,mu
-; LMULMAX1-NEXT: vse16.v v26, (a0)
-; LMULMAX1-NEXT: vse16.v v25, (a1)
-; LMULMAX1-NEXT: addi sp, sp, 32
-; LMULMAX1-NEXT: ret
+; LMULMAX1-RV32-LABEL: bitreverse_v16i16:
+; LMULMAX1-RV32: # %bb.0:
+; LMULMAX1-RV32-NEXT: vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX1-RV32-NEXT: addi a6, a0, 16
+; LMULMAX1-RV32-NEXT: vle16.v v25, (a6)
+; LMULMAX1-RV32-NEXT: vle16.v v26, (a0)
+; LMULMAX1-RV32-NEXT: vsrl.vi v27, v25, 8
+; LMULMAX1-RV32-NEXT: vsll.vi v25, v25, 8
+; LMULMAX1-RV32-NEXT: vor.vv v25, v25, v27
+; LMULMAX1-RV32-NEXT: lui a2, 1
+; LMULMAX1-RV32-NEXT: addi a7, a2, -241
+; LMULMAX1-RV32-NEXT: vand.vx v27, v25, a7
+; LMULMAX1-RV32-NEXT: vsll.vi v27, v27, 4
+; LMULMAX1-RV32-NEXT: lui a3, 15
+; LMULMAX1-RV32-NEXT: addi a3, a3, 240
+; LMULMAX1-RV32-NEXT: vand.vx v25, v25, a3
+; LMULMAX1-RV32-NEXT: vsrl.vi v25, v25, 4
+; LMULMAX1-RV32-NEXT: vor.vv v25, v25, v27
+; LMULMAX1-RV32-NEXT: lui a4, 3
+; LMULMAX1-RV32-NEXT: addi a4, a4, 819
+; LMULMAX1-RV32-NEXT: vand.vx v27, v25, a4
+; LMULMAX1-RV32-NEXT: vsll.vi v27, v27, 2
+; LMULMAX1-RV32-NEXT: lui a5, 13
+; LMULMAX1-RV32-NEXT: addi a5, a5, -820
+; LMULMAX1-RV32-NEXT: vand.vx v25, v25, a5
+; LMULMAX1-RV32-NEXT: vsrl.vi v25, v25, 2
+; LMULMAX1-RV32-NEXT: vor.vv v25, v25, v27
+; LMULMAX1-RV32-NEXT: lui a1, 5
+; LMULMAX1-RV32-NEXT: addi a1, a1, 1365
+; LMULMAX1-RV32-NEXT: vand.vx v27, v25, a1
+; LMULMAX1-RV32-NEXT: vsll.vi v27, v27, 1
+; LMULMAX1-RV32-NEXT: lui a2, 11
+; LMULMAX1-RV32-NEXT: addi a2, a2, -1366
+; LMULMAX1-RV32-NEXT: vand.vx v25, v25, a2
+; LMULMAX1-RV32-NEXT: vsrl.vi v25, v25, 1
+; LMULMAX1-RV32-NEXT: vor.vv v25, v25, v27
+; LMULMAX1-RV32-NEXT: vsrl.vi v27, v26, 8
+; LMULMAX1-RV32-NEXT: vsll.vi v26, v26, 8
+; LMULMAX1-RV32-NEXT: vor.vv v26, v26, v27
+; LMULMAX1-RV32-NEXT: vand.vx v27, v26, a7
+; LMULMAX1-RV32-NEXT: vsll.vi v27, v27, 4
+; LMULMAX1-RV32-NEXT: vand.vx v26, v26, a3
+; LMULMAX1-RV32-NEXT: vsrl.vi v26, v26, 4
+; LMULMAX1-RV32-NEXT: vor.vv v26, v26, v27
+; LMULMAX1-RV32-NEXT: vand.vx v27, v26, a4
+; LMULMAX1-RV32-NEXT: vsll.vi v27, v27, 2
+; LMULMAX1-RV32-NEXT: vand.vx v26, v26, a5
+; LMULMAX1-RV32-NEXT: vsrl.vi v26, v26, 2
+; LMULMAX1-RV32-NEXT: vor.vv v26, v26, v27
+; LMULMAX1-RV32-NEXT: vand.vx v27, v26, a1
+; LMULMAX1-RV32-NEXT: vsll.vi v27, v27, 1
+; LMULMAX1-RV32-NEXT: vand.vx v26, v26, a2
+; LMULMAX1-RV32-NEXT: vsrl.vi v26, v26, 1
+; LMULMAX1-RV32-NEXT: vor.vv v26, v26, v27
+; LMULMAX1-RV32-NEXT: vse16.v v26, (a0)
+; LMULMAX1-RV32-NEXT: vse16.v v25, (a6)
+; LMULMAX1-RV32-NEXT: ret
+;
+; LMULMAX1-RV64-LABEL: bitreverse_v16i16:
+; LMULMAX1-RV64: # %bb.0:
+; LMULMAX1-RV64-NEXT: vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX1-RV64-NEXT: addi a6, a0, 16
+; LMULMAX1-RV64-NEXT: vle16.v v25, (a6)
+; LMULMAX1-RV64-NEXT: vle16.v v26, (a0)
+; LMULMAX1-RV64-NEXT: vsrl.vi v27, v25, 8
+; LMULMAX1-RV64-NEXT: vsll.vi v25, v25, 8
+; LMULMAX1-RV64-NEXT: vor.vv v25, v25, v27
+; LMULMAX1-RV64-NEXT: lui a2, 1
+; LMULMAX1-RV64-NEXT: addiw a7, a2, -241
+; LMULMAX1-RV64-NEXT: vand.vx v27, v25, a7
+; LMULMAX1-RV64-NEXT: vsll.vi v27, v27, 4
+; LMULMAX1-RV64-NEXT: lui a3, 15
+; LMULMAX1-RV64-NEXT: addiw a3, a3, 240
+; LMULMAX1-RV64-NEXT: vand.vx v25, v25, a3
+; LMULMAX1-RV64-NEXT: vsrl.vi v25, v25, 4
+; LMULMAX1-RV64-NEXT: vor.vv v25, v25, v27
+; LMULMAX1-RV64-NEXT: lui a4, 3
+; LMULMAX1-RV64-NEXT: addiw a4, a4, 819
+; LMULMAX1-RV64-NEXT: vand.vx v27, v25, a4
+; LMULMAX1-RV64-NEXT: vsll.vi v27, v27, 2
+; LMULMAX1-RV64-NEXT: lui a5, 13
+; LMULMAX1-RV64-NEXT: addiw a5, a5, -820
+; LMULMAX1-RV64-NEXT: vand.vx v25, v25, a5
+; LMULMAX1-RV64-NEXT: vsrl.vi v25, v25, 2
+; LMULMAX1-RV64-NEXT: vor.vv v25, v25, v27
+; LMULMAX1-RV64-NEXT: lui a1, 5
+; LMULMAX1-RV64-NEXT: addiw a1, a1, 1365
+; LMULMAX1-RV64-NEXT: vand.vx v27, v25, a1
+; LMULMAX1-RV64-NEXT: vsll.vi v27, v27, 1
+; LMULMAX1-RV64-NEXT: lui a2, 11
+; LMULMAX1-RV64-NEXT: addiw a2, a2, -1366
+; LMULMAX1-RV64-NEXT: vand.vx v25, v25, a2
+; LMULMAX1-RV64-NEXT: vsrl.vi v25, v25, 1
+; LMULMAX1-RV64-NEXT: vor.vv v25, v25, v27
+; LMULMAX1-RV64-NEXT: vsrl.vi v27, v26, 8
+; LMULMAX1-RV64-NEXT: vsll.vi v26, v26, 8
+; LMULMAX1-RV64-NEXT: vor.vv v26, v26, v27
+; LMULMAX1-RV64-NEXT: vand.vx v27, v26, a7
+; LMULMAX1-RV64-NEXT: vsll.vi v27, v27, 4
+; LMULMAX1-RV64-NEXT: vand.vx v26, v26, a3
+; LMULMAX1-RV64-NEXT: vsrl.vi v26, v26, 4
+; LMULMAX1-RV64-NEXT: vor.vv v26, v26, v27
+; LMULMAX1-RV64-NEXT: vand.vx v27, v26, a4
+; LMULMAX1-RV64-NEXT: vsll.vi v27, v27, 2
+; LMULMAX1-RV64-NEXT: vand.vx v26, v26, a5
+; LMULMAX1-RV64-NEXT: vsrl.vi v26, v26, 2
+; LMULMAX1-RV64-NEXT: vor.vv v26, v26, v27
+; LMULMAX1-RV64-NEXT: vand.vx v27, v26, a1
+; LMULMAX1-RV64-NEXT: vsll.vi v27, v27, 1
+; LMULMAX1-RV64-NEXT: vand.vx v26, v26, a2
+; LMULMAX1-RV64-NEXT: vsrl.vi v26, v26, 1
+; LMULMAX1-RV64-NEXT: vor.vv v26, v26, v27
+; LMULMAX1-RV64-NEXT: vse16.v v26, (a0)
+; LMULMAX1-RV64-NEXT: vse16.v v25, (a6)
+; LMULMAX1-RV64-NEXT: ret
%a = load <16 x i16>, <16 x i16>* %x
%b = load <16 x i16>, <16 x i16>* %y
%c = call <16 x i16> @llvm.bitreverse.v16i16(<16 x i16> %a)
@@ -710,431 +1051,253 @@ declare <16 x i16> @llvm.bitreverse.v16i16(<16 x i16>)
define void @bitreverse_v8i32(<8 x i32>* %x, <8 x i32>* %y) {
; LMULMAX2-RV32-LABEL: bitreverse_v8i32:
; LMULMAX2-RV32: # %bb.0:
-; LMULMAX2-RV32-NEXT: addi sp, sp, -64
-; LMULMAX2-RV32-NEXT: .cfi_def_cfa_offset 64
-; LMULMAX2-RV32-NEXT: sw ra, 60(sp) # 4-byte Folded Spill
-; LMULMAX2-RV32-NEXT: sw s0, 56(sp) # 4-byte Folded Spill
-; LMULMAX2-RV32-NEXT: .cfi_offset ra, -4
-; LMULMAX2-RV32-NEXT: .cfi_offset s0, -8
-; LMULMAX2-RV32-NEXT: addi s0, sp, 64
-; LMULMAX2-RV32-NEXT: .cfi_def_cfa s0, 0
-; LMULMAX2-RV32-NEXT: andi sp, sp, -32
; LMULMAX2-RV32-NEXT: vsetivli a1, 8, e32,m2,ta,mu
; LMULMAX2-RV32-NEXT: vle32.v v26, (a0)
-; LMULMAX2-RV32-NEXT: vsetvli zero, zero, e8,m2,ta,mu
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v26
-; LMULMAX2-RV32-NEXT: sb a1, 3(sp)
-; LMULMAX2-RV32-NEXT: vsetivli a1, 1, e8,m2,ta,mu
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 28
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 31(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 29
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 30(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 30
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 29(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 31
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 28(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 24
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 27(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 25
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 26(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 26
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 25(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 27
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 24(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 20
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 23(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 21
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 22(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 22
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 21(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 23
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 20(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 16
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 19(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 17
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 18(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 18
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 17(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 19
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 16(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 12
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 15(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 13
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 14(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 14
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 13(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 15
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 12(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 8
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 11(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 9
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 10(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 10
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 9(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 11
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 8(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 4
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 7(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 5
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 6(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 6
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 5(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 7
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 4(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 1
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 2(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 2
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 1(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v26, v26, 3
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v26
-; LMULMAX2-RV32-NEXT: sb a1, 0(sp)
-; LMULMAX2-RV32-NEXT: addi a1, zero, 32
-; LMULMAX2-RV32-NEXT: vsetvli a1, a1, e8,m2,ta,mu
-; LMULMAX2-RV32-NEXT: vle8.v v26, (sp)
-; LMULMAX2-RV32-NEXT: vand.vi v28, v26, 15
+; LMULMAX2-RV32-NEXT: vsrl.vi v28, v26, 8
+; LMULMAX2-RV32-NEXT: lui a1, 16
+; LMULMAX2-RV32-NEXT: addi a1, a1, -256
+; LMULMAX2-RV32-NEXT: vand.vx v28, v28, a1
+; LMULMAX2-RV32-NEXT: vsrl.vi v30, v26, 24
+; LMULMAX2-RV32-NEXT: vor.vv v28, v28, v30
+; LMULMAX2-RV32-NEXT: vsll.vi v30, v26, 8
+; LMULMAX2-RV32-NEXT: lui a1, 4080
+; LMULMAX2-RV32-NEXT: vand.vx v30, v30, a1
+; LMULMAX2-RV32-NEXT: vsll.vi v26, v26, 24
+; LMULMAX2-RV32-NEXT: vor.vv v26, v26, v30
+; LMULMAX2-RV32-NEXT: vor.vv v26, v26, v28
+; LMULMAX2-RV32-NEXT: lui a1, 61681
+; LMULMAX2-RV32-NEXT: addi a1, a1, -241
+; LMULMAX2-RV32-NEXT: vand.vx v28, v26, a1
; LMULMAX2-RV32-NEXT: vsll.vi v28, v28, 4
-; LMULMAX2-RV32-NEXT: vand.vi v26, v26, -16
+; LMULMAX2-RV32-NEXT: lui a1, 986895
+; LMULMAX2-RV32-NEXT: addi a1, a1, 240
+; LMULMAX2-RV32-NEXT: vand.vx v26, v26, a1
; LMULMAX2-RV32-NEXT: vsrl.vi v26, v26, 4
; LMULMAX2-RV32-NEXT: vor.vv v26, v26, v28
-; LMULMAX2-RV32-NEXT: addi a1, zero, 51
+; LMULMAX2-RV32-NEXT: lui a1, 209715
+; LMULMAX2-RV32-NEXT: addi a1, a1, 819
; LMULMAX2-RV32-NEXT: vand.vx v28, v26, a1
; LMULMAX2-RV32-NEXT: vsll.vi v28, v28, 2
-; LMULMAX2-RV32-NEXT: addi a1, zero, 204
+; LMULMAX2-RV32-NEXT: lui a1, 838861
+; LMULMAX2-RV32-NEXT: addi a1, a1, -820
; LMULMAX2-RV32-NEXT: vand.vx v26, v26, a1
; LMULMAX2-RV32-NEXT: vsrl.vi v26, v26, 2
; LMULMAX2-RV32-NEXT: vor.vv v26, v26, v28
-; LMULMAX2-RV32-NEXT: addi a1, zero, 85
+; LMULMAX2-RV32-NEXT: lui a1, 349525
+; LMULMAX2-RV32-NEXT: addi a1, a1, 1365
; LMULMAX2-RV32-NEXT: vand.vx v28, v26, a1
; LMULMAX2-RV32-NEXT: vsll.vi v28, v28, 1
-; LMULMAX2-RV32-NEXT: addi a1, zero, 170
+; LMULMAX2-RV32-NEXT: lui a1, 699051
+; LMULMAX2-RV32-NEXT: addi a1, a1, -1366
; LMULMAX2-RV32-NEXT: vand.vx v26, v26, a1
; LMULMAX2-RV32-NEXT: vsrl.vi v26, v26, 1
; LMULMAX2-RV32-NEXT: vor.vv v26, v26, v28
-; LMULMAX2-RV32-NEXT: vsetivli a1, 8, e32,m2,ta,mu
; LMULMAX2-RV32-NEXT: vse32.v v26, (a0)
-; LMULMAX2-RV32-NEXT: addi sp, s0, -64
-; LMULMAX2-RV32-NEXT: lw s0, 56(sp) # 4-byte Folded Reload
-; LMULMAX2-RV32-NEXT: lw ra, 60(sp) # 4-byte Folded Reload
-; LMULMAX2-RV32-NEXT: addi sp, sp, 64
; LMULMAX2-RV32-NEXT: ret
;
; LMULMAX2-RV64-LABEL: bitreverse_v8i32:
; LMULMAX2-RV64: # %bb.0:
-; LMULMAX2-RV64-NEXT: addi sp, sp, -64
-; LMULMAX2-RV64-NEXT: .cfi_def_cfa_offset 64
-; LMULMAX2-RV64-NEXT: sd ra, 56(sp) # 8-byte Folded Spill
-; LMULMAX2-RV64-NEXT: sd s0, 48(sp) # 8-byte Folded Spill
-; LMULMAX2-RV64-NEXT: .cfi_offset ra, -8
-; LMULMAX2-RV64-NEXT: .cfi_offset s0, -16
-; LMULMAX2-RV64-NEXT: addi s0, sp, 64
-; LMULMAX2-RV64-NEXT: .cfi_def_cfa s0, 0
-; LMULMAX2-RV64-NEXT: andi sp, sp, -32
; LMULMAX2-RV64-NEXT: vsetivli a1, 8, e32,m2,ta,mu
; LMULMAX2-RV64-NEXT: vle32.v v26, (a0)
-; LMULMAX2-RV64-NEXT: vsetvli zero, zero, e8,m2,ta,mu
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v26
-; LMULMAX2-RV64-NEXT: sb a1, 3(sp)
-; LMULMAX2-RV64-NEXT: vsetivli a1, 1, e8,m2,ta,mu
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 28
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 31(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 29
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 30(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 30
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 29(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 31
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 28(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 24
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 27(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 25
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 26(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 26
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 25(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 27
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 24(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 20
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 23(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 21
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 22(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 22
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 21(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 23
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 20(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 16
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 19(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 17
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 18(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 18
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 17(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 19
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 16(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 12
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 15(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 13
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 14(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 14
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 13(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 15
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 12(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 8
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 11(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 9
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 10(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 10
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 9(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 11
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 8(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 4
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 7(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 5
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 6(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 6
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 5(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 7
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 4(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 1
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 2(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 2
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 1(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v26, v26, 3
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v26
-; LMULMAX2-RV64-NEXT: sb a1, 0(sp)
-; LMULMAX2-RV64-NEXT: addi a1, zero, 32
-; LMULMAX2-RV64-NEXT: vsetvli a1, a1, e8,m2,ta,mu
-; LMULMAX2-RV64-NEXT: vle8.v v26, (sp)
-; LMULMAX2-RV64-NEXT: vand.vi v28, v26, 15
+; LMULMAX2-RV64-NEXT: vsrl.vi v28, v26, 8
+; LMULMAX2-RV64-NEXT: lui a1, 16
+; LMULMAX2-RV64-NEXT: addiw a1, a1, -256
+; LMULMAX2-RV64-NEXT: vand.vx v28, v28, a1
+; LMULMAX2-RV64-NEXT: vsrl.vi v30, v26, 24
+; LMULMAX2-RV64-NEXT: vor.vv v28, v28, v30
+; LMULMAX2-RV64-NEXT: vsll.vi v30, v26, 8
+; LMULMAX2-RV64-NEXT: lui a1, 4080
+; LMULMAX2-RV64-NEXT: vand.vx v30, v30, a1
+; LMULMAX2-RV64-NEXT: vsll.vi v26, v26, 24
+; LMULMAX2-RV64-NEXT: vor.vv v26, v26, v30
+; LMULMAX2-RV64-NEXT: vor.vv v26, v26, v28
+; LMULMAX2-RV64-NEXT: lui a1, 61681
+; LMULMAX2-RV64-NEXT: addiw a1, a1, -241
+; LMULMAX2-RV64-NEXT: vand.vx v28, v26, a1
; LMULMAX2-RV64-NEXT: vsll.vi v28, v28, 4
-; LMULMAX2-RV64-NEXT: vand.vi v26, v26, -16
+; LMULMAX2-RV64-NEXT: lui a1, 241
+; LMULMAX2-RV64-NEXT: addiw a1, a1, -241
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, 240
+; LMULMAX2-RV64-NEXT: vand.vx v26, v26, a1
; LMULMAX2-RV64-NEXT: vsrl.vi v26, v26, 4
; LMULMAX2-RV64-NEXT: vor.vv v26, v26, v28
-; LMULMAX2-RV64-NEXT: addi a1, zero, 51
+; LMULMAX2-RV64-NEXT: lui a1, 209715
+; LMULMAX2-RV64-NEXT: addiw a1, a1, 819
; LMULMAX2-RV64-NEXT: vand.vx v28, v26, a1
; LMULMAX2-RV64-NEXT: vsll.vi v28, v28, 2
-; LMULMAX2-RV64-NEXT: addi a1, zero, 204
+; LMULMAX2-RV64-NEXT: lui a1, 205
+; LMULMAX2-RV64-NEXT: addiw a1, a1, -819
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, -820
; LMULMAX2-RV64-NEXT: vand.vx v26, v26, a1
; LMULMAX2-RV64-NEXT: vsrl.vi v26, v26, 2
; LMULMAX2-RV64-NEXT: vor.vv v26, v26, v28
-; LMULMAX2-RV64-NEXT: addi a1, zero, 85
+; LMULMAX2-RV64-NEXT: lui a1, 349525
+; LMULMAX2-RV64-NEXT: addiw a1, a1, 1365
; LMULMAX2-RV64-NEXT: vand.vx v28, v26, a1
; LMULMAX2-RV64-NEXT: vsll.vi v28, v28, 1
-; LMULMAX2-RV64-NEXT: addi a1, zero, 170
+; LMULMAX2-RV64-NEXT: lui a1, 171
+; LMULMAX2-RV64-NEXT: addiw a1, a1, -1365
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, -1366
; LMULMAX2-RV64-NEXT: vand.vx v26, v26, a1
; LMULMAX2-RV64-NEXT: vsrl.vi v26, v26, 1
; LMULMAX2-RV64-NEXT: vor.vv v26, v26, v28
-; LMULMAX2-RV64-NEXT: vsetivli a1, 8, e32,m2,ta,mu
; LMULMAX2-RV64-NEXT: vse32.v v26, (a0)
-; LMULMAX2-RV64-NEXT: addi sp, s0, -64
-; LMULMAX2-RV64-NEXT: ld s0, 48(sp) # 8-byte Folded Reload
-; LMULMAX2-RV64-NEXT: ld ra, 56(sp) # 8-byte Folded Reload
-; LMULMAX2-RV64-NEXT: addi sp, sp, 64
; LMULMAX2-RV64-NEXT: ret
;
-; LMULMAX1-LABEL: bitreverse_v8i32:
-; LMULMAX1: # %bb.0:
-; LMULMAX1-NEXT: addi sp, sp, -32
-; LMULMAX1-NEXT: .cfi_def_cfa_offset 32
-; LMULMAX1-NEXT: vsetivli a1, 4, e32,m1,ta,mu
-; LMULMAX1-NEXT: addi a1, a0, 16
-; LMULMAX1-NEXT: vle32.v v26, (a1)
-; LMULMAX1-NEXT: vle32.v v25, (a0)
-; LMULMAX1-NEXT: vsetvli zero, zero, e8,m1,ta,mu
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 3(sp)
-; LMULMAX1-NEXT: vsetivli a2, 1, e8,m1,ta,mu
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 12
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 15(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 13
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 14(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 14
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 13(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 15
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 12(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 8
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 11(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 9
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 10(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 10
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 9(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 11
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 8(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 4
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 7(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 5
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 6(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 6
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 5(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 7
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 4(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 1
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 2(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 2
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 1(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v26, 3
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 0(sp)
-; LMULMAX1-NEXT: vmv.x.s a2, v25
-; LMULMAX1-NEXT: sb a2, 19(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 12
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 31(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 13
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 30(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 14
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 29(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 15
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 28(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 8
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 27(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 9
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 26(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 10
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 25(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 11
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 24(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 4
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 23(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 5
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 22(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 6
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 21(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 7
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 20(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 1
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 18(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 2
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 17(sp)
-; LMULMAX1-NEXT: vslidedown.vi v25, v25, 3
-; LMULMAX1-NEXT: vmv.x.s a2, v25
-; LMULMAX1-NEXT: sb a2, 16(sp)
-; LMULMAX1-NEXT: vsetivli a2, 16, e8,m1,ta,mu
-; LMULMAX1-NEXT: vle8.v v25, (sp)
-; LMULMAX1-NEXT: vand.vi v26, v25, 15
-; LMULMAX1-NEXT: vsll.vi v26, v26, 4
-; LMULMAX1-NEXT: vand.vi v25, v25, -16
-; LMULMAX1-NEXT: vsrl.vi v25, v25, 4
-; LMULMAX1-NEXT: vor.vv v25, v25, v26
-; LMULMAX1-NEXT: addi a2, zero, 51
-; LMULMAX1-NEXT: vand.vx v26, v25, a2
-; LMULMAX1-NEXT: vsll.vi v26, v26, 2
-; LMULMAX1-NEXT: addi a3, zero, 204
-; LMULMAX1-NEXT: vand.vx v25, v25, a3
-; LMULMAX1-NEXT: vsrl.vi v25, v25, 2
-; LMULMAX1-NEXT: vor.vv v25, v25, v26
-; LMULMAX1-NEXT: addi a4, zero, 85
-; LMULMAX1-NEXT: vand.vx v26, v25, a4
-; LMULMAX1-NEXT: vsll.vi v26, v26, 1
-; LMULMAX1-NEXT: addi a5, zero, 170
-; LMULMAX1-NEXT: addi a6, sp, 16
-; LMULMAX1-NEXT: vle8.v v27, (a6)
-; LMULMAX1-NEXT: vand.vx v25, v25, a5
-; LMULMAX1-NEXT: vsrl.vi v25, v25, 1
-; LMULMAX1-NEXT: vor.vv v25, v25, v26
-; LMULMAX1-NEXT: vand.vi v26, v27, 15
-; LMULMAX1-NEXT: vsll.vi v26, v26, 4
-; LMULMAX1-NEXT: vand.vi v27, v27, -16
-; LMULMAX1-NEXT: vsrl.vi v27, v27, 4
-; LMULMAX1-NEXT: vor.vv v26, v27, v26
-; LMULMAX1-NEXT: vand.vx v27, v26, a2
-; LMULMAX1-NEXT: vsll.vi v27, v27, 2
-; LMULMAX1-NEXT: vand.vx v26, v26, a3
-; LMULMAX1-NEXT: vsrl.vi v26, v26, 2
-; LMULMAX1-NEXT: vor.vv v26, v26, v27
-; LMULMAX1-NEXT: vand.vx v27, v26, a4
-; LMULMAX1-NEXT: vsll.vi v27, v27, 1
-; LMULMAX1-NEXT: vand.vx v26, v26, a5
-; LMULMAX1-NEXT: vsrl.vi v26, v26, 1
-; LMULMAX1-NEXT: vor.vv v26, v26, v27
-; LMULMAX1-NEXT: vsetivli a2, 4, e32,m1,ta,mu
-; LMULMAX1-NEXT: vse32.v v26, (a0)
-; LMULMAX1-NEXT: vse32.v v25, (a1)
-; LMULMAX1-NEXT: addi sp, sp, 32
-; LMULMAX1-NEXT: ret
+; LMULMAX1-RV32-LABEL: bitreverse_v8i32:
+; LMULMAX1-RV32: # %bb.0:
+; LMULMAX1-RV32-NEXT: vsetivli a1, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: addi a7, a0, 16
+; LMULMAX1-RV32-NEXT: vle32.v v25, (a7)
+; LMULMAX1-RV32-NEXT: vle32.v v26, (a0)
+; LMULMAX1-RV32-NEXT: vsrl.vi v27, v25, 8
+; LMULMAX1-RV32-NEXT: lui a2, 16
+; LMULMAX1-RV32-NEXT: addi t0, a2, -256
+; LMULMAX1-RV32-NEXT: vand.vx v27, v27, t0
+; LMULMAX1-RV32-NEXT: vsrl.vi v28, v25, 24
+; LMULMAX1-RV32-NEXT: vor.vv v27, v27, v28
+; LMULMAX1-RV32-NEXT: vsll.vi v28, v25, 8
+; LMULMAX1-RV32-NEXT: lui a6, 4080
+; LMULMAX1-RV32-NEXT: vand.vx v28, v28, a6
+; LMULMAX1-RV32-NEXT: vsll.vi v25, v25, 24
+; LMULMAX1-RV32-NEXT: vor.vv v25, v25, v28
+; LMULMAX1-RV32-NEXT: vor.vv v25, v25, v27
+; LMULMAX1-RV32-NEXT: lui a4, 61681
+; LMULMAX1-RV32-NEXT: addi t1, a4, -241
+; LMULMAX1-RV32-NEXT: vand.vx v27, v25, t1
+; LMULMAX1-RV32-NEXT: vsll.vi v27, v27, 4
+; LMULMAX1-RV32-NEXT: lui a5, 986895
+; LMULMAX1-RV32-NEXT: addi a5, a5, 240
+; LMULMAX1-RV32-NEXT: vand.vx v25, v25, a5
+; LMULMAX1-RV32-NEXT: vsrl.vi v25, v25, 4
+; LMULMAX1-RV32-NEXT: vor.vv v25, v25, v27
+; LMULMAX1-RV32-NEXT: lui a3, 209715
+; LMULMAX1-RV32-NEXT: addi a3, a3, 819
+; LMULMAX1-RV32-NEXT: vand.vx v27, v25, a3
+; LMULMAX1-RV32-NEXT: vsll.vi v27, v27, 2
+; LMULMAX1-RV32-NEXT: lui a1, 838861
+; LMULMAX1-RV32-NEXT: addi a1, a1, -820
+; LMULMAX1-RV32-NEXT: vand.vx v25, v25, a1
+; LMULMAX1-RV32-NEXT: vsrl.vi v25, v25, 2
+; LMULMAX1-RV32-NEXT: vor.vv v25, v25, v27
+; LMULMAX1-RV32-NEXT: lui a2, 349525
+; LMULMAX1-RV32-NEXT: addi a2, a2, 1365
+; LMULMAX1-RV32-NEXT: vand.vx v27, v25, a2
+; LMULMAX1-RV32-NEXT: vsll.vi v27, v27, 1
+; LMULMAX1-RV32-NEXT: lui a4, 699051
+; LMULMAX1-RV32-NEXT: addi a4, a4, -1366
+; LMULMAX1-RV32-NEXT: vand.vx v25, v25, a4
+; LMULMAX1-RV32-NEXT: vsrl.vi v25, v25, 1
+; LMULMAX1-RV32-NEXT: vor.vv v25, v25, v27
+; LMULMAX1-RV32-NEXT: vsrl.vi v27, v26, 8
+; LMULMAX1-RV32-NEXT: vand.vx v27, v27, t0
+; LMULMAX1-RV32-NEXT: vsrl.vi v28, v26, 24
+; LMULMAX1-RV32-NEXT: vor.vv v27, v27, v28
+; LMULMAX1-RV32-NEXT: vsll.vi v28, v26, 8
+; LMULMAX1-RV32-NEXT: vand.vx v28, v28, a6
+; LMULMAX1-RV32-NEXT: vsll.vi v26, v26, 24
+; LMULMAX1-RV32-NEXT: vor.vv v26, v26, v28
+; LMULMAX1-RV32-NEXT: vor.vv v26, v26, v27
+; LMULMAX1-RV32-NEXT: vand.vx v27, v26, t1
+; LMULMAX1-RV32-NEXT: vsll.vi v27, v27, 4
+; LMULMAX1-RV32-NEXT: vand.vx v26, v26, a5
+; LMULMAX1-RV32-NEXT: vsrl.vi v26, v26, 4
+; LMULMAX1-RV32-NEXT: vor.vv v26, v26, v27
+; LMULMAX1-RV32-NEXT: vand.vx v27, v26, a3
+; LMULMAX1-RV32-NEXT: vsll.vi v27, v27, 2
+; LMULMAX1-RV32-NEXT: vand.vx v26, v26, a1
+; LMULMAX1-RV32-NEXT: vsrl.vi v26, v26, 2
+; LMULMAX1-RV32-NEXT: vor.vv v26, v26, v27
+; LMULMAX1-RV32-NEXT: vand.vx v27, v26, a2
+; LMULMAX1-RV32-NEXT: vsll.vi v27, v27, 1
+; LMULMAX1-RV32-NEXT: vand.vx v26, v26, a4
+; LMULMAX1-RV32-NEXT: vsrl.vi v26, v26, 1
+; LMULMAX1-RV32-NEXT: vor.vv v26, v26, v27
+; LMULMAX1-RV32-NEXT: vse32.v v26, (a0)
+; LMULMAX1-RV32-NEXT: vse32.v v25, (a7)
+; LMULMAX1-RV32-NEXT: ret
+;
+; LMULMAX1-RV64-LABEL: bitreverse_v8i32:
+; LMULMAX1-RV64: # %bb.0:
+; LMULMAX1-RV64-NEXT: vsetivli a1, 4, e32,m1,ta,mu
+; LMULMAX1-RV64-NEXT: addi a7, a0, 16
+; LMULMAX1-RV64-NEXT: vle32.v v25, (a7)
+; LMULMAX1-RV64-NEXT: vle32.v v26, (a0)
+; LMULMAX1-RV64-NEXT: vsrl.vi v27, v25, 8
+; LMULMAX1-RV64-NEXT: lui a2, 16
+; LMULMAX1-RV64-NEXT: addiw t0, a2, -256
+; LMULMAX1-RV64-NEXT: vand.vx v27, v27, t0
+; LMULMAX1-RV64-NEXT: vsrl.vi v28, v25, 24
+; LMULMAX1-RV64-NEXT: vor.vv v27, v27, v28
+; LMULMAX1-RV64-NEXT: vsll.vi v28, v25, 8
+; LMULMAX1-RV64-NEXT: lui a6, 4080
+; LMULMAX1-RV64-NEXT: vand.vx v28, v28, a6
+; LMULMAX1-RV64-NEXT: vsll.vi v25, v25, 24
+; LMULMAX1-RV64-NEXT: vor.vv v25, v25, v28
+; LMULMAX1-RV64-NEXT: vor.vv v25, v25, v27
+; LMULMAX1-RV64-NEXT: lui a4, 61681
+; LMULMAX1-RV64-NEXT: addiw t1, a4, -241
+; LMULMAX1-RV64-NEXT: vand.vx v27, v25, t1
+; LMULMAX1-RV64-NEXT: vsll.vi v27, v27, 4
+; LMULMAX1-RV64-NEXT: lui a5, 241
+; LMULMAX1-RV64-NEXT: addiw a5, a5, -241
+; LMULMAX1-RV64-NEXT: slli a5, a5, 12
+; LMULMAX1-RV64-NEXT: addi a5, a5, 240
+; LMULMAX1-RV64-NEXT: vand.vx v25, v25, a5
+; LMULMAX1-RV64-NEXT: vsrl.vi v25, v25, 4
+; LMULMAX1-RV64-NEXT: vor.vv v25, v25, v27
+; LMULMAX1-RV64-NEXT: lui a3, 209715
+; LMULMAX1-RV64-NEXT: addiw a3, a3, 819
+; LMULMAX1-RV64-NEXT: vand.vx v27, v25, a3
+; LMULMAX1-RV64-NEXT: vsll.vi v27, v27, 2
+; LMULMAX1-RV64-NEXT: lui a1, 205
+; LMULMAX1-RV64-NEXT: addiw a1, a1, -819
+; LMULMAX1-RV64-NEXT: slli a1, a1, 12
+; LMULMAX1-RV64-NEXT: addi a1, a1, -820
+; LMULMAX1-RV64-NEXT: vand.vx v25, v25, a1
+; LMULMAX1-RV64-NEXT: vsrl.vi v25, v25, 2
+; LMULMAX1-RV64-NEXT: vor.vv v25, v25, v27
+; LMULMAX1-RV64-NEXT: lui a2, 349525
+; LMULMAX1-RV64-NEXT: addiw a2, a2, 1365
+; LMULMAX1-RV64-NEXT: vand.vx v27, v25, a2
+; LMULMAX1-RV64-NEXT: vsll.vi v27, v27, 1
+; LMULMAX1-RV64-NEXT: lui a4, 171
+; LMULMAX1-RV64-NEXT: addiw a4, a4, -1365
+; LMULMAX1-RV64-NEXT: slli a4, a4, 12
+; LMULMAX1-RV64-NEXT: addi a4, a4, -1366
+; LMULMAX1-RV64-NEXT: vand.vx v25, v25, a4
+; LMULMAX1-RV64-NEXT: vsrl.vi v25, v25, 1
+; LMULMAX1-RV64-NEXT: vor.vv v25, v25, v27
+; LMULMAX1-RV64-NEXT: vsrl.vi v27, v26, 8
+; LMULMAX1-RV64-NEXT: vand.vx v27, v27, t0
+; LMULMAX1-RV64-NEXT: vsrl.vi v28, v26, 24
+; LMULMAX1-RV64-NEXT: vor.vv v27, v27, v28
+; LMULMAX1-RV64-NEXT: vsll.vi v28, v26, 8
+; LMULMAX1-RV64-NEXT: vand.vx v28, v28, a6
+; LMULMAX1-RV64-NEXT: vsll.vi v26, v26, 24
+; LMULMAX1-RV64-NEXT: vor.vv v26, v26, v28
+; LMULMAX1-RV64-NEXT: vor.vv v26, v26, v27
+; LMULMAX1-RV64-NEXT: vand.vx v27, v26, t1
+; LMULMAX1-RV64-NEXT: vsll.vi v27, v27, 4
+; LMULMAX1-RV64-NEXT: vand.vx v26, v26, a5
+; LMULMAX1-RV64-NEXT: vsrl.vi v26, v26, 4
+; LMULMAX1-RV64-NEXT: vor.vv v26, v26, v27
+; LMULMAX1-RV64-NEXT: vand.vx v27, v26, a3
+; LMULMAX1-RV64-NEXT: vsll.vi v27, v27, 2
+; LMULMAX1-RV64-NEXT: vand.vx v26, v26, a1
+; LMULMAX1-RV64-NEXT: vsrl.vi v26, v26, 2
+; LMULMAX1-RV64-NEXT: vor.vv v26, v26, v27
+; LMULMAX1-RV64-NEXT: vand.vx v27, v26, a2
+; LMULMAX1-RV64-NEXT: vsll.vi v27, v27, 1
+; LMULMAX1-RV64-NEXT: vand.vx v26, v26, a4
+; LMULMAX1-RV64-NEXT: vsrl.vi v26, v26, 1
+; LMULMAX1-RV64-NEXT: vor.vv v26, v26, v27
+; LMULMAX1-RV64-NEXT: vse32.v v26, (a0)
+; LMULMAX1-RV64-NEXT: vse32.v v25, (a7)
+; LMULMAX1-RV64-NEXT: ret
%a = load <8 x i32>, <8 x i32>* %x
%b = load <8 x i32>, <8 x i32>* %y
%c = call <8 x i32> @llvm.bitreverse.v8i32(<8 x i32> %a)
@@ -1146,431 +1309,565 @@ declare <8 x i32> @llvm.bitreverse.v8i32(<8 x i32>)
define void @bitreverse_v4i64(<4 x i64>* %x, <4 x i64>* %y) {
; LMULMAX2-RV32-LABEL: bitreverse_v4i64:
; LMULMAX2-RV32: # %bb.0:
-; LMULMAX2-RV32-NEXT: addi sp, sp, -64
-; LMULMAX2-RV32-NEXT: .cfi_def_cfa_offset 64
-; LMULMAX2-RV32-NEXT: sw ra, 60(sp) # 4-byte Folded Spill
-; LMULMAX2-RV32-NEXT: sw s0, 56(sp) # 4-byte Folded Spill
-; LMULMAX2-RV32-NEXT: .cfi_offset ra, -4
-; LMULMAX2-RV32-NEXT: .cfi_offset s0, -8
-; LMULMAX2-RV32-NEXT: addi s0, sp, 64
-; LMULMAX2-RV32-NEXT: .cfi_def_cfa s0, 0
-; LMULMAX2-RV32-NEXT: andi sp, sp, -32
; LMULMAX2-RV32-NEXT: vsetivli a1, 4, e64,m2,ta,mu
; LMULMAX2-RV32-NEXT: vle64.v v26, (a0)
-; LMULMAX2-RV32-NEXT: vsetvli zero, zero, e8,m2,ta,mu
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v26
-; LMULMAX2-RV32-NEXT: sb a1, 7(sp)
-; LMULMAX2-RV32-NEXT: vsetivli a1, 1, e8,m2,ta,mu
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 24
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 31(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 25
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 30(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 26
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 29(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 27
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 28(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 28
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 27(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 29
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 26(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 30
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 25(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 31
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 24(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 16
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 23(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 17
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 22(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 18
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 21(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 19
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 20(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 20
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 19(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 21
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 18(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 22
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 17(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 23
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 16(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 8
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 15(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 9
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 14(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 10
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 13(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 11
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 12(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 12
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 11(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 13
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 10(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 14
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 9(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 15
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 8(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 1
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 6(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 2
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 5(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 3
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 4(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 4
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 3(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 5
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 2(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 6
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 1(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v26, v26, 7
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v26
-; LMULMAX2-RV32-NEXT: sb a1, 0(sp)
-; LMULMAX2-RV32-NEXT: addi a1, zero, 32
-; LMULMAX2-RV32-NEXT: vsetvli a1, a1, e8,m2,ta,mu
-; LMULMAX2-RV32-NEXT: vle8.v v26, (sp)
-; LMULMAX2-RV32-NEXT: vand.vi v28, v26, 15
-; LMULMAX2-RV32-NEXT: vsll.vi v28, v28, 4
-; LMULMAX2-RV32-NEXT: vand.vi v26, v26, -16
-; LMULMAX2-RV32-NEXT: vsrl.vi v26, v26, 4
-; LMULMAX2-RV32-NEXT: vor.vv v26, v26, v28
-; LMULMAX2-RV32-NEXT: addi a1, zero, 51
-; LMULMAX2-RV32-NEXT: vand.vx v28, v26, a1
-; LMULMAX2-RV32-NEXT: vsll.vi v28, v28, 2
-; LMULMAX2-RV32-NEXT: addi a1, zero, 204
-; LMULMAX2-RV32-NEXT: vand.vx v26, v26, a1
-; LMULMAX2-RV32-NEXT: vsrl.vi v26, v26, 2
+; LMULMAX2-RV32-NEXT: lui a1, %hi(.LCPI5_0)
+; LMULMAX2-RV32-NEXT: addi a1, a1, %lo(.LCPI5_0)
+; LMULMAX2-RV32-NEXT: vsetivli a2, 8, e32,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vle32.v v28, (a1)
+; LMULMAX2-RV32-NEXT: vsetivli a1, 4, e64,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vsrl.vv v8, v26, v28
+; LMULMAX2-RV32-NEXT: lui a1, %hi(.LCPI5_1)
+; LMULMAX2-RV32-NEXT: addi a1, a1, %lo(.LCPI5_1)
+; LMULMAX2-RV32-NEXT: vsetivli a2, 8, e32,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vle32.v v30, (a1)
+; LMULMAX2-RV32-NEXT: vsetivli a1, 4, e64,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vsrl.vv v10, v26, v30
+; LMULMAX2-RV32-NEXT: lui a1, %hi(.LCPI5_2)
+; LMULMAX2-RV32-NEXT: addi a1, a1, %lo(.LCPI5_2)
+; LMULMAX2-RV32-NEXT: vsetivli a2, 8, e32,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vle32.v v12, (a1)
+; LMULMAX2-RV32-NEXT: vsetivli a1, 4, e64,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vand.vv v10, v10, v12
+; LMULMAX2-RV32-NEXT: vor.vv v10, v10, v8
+; LMULMAX2-RV32-NEXT: lui a1, %hi(.LCPI5_3)
+; LMULMAX2-RV32-NEXT: addi a1, a1, %lo(.LCPI5_3)
+; LMULMAX2-RV32-NEXT: vsetivli a2, 8, e32,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vle32.v v8, (a1)
+; LMULMAX2-RV32-NEXT: vsetivli a1, 4, e64,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vsrl.vv v12, v26, v8
+; LMULMAX2-RV32-NEXT: lui a1, %hi(.LCPI5_4)
+; LMULMAX2-RV32-NEXT: addi a1, a1, %lo(.LCPI5_4)
+; LMULMAX2-RV32-NEXT: vsetivli a2, 8, e32,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vle32.v v14, (a1)
+; LMULMAX2-RV32-NEXT: vsetivli a1, 4, e64,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vand.vv v12, v12, v14
+; LMULMAX2-RV32-NEXT: lui a1, %hi(.LCPI5_5)
+; LMULMAX2-RV32-NEXT: addi a1, a1, %lo(.LCPI5_5)
+; LMULMAX2-RV32-NEXT: vsetivli a2, 8, e32,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vle32.v v14, (a1)
+; LMULMAX2-RV32-NEXT: vsetivli a1, 4, e64,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vsrl.vv v16, v26, v14
+; LMULMAX2-RV32-NEXT: lui a1, %hi(.LCPI5_6)
+; LMULMAX2-RV32-NEXT: addi a1, a1, %lo(.LCPI5_6)
+; LMULMAX2-RV32-NEXT: vsetivli a2, 8, e32,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vle32.v v18, (a1)
+; LMULMAX2-RV32-NEXT: vsetivli a1, 4, e64,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vand.vv v16, v16, v18
+; LMULMAX2-RV32-NEXT: vor.vv v12, v16, v12
+; LMULMAX2-RV32-NEXT: vor.vv v10, v12, v10
+; LMULMAX2-RV32-NEXT: vsll.vv v12, v26, v14
+; LMULMAX2-RV32-NEXT: lui a1, %hi(.LCPI5_7)
+; LMULMAX2-RV32-NEXT: addi a1, a1, %lo(.LCPI5_7)
+; LMULMAX2-RV32-NEXT: vsetivli a2, 8, e32,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vle32.v v14, (a1)
+; LMULMAX2-RV32-NEXT: vsetivli a1, 4, e64,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vand.vv v12, v12, v14
+; LMULMAX2-RV32-NEXT: vsll.vv v8, v26, v8
+; LMULMAX2-RV32-NEXT: lui a1, %hi(.LCPI5_8)
+; LMULMAX2-RV32-NEXT: addi a1, a1, %lo(.LCPI5_8)
+; LMULMAX2-RV32-NEXT: vsetivli a2, 8, e32,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vle32.v v14, (a1)
+; LMULMAX2-RV32-NEXT: vsetivli a1, 4, e64,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vand.vv v8, v8, v14
+; LMULMAX2-RV32-NEXT: vor.vv v8, v8, v12
+; LMULMAX2-RV32-NEXT: vsll.vv v28, v26, v28
+; LMULMAX2-RV32-NEXT: vsll.vv v26, v26, v30
+; LMULMAX2-RV32-NEXT: lui a1, %hi(.LCPI5_9)
+; LMULMAX2-RV32-NEXT: addi a1, a1, %lo(.LCPI5_9)
+; LMULMAX2-RV32-NEXT: vsetivli a2, 8, e32,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vle32.v v30, (a1)
+; LMULMAX2-RV32-NEXT: vsetivli a1, 4, e64,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vand.vv v26, v26, v30
+; LMULMAX2-RV32-NEXT: vor.vv v26, v28, v26
+; LMULMAX2-RV32-NEXT: vor.vv v26, v26, v8
+; LMULMAX2-RV32-NEXT: vor.vv v26, v26, v10
+; LMULMAX2-RV32-NEXT: lui a1, 61681
+; LMULMAX2-RV32-NEXT: addi a1, a1, -241
+; LMULMAX2-RV32-NEXT: vsetivli a2, 8, e32,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vmv.v.x v28, a1
+; LMULMAX2-RV32-NEXT: vsetivli a1, 4, e64,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vand.vv v28, v26, v28
+; LMULMAX2-RV32-NEXT: lui a1, %hi(.LCPI5_10)
+; LMULMAX2-RV32-NEXT: addi a1, a1, %lo(.LCPI5_10)
+; LMULMAX2-RV32-NEXT: vsetivli a2, 8, e32,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vle32.v v30, (a1)
+; LMULMAX2-RV32-NEXT: vsetivli a1, 4, e64,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vsll.vv v28, v28, v30
+; LMULMAX2-RV32-NEXT: lui a1, 986895
+; LMULMAX2-RV32-NEXT: addi a1, a1, 240
+; LMULMAX2-RV32-NEXT: vsetivli a2, 8, e32,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vmv.v.x v8, a1
+; LMULMAX2-RV32-NEXT: vsetivli a1, 4, e64,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vand.vv v26, v26, v8
+; LMULMAX2-RV32-NEXT: vsrl.vv v26, v26, v30
; LMULMAX2-RV32-NEXT: vor.vv v26, v26, v28
-; LMULMAX2-RV32-NEXT: addi a1, zero, 85
-; LMULMAX2-RV32-NEXT: vand.vx v28, v26, a1
-; LMULMAX2-RV32-NEXT: vsll.vi v28, v28, 1
-; LMULMAX2-RV32-NEXT: addi a1, zero, 170
-; LMULMAX2-RV32-NEXT: vand.vx v26, v26, a1
-; LMULMAX2-RV32-NEXT: vsrl.vi v26, v26, 1
+; LMULMAX2-RV32-NEXT: lui a1, 209715
+; LMULMAX2-RV32-NEXT: addi a1, a1, 819
+; LMULMAX2-RV32-NEXT: vsetivli a2, 8, e32,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vmv.v.x v28, a1
+; LMULMAX2-RV32-NEXT: vsetivli a1, 4, e64,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vand.vv v28, v26, v28
+; LMULMAX2-RV32-NEXT: lui a1, %hi(.LCPI5_11)
+; LMULMAX2-RV32-NEXT: addi a1, a1, %lo(.LCPI5_11)
+; LMULMAX2-RV32-NEXT: vsetivli a2, 8, e32,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vle32.v v30, (a1)
+; LMULMAX2-RV32-NEXT: vsetivli a1, 4, e64,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vsll.vv v28, v28, v30
+; LMULMAX2-RV32-NEXT: lui a1, 838861
+; LMULMAX2-RV32-NEXT: addi a1, a1, -820
+; LMULMAX2-RV32-NEXT: vsetivli a2, 8, e32,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vmv.v.x v8, a1
+; LMULMAX2-RV32-NEXT: vsetivli a1, 4, e64,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vand.vv v26, v26, v8
+; LMULMAX2-RV32-NEXT: vsrl.vv v26, v26, v30
; LMULMAX2-RV32-NEXT: vor.vv v26, v26, v28
+; LMULMAX2-RV32-NEXT: lui a1, 349525
+; LMULMAX2-RV32-NEXT: addi a1, a1, 1365
+; LMULMAX2-RV32-NEXT: vsetivli a2, 8, e32,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vmv.v.x v28, a1
; LMULMAX2-RV32-NEXT: vsetivli a1, 4, e64,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vand.vv v28, v26, v28
+; LMULMAX2-RV32-NEXT: lui a1, %hi(.LCPI5_12)
+; LMULMAX2-RV32-NEXT: addi a1, a1, %lo(.LCPI5_12)
+; LMULMAX2-RV32-NEXT: vsetivli a2, 8, e32,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vle32.v v30, (a1)
+; LMULMAX2-RV32-NEXT: vsetivli a1, 4, e64,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vsll.vv v28, v28, v30
+; LMULMAX2-RV32-NEXT: lui a1, 699051
+; LMULMAX2-RV32-NEXT: addi a1, a1, -1366
+; LMULMAX2-RV32-NEXT: vsetivli a2, 8, e32,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vmv.v.x v8, a1
+; LMULMAX2-RV32-NEXT: vsetivli a1, 4, e64,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vand.vv v26, v26, v8
+; LMULMAX2-RV32-NEXT: vsrl.vv v26, v26, v30
+; LMULMAX2-RV32-NEXT: vor.vv v26, v26, v28
; LMULMAX2-RV32-NEXT: vse64.v v26, (a0)
-; LMULMAX2-RV32-NEXT: addi sp, s0, -64
-; LMULMAX2-RV32-NEXT: lw s0, 56(sp) # 4-byte Folded Reload
-; LMULMAX2-RV32-NEXT: lw ra, 60(sp) # 4-byte Folded Reload
-; LMULMAX2-RV32-NEXT: addi sp, sp, 64
; LMULMAX2-RV32-NEXT: ret
;
; LMULMAX2-RV64-LABEL: bitreverse_v4i64:
; LMULMAX2-RV64: # %bb.0:
-; LMULMAX2-RV64-NEXT: addi sp, sp, -64
-; LMULMAX2-RV64-NEXT: .cfi_def_cfa_offset 64
-; LMULMAX2-RV64-NEXT: sd ra, 56(sp) # 8-byte Folded Spill
-; LMULMAX2-RV64-NEXT: sd s0, 48(sp) # 8-byte Folded Spill
-; LMULMAX2-RV64-NEXT: .cfi_offset ra, -8
-; LMULMAX2-RV64-NEXT: .cfi_offset s0, -16
-; LMULMAX2-RV64-NEXT: addi s0, sp, 64
-; LMULMAX2-RV64-NEXT: .cfi_def_cfa s0, 0
-; LMULMAX2-RV64-NEXT: andi sp, sp, -32
; LMULMAX2-RV64-NEXT: vsetivli a1, 4, e64,m2,ta,mu
; LMULMAX2-RV64-NEXT: vle64.v v26, (a0)
-; LMULMAX2-RV64-NEXT: vsetvli zero, zero, e8,m2,ta,mu
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v26
-; LMULMAX2-RV64-NEXT: sb a1, 7(sp)
-; LMULMAX2-RV64-NEXT: vsetivli a1, 1, e8,m2,ta,mu
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 24
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 31(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 25
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 30(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 26
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 29(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 27
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 28(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 28
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 27(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 29
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 26(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 30
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 25(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 31
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 24(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 16
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 23(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 17
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 22(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 18
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 21(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 19
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 20(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 20
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 19(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 21
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 18(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 22
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 17(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 23
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 16(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 8
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 15(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 9
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 14(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 10
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 13(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 11
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 12(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 12
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 11(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 13
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 10(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 14
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 9(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 15
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 8(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 1
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 6(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 2
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 5(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 3
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 4(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 4
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 3(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 5
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 2(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 6
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 1(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v26, v26, 7
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v26
-; LMULMAX2-RV64-NEXT: sb a1, 0(sp)
-; LMULMAX2-RV64-NEXT: addi a1, zero, 32
-; LMULMAX2-RV64-NEXT: vsetvli a1, a1, e8,m2,ta,mu
-; LMULMAX2-RV64-NEXT: vle8.v v26, (sp)
-; LMULMAX2-RV64-NEXT: vand.vi v28, v26, 15
+; LMULMAX2-RV64-NEXT: addi a1, zero, 56
+; LMULMAX2-RV64-NEXT: vsrl.vx v28, v26, a1
+; LMULMAX2-RV64-NEXT: addi a2, zero, 40
+; LMULMAX2-RV64-NEXT: vsrl.vx v30, v26, a2
+; LMULMAX2-RV64-NEXT: lui a3, 16
+; LMULMAX2-RV64-NEXT: addiw a3, a3, -256
+; LMULMAX2-RV64-NEXT: vand.vx v30, v30, a3
+; LMULMAX2-RV64-NEXT: vor.vv v28, v30, v28
+; LMULMAX2-RV64-NEXT: vsrl.vi v30, v26, 24
+; LMULMAX2-RV64-NEXT: lui a3, 4080
+; LMULMAX2-RV64-NEXT: vand.vx v30, v30, a3
+; LMULMAX2-RV64-NEXT: vsrl.vi v8, v26, 8
+; LMULMAX2-RV64-NEXT: addi a3, zero, 255
+; LMULMAX2-RV64-NEXT: slli a4, a3, 24
+; LMULMAX2-RV64-NEXT: vand.vx v8, v8, a4
+; LMULMAX2-RV64-NEXT: vor.vv v30, v8, v30
+; LMULMAX2-RV64-NEXT: vor.vv v28, v30, v28
+; LMULMAX2-RV64-NEXT: vsll.vi v30, v26, 8
+; LMULMAX2-RV64-NEXT: slli a4, a3, 32
+; LMULMAX2-RV64-NEXT: vand.vx v30, v30, a4
+; LMULMAX2-RV64-NEXT: vsll.vi v8, v26, 24
+; LMULMAX2-RV64-NEXT: slli a4, a3, 40
+; LMULMAX2-RV64-NEXT: vand.vx v8, v8, a4
+; LMULMAX2-RV64-NEXT: vor.vv v30, v8, v30
+; LMULMAX2-RV64-NEXT: vsll.vx v8, v26, a1
+; LMULMAX2-RV64-NEXT: vsll.vx v26, v26, a2
+; LMULMAX2-RV64-NEXT: slli a1, a3, 48
+; LMULMAX2-RV64-NEXT: vand.vx v26, v26, a1
+; LMULMAX2-RV64-NEXT: vor.vv v26, v8, v26
+; LMULMAX2-RV64-NEXT: vor.vv v26, v26, v30
+; LMULMAX2-RV64-NEXT: vor.vv v26, v26, v28
+; LMULMAX2-RV64-NEXT: lui a1, 3855
+; LMULMAX2-RV64-NEXT: addiw a1, a1, 241
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, -241
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, 241
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, -241
+; LMULMAX2-RV64-NEXT: vand.vx v28, v26, a1
; LMULMAX2-RV64-NEXT: vsll.vi v28, v28, 4
-; LMULMAX2-RV64-NEXT: vand.vi v26, v26, -16
+; LMULMAX2-RV64-NEXT: lui a1, 1044721
+; LMULMAX2-RV64-NEXT: addiw a1, a1, -241
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, 241
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, -241
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, 240
+; LMULMAX2-RV64-NEXT: vand.vx v26, v26, a1
; LMULMAX2-RV64-NEXT: vsrl.vi v26, v26, 4
; LMULMAX2-RV64-NEXT: vor.vv v26, v26, v28
-; LMULMAX2-RV64-NEXT: addi a1, zero, 51
+; LMULMAX2-RV64-NEXT: lui a1, 13107
+; LMULMAX2-RV64-NEXT: addiw a1, a1, 819
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, 819
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, 819
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, 819
; LMULMAX2-RV64-NEXT: vand.vx v28, v26, a1
; LMULMAX2-RV64-NEXT: vsll.vi v28, v28, 2
-; LMULMAX2-RV64-NEXT: addi a1, zero, 204
+; LMULMAX2-RV64-NEXT: lui a1, 1035469
+; LMULMAX2-RV64-NEXT: addiw a1, a1, -819
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, -819
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, -819
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, -820
; LMULMAX2-RV64-NEXT: vand.vx v26, v26, a1
; LMULMAX2-RV64-NEXT: vsrl.vi v26, v26, 2
; LMULMAX2-RV64-NEXT: vor.vv v26, v26, v28
-; LMULMAX2-RV64-NEXT: addi a1, zero, 85
+; LMULMAX2-RV64-NEXT: lui a1, 21845
+; LMULMAX2-RV64-NEXT: addiw a1, a1, 1365
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, 1365
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, 1365
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, 1365
; LMULMAX2-RV64-NEXT: vand.vx v28, v26, a1
; LMULMAX2-RV64-NEXT: vsll.vi v28, v28, 1
-; LMULMAX2-RV64-NEXT: addi a1, zero, 170
+; LMULMAX2-RV64-NEXT: lui a1, 1026731
+; LMULMAX2-RV64-NEXT: addiw a1, a1, -1365
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, -1365
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, -1365
+; LMULMAX2-RV64-NEXT: slli a1, a1, 12
+; LMULMAX2-RV64-NEXT: addi a1, a1, -1366
; LMULMAX2-RV64-NEXT: vand.vx v26, v26, a1
; LMULMAX2-RV64-NEXT: vsrl.vi v26, v26, 1
; LMULMAX2-RV64-NEXT: vor.vv v26, v26, v28
-; LMULMAX2-RV64-NEXT: vsetivli a1, 4, e64,m2,ta,mu
; LMULMAX2-RV64-NEXT: vse64.v v26, (a0)
-; LMULMAX2-RV64-NEXT: addi sp, s0, -64
-; LMULMAX2-RV64-NEXT: ld s0, 48(sp) # 8-byte Folded Reload
-; LMULMAX2-RV64-NEXT: ld ra, 56(sp) # 8-byte Folded Reload
-; LMULMAX2-RV64-NEXT: addi sp, sp, 64
; LMULMAX2-RV64-NEXT: ret
;
-; LMULMAX1-LABEL: bitreverse_v4i64:
-; LMULMAX1: # %bb.0:
-; LMULMAX1-NEXT: addi sp, sp, -32
-; LMULMAX1-NEXT: .cfi_def_cfa_offset 32
-; LMULMAX1-NEXT: vsetivli a1, 2, e64,m1,ta,mu
-; LMULMAX1-NEXT: addi a1, a0, 16
-; LMULMAX1-NEXT: vle64.v v26, (a1)
-; LMULMAX1-NEXT: vle64.v v25, (a0)
-; LMULMAX1-NEXT: vsetvli zero, zero, e8,m1,ta,mu
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 7(sp)
-; LMULMAX1-NEXT: vsetivli a2, 1, e8,m1,ta,mu
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 8
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 15(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 9
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 14(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 10
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 13(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 11
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 12(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 12
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 11(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 13
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 10(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 14
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 9(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 15
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 8(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 1
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 6(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 2
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 5(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 3
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 4(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 4
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 3(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 5
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 2(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 6
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 1(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v26, 7
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 0(sp)
-; LMULMAX1-NEXT: vmv.x.s a2, v25
-; LMULMAX1-NEXT: sb a2, 23(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 8
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 31(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 9
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 30(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 10
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 29(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 11
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 28(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 12
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 27(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 13
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 26(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 14
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 25(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 15
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 24(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 1
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 22(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 2
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 21(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 3
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 20(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 4
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 19(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 5
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 18(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 6
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 17(sp)
-; LMULMAX1-NEXT: vslidedown.vi v25, v25, 7
-; LMULMAX1-NEXT: vmv.x.s a2, v25
-; LMULMAX1-NEXT: sb a2, 16(sp)
-; LMULMAX1-NEXT: vsetivli a2, 16, e8,m1,ta,mu
-; LMULMAX1-NEXT: vle8.v v25, (sp)
-; LMULMAX1-NEXT: vand.vi v26, v25, 15
-; LMULMAX1-NEXT: vsll.vi v26, v26, 4
-; LMULMAX1-NEXT: vand.vi v25, v25, -16
-; LMULMAX1-NEXT: vsrl.vi v25, v25, 4
-; LMULMAX1-NEXT: vor.vv v25, v25, v26
-; LMULMAX1-NEXT: addi a2, zero, 51
-; LMULMAX1-NEXT: vand.vx v26, v25, a2
-; LMULMAX1-NEXT: vsll.vi v26, v26, 2
-; LMULMAX1-NEXT: addi a3, zero, 204
-; LMULMAX1-NEXT: vand.vx v25, v25, a3
-; LMULMAX1-NEXT: vsrl.vi v25, v25, 2
-; LMULMAX1-NEXT: vor.vv v25, v25, v26
-; LMULMAX1-NEXT: addi a4, zero, 85
-; LMULMAX1-NEXT: vand.vx v26, v25, a4
-; LMULMAX1-NEXT: vsll.vi v26, v26, 1
-; LMULMAX1-NEXT: addi a5, zero, 170
-; LMULMAX1-NEXT: addi a6, sp, 16
-; LMULMAX1-NEXT: vle8.v v27, (a6)
-; LMULMAX1-NEXT: vand.vx v25, v25, a5
-; LMULMAX1-NEXT: vsrl.vi v25, v25, 1
-; LMULMAX1-NEXT: vor.vv v25, v25, v26
-; LMULMAX1-NEXT: vand.vi v26, v27, 15
-; LMULMAX1-NEXT: vsll.vi v26, v26, 4
-; LMULMAX1-NEXT: vand.vi v27, v27, -16
-; LMULMAX1-NEXT: vsrl.vi v27, v27, 4
-; LMULMAX1-NEXT: vor.vv v26, v27, v26
-; LMULMAX1-NEXT: vand.vx v27, v26, a2
-; LMULMAX1-NEXT: vsll.vi v27, v27, 2
-; LMULMAX1-NEXT: vand.vx v26, v26, a3
-; LMULMAX1-NEXT: vsrl.vi v26, v26, 2
-; LMULMAX1-NEXT: vor.vv v26, v26, v27
-; LMULMAX1-NEXT: vand.vx v27, v26, a4
-; LMULMAX1-NEXT: vsll.vi v27, v27, 1
-; LMULMAX1-NEXT: vand.vx v26, v26, a5
-; LMULMAX1-NEXT: vsrl.vi v26, v26, 1
-; LMULMAX1-NEXT: vor.vv v26, v26, v27
-; LMULMAX1-NEXT: vsetivli a2, 2, e64,m1,ta,mu
-; LMULMAX1-NEXT: vse64.v v26, (a0)
-; LMULMAX1-NEXT: vse64.v v25, (a1)
-; LMULMAX1-NEXT: addi sp, sp, 32
-; LMULMAX1-NEXT: ret
+; LMULMAX1-RV32-LABEL: bitreverse_v4i64:
+; LMULMAX1-RV32: # %bb.0:
+; LMULMAX1-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vle64.v v25, (a0)
+; LMULMAX1-RV32-NEXT: addi a1, a0, 16
+; LMULMAX1-RV32-NEXT: vle64.v v11, (a1)
+; LMULMAX1-RV32-NEXT: lui a2, %hi(.LCPI5_0)
+; LMULMAX1-RV32-NEXT: addi a2, a2, %lo(.LCPI5_0)
+; LMULMAX1-RV32-NEXT: vsetivli a3, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vle32.v v26, (a2)
+; LMULMAX1-RV32-NEXT: vsetivli a2, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vsrl.vv v28, v11, v26
+; LMULMAX1-RV32-NEXT: lui a2, %hi(.LCPI5_1)
+; LMULMAX1-RV32-NEXT: addi a2, a2, %lo(.LCPI5_1)
+; LMULMAX1-RV32-NEXT: vsetivli a3, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vle32.v v27, (a2)
+; LMULMAX1-RV32-NEXT: vsetivli a2, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vsrl.vv v30, v11, v27
+; LMULMAX1-RV32-NEXT: lui a2, %hi(.LCPI5_2)
+; LMULMAX1-RV32-NEXT: addi a2, a2, %lo(.LCPI5_2)
+; LMULMAX1-RV32-NEXT: vsetivli a3, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vle32.v v29, (a2)
+; LMULMAX1-RV32-NEXT: vsetivli a2, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vand.vv v30, v30, v29
+; LMULMAX1-RV32-NEXT: vor.vv v9, v30, v28
+; LMULMAX1-RV32-NEXT: lui a2, %hi(.LCPI5_3)
+; LMULMAX1-RV32-NEXT: addi a2, a2, %lo(.LCPI5_3)
+; LMULMAX1-RV32-NEXT: vsetivli a3, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vle32.v v28, (a2)
+; LMULMAX1-RV32-NEXT: vsetivli a2, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vsrl.vv v31, v11, v28
+; LMULMAX1-RV32-NEXT: lui a2, %hi(.LCPI5_4)
+; LMULMAX1-RV32-NEXT: addi a2, a2, %lo(.LCPI5_4)
+; LMULMAX1-RV32-NEXT: vsetivli a3, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vle32.v v30, (a2)
+; LMULMAX1-RV32-NEXT: vsetivli a2, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vand.vv v10, v31, v30
+; LMULMAX1-RV32-NEXT: lui a2, %hi(.LCPI5_5)
+; LMULMAX1-RV32-NEXT: addi a2, a2, %lo(.LCPI5_5)
+; LMULMAX1-RV32-NEXT: vsetivli a3, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vle32.v v31, (a2)
+; LMULMAX1-RV32-NEXT: vsetivli a2, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vsrl.vv v12, v11, v31
+; LMULMAX1-RV32-NEXT: lui a2, %hi(.LCPI5_6)
+; LMULMAX1-RV32-NEXT: addi a2, a2, %lo(.LCPI5_6)
+; LMULMAX1-RV32-NEXT: vsetivli a3, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vle32.v v8, (a2)
+; LMULMAX1-RV32-NEXT: vsetivli a2, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vand.vv v12, v12, v8
+; LMULMAX1-RV32-NEXT: vor.vv v10, v12, v10
+; LMULMAX1-RV32-NEXT: vor.vv v12, v10, v9
+; LMULMAX1-RV32-NEXT: vsll.vv v10, v11, v31
+; LMULMAX1-RV32-NEXT: lui a2, %hi(.LCPI5_7)
+; LMULMAX1-RV32-NEXT: addi a2, a2, %lo(.LCPI5_7)
+; LMULMAX1-RV32-NEXT: vsetivli a3, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vle32.v v9, (a2)
+; LMULMAX1-RV32-NEXT: vsetivli a2, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vand.vv v13, v10, v9
+; LMULMAX1-RV32-NEXT: vsll.vv v14, v11, v28
+; LMULMAX1-RV32-NEXT: lui a2, %hi(.LCPI5_8)
+; LMULMAX1-RV32-NEXT: addi a2, a2, %lo(.LCPI5_8)
+; LMULMAX1-RV32-NEXT: vsetivli a3, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vle32.v v10, (a2)
+; LMULMAX1-RV32-NEXT: vsetivli a2, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vand.vv v14, v14, v10
+; LMULMAX1-RV32-NEXT: vor.vv v13, v14, v13
+; LMULMAX1-RV32-NEXT: vsll.vv v14, v11, v26
+; LMULMAX1-RV32-NEXT: vsll.vv v15, v11, v27
+; LMULMAX1-RV32-NEXT: lui a2, %hi(.LCPI5_9)
+; LMULMAX1-RV32-NEXT: addi a2, a2, %lo(.LCPI5_9)
+; LMULMAX1-RV32-NEXT: vsetivli a3, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vle32.v v11, (a2)
+; LMULMAX1-RV32-NEXT: vsetivli a2, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vand.vv v15, v15, v11
+; LMULMAX1-RV32-NEXT: vor.vv v14, v14, v15
+; LMULMAX1-RV32-NEXT: vor.vv v13, v14, v13
+; LMULMAX1-RV32-NEXT: vor.vv v15, v13, v12
+; LMULMAX1-RV32-NEXT: lui a2, 61681
+; LMULMAX1-RV32-NEXT: addi a2, a2, -241
+; LMULMAX1-RV32-NEXT: vsetivli a3, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vmv.v.x v12, a2
+; LMULMAX1-RV32-NEXT: vsetivli a2, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vand.vv v14, v15, v12
+; LMULMAX1-RV32-NEXT: lui a2, %hi(.LCPI5_10)
+; LMULMAX1-RV32-NEXT: addi a2, a2, %lo(.LCPI5_10)
+; LMULMAX1-RV32-NEXT: vsetivli a3, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vle32.v v13, (a2)
+; LMULMAX1-RV32-NEXT: vsetivli a2, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vsll.vv v16, v14, v13
+; LMULMAX1-RV32-NEXT: lui a2, 986895
+; LMULMAX1-RV32-NEXT: addi a2, a2, 240
+; LMULMAX1-RV32-NEXT: vsetivli a3, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vmv.v.x v14, a2
+; LMULMAX1-RV32-NEXT: vsetivli a2, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vand.vv v15, v15, v14
+; LMULMAX1-RV32-NEXT: vsrl.vv v15, v15, v13
+; LMULMAX1-RV32-NEXT: vor.vv v17, v15, v16
+; LMULMAX1-RV32-NEXT: lui a2, 209715
+; LMULMAX1-RV32-NEXT: addi a2, a2, 819
+; LMULMAX1-RV32-NEXT: vsetivli a3, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vmv.v.x v15, a2
+; LMULMAX1-RV32-NEXT: vsetivli a2, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vand.vv v18, v17, v15
+; LMULMAX1-RV32-NEXT: lui a2, %hi(.LCPI5_11)
+; LMULMAX1-RV32-NEXT: addi a2, a2, %lo(.LCPI5_11)
+; LMULMAX1-RV32-NEXT: vsetivli a3, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vle32.v v16, (a2)
+; LMULMAX1-RV32-NEXT: vsetivli a2, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vsll.vv v18, v18, v16
+; LMULMAX1-RV32-NEXT: lui a2, 838861
+; LMULMAX1-RV32-NEXT: addi a2, a2, -820
+; LMULMAX1-RV32-NEXT: vsetivli a3, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vmv.v.x v19, a2
+; LMULMAX1-RV32-NEXT: vsetivli a2, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vand.vv v17, v17, v19
+; LMULMAX1-RV32-NEXT: vsrl.vv v17, v17, v16
+; LMULMAX1-RV32-NEXT: vor.vv v17, v17, v18
+; LMULMAX1-RV32-NEXT: lui a2, 349525
+; LMULMAX1-RV32-NEXT: addi a2, a2, 1365
+; LMULMAX1-RV32-NEXT: vsetivli a3, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vmv.v.x v18, a2
+; LMULMAX1-RV32-NEXT: vsetivli a2, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vand.vv v20, v17, v18
+; LMULMAX1-RV32-NEXT: lui a2, %hi(.LCPI5_12)
+; LMULMAX1-RV32-NEXT: addi a2, a2, %lo(.LCPI5_12)
+; LMULMAX1-RV32-NEXT: vsetivli a3, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vle32.v v21, (a2)
+; LMULMAX1-RV32-NEXT: vsetivli a2, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vsll.vv v20, v20, v21
+; LMULMAX1-RV32-NEXT: lui a2, 699051
+; LMULMAX1-RV32-NEXT: addi a2, a2, -1366
+; LMULMAX1-RV32-NEXT: vsetivli a3, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vmv.v.x v22, a2
+; LMULMAX1-RV32-NEXT: vsetivli a2, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vand.vv v17, v17, v22
+; LMULMAX1-RV32-NEXT: vsrl.vv v17, v17, v21
+; LMULMAX1-RV32-NEXT: vor.vv v17, v17, v20
+; LMULMAX1-RV32-NEXT: vsrl.vv v20, v25, v26
+; LMULMAX1-RV32-NEXT: vsrl.vv v23, v25, v27
+; LMULMAX1-RV32-NEXT: vand.vv v29, v23, v29
+; LMULMAX1-RV32-NEXT: vor.vv v29, v29, v20
+; LMULMAX1-RV32-NEXT: vsrl.vv v20, v25, v28
+; LMULMAX1-RV32-NEXT: vand.vv v30, v20, v30
+; LMULMAX1-RV32-NEXT: vsrl.vv v20, v25, v31
+; LMULMAX1-RV32-NEXT: vand.vv v8, v20, v8
+; LMULMAX1-RV32-NEXT: vor.vv v30, v8, v30
+; LMULMAX1-RV32-NEXT: vor.vv v29, v30, v29
+; LMULMAX1-RV32-NEXT: vsll.vv v30, v25, v31
+; LMULMAX1-RV32-NEXT: vand.vv v30, v30, v9
+; LMULMAX1-RV32-NEXT: vsll.vv v28, v25, v28
+; LMULMAX1-RV32-NEXT: vand.vv v28, v28, v10
+; LMULMAX1-RV32-NEXT: vor.vv v28, v28, v30
+; LMULMAX1-RV32-NEXT: vsll.vv v26, v25, v26
+; LMULMAX1-RV32-NEXT: vsll.vv v25, v25, v27
+; LMULMAX1-RV32-NEXT: vand.vv v25, v25, v11
+; LMULMAX1-RV32-NEXT: vor.vv v25, v26, v25
+; LMULMAX1-RV32-NEXT: vor.vv v25, v25, v28
+; LMULMAX1-RV32-NEXT: vor.vv v25, v25, v29
+; LMULMAX1-RV32-NEXT: vand.vv v26, v25, v12
+; LMULMAX1-RV32-NEXT: vsll.vv v26, v26, v13
+; LMULMAX1-RV32-NEXT: vand.vv v25, v25, v14
+; LMULMAX1-RV32-NEXT: vsrl.vv v25, v25, v13
+; LMULMAX1-RV32-NEXT: vor.vv v25, v25, v26
+; LMULMAX1-RV32-NEXT: vand.vv v26, v25, v15
+; LMULMAX1-RV32-NEXT: vsll.vv v26, v26, v16
+; LMULMAX1-RV32-NEXT: vand.vv v25, v25, v19
+; LMULMAX1-RV32-NEXT: vsrl.vv v25, v25, v16
+; LMULMAX1-RV32-NEXT: vor.vv v25, v25, v26
+; LMULMAX1-RV32-NEXT: vand.vv v26, v25, v18
+; LMULMAX1-RV32-NEXT: vsll.vv v26, v26, v21
+; LMULMAX1-RV32-NEXT: vand.vv v25, v25, v22
+; LMULMAX1-RV32-NEXT: vsrl.vv v25, v25, v21
+; LMULMAX1-RV32-NEXT: vor.vv v25, v25, v26
+; LMULMAX1-RV32-NEXT: vse64.v v25, (a0)
+; LMULMAX1-RV32-NEXT: vse64.v v17, (a1)
+; LMULMAX1-RV32-NEXT: ret
+;
+; LMULMAX1-RV64-LABEL: bitreverse_v4i64:
+; LMULMAX1-RV64: # %bb.0:
+; LMULMAX1-RV64-NEXT: addi sp, sp, -16
+; LMULMAX1-RV64-NEXT: .cfi_def_cfa_offset 16
+; LMULMAX1-RV64-NEXT: sd s0, 8(sp) # 8-byte Folded Spill
+; LMULMAX1-RV64-NEXT: .cfi_offset s0, -8
+; LMULMAX1-RV64-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX1-RV64-NEXT: addi t1, a0, 16
+; LMULMAX1-RV64-NEXT: vle64.v v26, (t1)
+; LMULMAX1-RV64-NEXT: vle64.v v25, (a0)
+; LMULMAX1-RV64-NEXT: addi a7, zero, 56
+; LMULMAX1-RV64-NEXT: vsrl.vx v27, v26, a7
+; LMULMAX1-RV64-NEXT: addi t0, zero, 40
+; LMULMAX1-RV64-NEXT: vsrl.vx v28, v26, t0
+; LMULMAX1-RV64-NEXT: lui a1, 16
+; LMULMAX1-RV64-NEXT: addiw t2, a1, -256
+; LMULMAX1-RV64-NEXT: vand.vx v28, v28, t2
+; LMULMAX1-RV64-NEXT: vor.vv v27, v28, v27
+; LMULMAX1-RV64-NEXT: vsrl.vi v28, v26, 24
+; LMULMAX1-RV64-NEXT: lui a6, 4080
+; LMULMAX1-RV64-NEXT: vand.vx v28, v28, a6
+; LMULMAX1-RV64-NEXT: vsrl.vi v29, v26, 8
+; LMULMAX1-RV64-NEXT: addi a1, zero, 255
+; LMULMAX1-RV64-NEXT: slli t3, a1, 24
+; LMULMAX1-RV64-NEXT: vand.vx v29, v29, t3
+; LMULMAX1-RV64-NEXT: vor.vv v28, v29, v28
+; LMULMAX1-RV64-NEXT: vor.vv v27, v28, v27
+; LMULMAX1-RV64-NEXT: vsll.vi v28, v26, 8
+; LMULMAX1-RV64-NEXT: slli t4, a1, 32
+; LMULMAX1-RV64-NEXT: vand.vx v28, v28, t4
+; LMULMAX1-RV64-NEXT: vsll.vi v29, v26, 24
+; LMULMAX1-RV64-NEXT: slli t5, a1, 40
+; LMULMAX1-RV64-NEXT: vand.vx v29, v29, t5
+; LMULMAX1-RV64-NEXT: vor.vv v28, v29, v28
+; LMULMAX1-RV64-NEXT: vsll.vx v29, v26, a7
+; LMULMAX1-RV64-NEXT: vsll.vx v26, v26, t0
+; LMULMAX1-RV64-NEXT: slli t6, a1, 48
+; LMULMAX1-RV64-NEXT: vand.vx v26, v26, t6
+; LMULMAX1-RV64-NEXT: vor.vv v26, v29, v26
+; LMULMAX1-RV64-NEXT: vor.vv v26, v26, v28
+; LMULMAX1-RV64-NEXT: vor.vv v26, v26, v27
+; LMULMAX1-RV64-NEXT: lui a4, 3855
+; LMULMAX1-RV64-NEXT: addiw a4, a4, 241
+; LMULMAX1-RV64-NEXT: slli a4, a4, 12
+; LMULMAX1-RV64-NEXT: addi a4, a4, -241
+; LMULMAX1-RV64-NEXT: slli a4, a4, 12
+; LMULMAX1-RV64-NEXT: addi a4, a4, 241
+; LMULMAX1-RV64-NEXT: slli a4, a4, 12
+; LMULMAX1-RV64-NEXT: addi a4, a4, -241
+; LMULMAX1-RV64-NEXT: vand.vx v27, v26, a4
+; LMULMAX1-RV64-NEXT: vsll.vi v27, v27, 4
+; LMULMAX1-RV64-NEXT: lui a5, 1044721
+; LMULMAX1-RV64-NEXT: addiw a5, a5, -241
+; LMULMAX1-RV64-NEXT: slli a5, a5, 12
+; LMULMAX1-RV64-NEXT: addi a5, a5, 241
+; LMULMAX1-RV64-NEXT: slli a5, a5, 12
+; LMULMAX1-RV64-NEXT: addi a5, a5, -241
+; LMULMAX1-RV64-NEXT: slli a5, a5, 12
+; LMULMAX1-RV64-NEXT: addi a5, a5, 240
+; LMULMAX1-RV64-NEXT: vand.vx v26, v26, a5
+; LMULMAX1-RV64-NEXT: vsrl.vi v26, v26, 4
+; LMULMAX1-RV64-NEXT: vor.vv v26, v26, v27
+; LMULMAX1-RV64-NEXT: lui a2, 13107
+; LMULMAX1-RV64-NEXT: addiw a2, a2, 819
+; LMULMAX1-RV64-NEXT: slli a2, a2, 12
+; LMULMAX1-RV64-NEXT: addi a2, a2, 819
+; LMULMAX1-RV64-NEXT: slli a2, a2, 12
+; LMULMAX1-RV64-NEXT: addi a2, a2, 819
+; LMULMAX1-RV64-NEXT: slli a2, a2, 12
+; LMULMAX1-RV64-NEXT: addi a2, a2, 819
+; LMULMAX1-RV64-NEXT: vand.vx v27, v26, a2
+; LMULMAX1-RV64-NEXT: vsll.vi v27, v27, 2
+; LMULMAX1-RV64-NEXT: lui a3, 1035469
+; LMULMAX1-RV64-NEXT: addiw a3, a3, -819
+; LMULMAX1-RV64-NEXT: slli a3, a3, 12
+; LMULMAX1-RV64-NEXT: addi a3, a3, -819
+; LMULMAX1-RV64-NEXT: slli a3, a3, 12
+; LMULMAX1-RV64-NEXT: addi a3, a3, -819
+; LMULMAX1-RV64-NEXT: slli a3, a3, 12
+; LMULMAX1-RV64-NEXT: addi a3, a3, -820
+; LMULMAX1-RV64-NEXT: vand.vx v26, v26, a3
+; LMULMAX1-RV64-NEXT: vsrl.vi v26, v26, 2
+; LMULMAX1-RV64-NEXT: vor.vv v26, v26, v27
+; LMULMAX1-RV64-NEXT: lui a1, 21845
+; LMULMAX1-RV64-NEXT: addiw a1, a1, 1365
+; LMULMAX1-RV64-NEXT: slli a1, a1, 12
+; LMULMAX1-RV64-NEXT: addi a1, a1, 1365
+; LMULMAX1-RV64-NEXT: slli a1, a1, 12
+; LMULMAX1-RV64-NEXT: addi a1, a1, 1365
+; LMULMAX1-RV64-NEXT: slli a1, a1, 12
+; LMULMAX1-RV64-NEXT: addi a1, a1, 1365
+; LMULMAX1-RV64-NEXT: vand.vx v27, v26, a1
+; LMULMAX1-RV64-NEXT: vsll.vi v27, v27, 1
+; LMULMAX1-RV64-NEXT: lui s0, 1026731
+; LMULMAX1-RV64-NEXT: addiw s0, s0, -1365
+; LMULMAX1-RV64-NEXT: slli s0, s0, 12
+; LMULMAX1-RV64-NEXT: addi s0, s0, -1365
+; LMULMAX1-RV64-NEXT: slli s0, s0, 12
+; LMULMAX1-RV64-NEXT: addi s0, s0, -1365
+; LMULMAX1-RV64-NEXT: slli s0, s0, 12
+; LMULMAX1-RV64-NEXT: addi s0, s0, -1366
+; LMULMAX1-RV64-NEXT: vand.vx v26, v26, s0
+; LMULMAX1-RV64-NEXT: vsrl.vi v26, v26, 1
+; LMULMAX1-RV64-NEXT: vor.vv v26, v26, v27
+; LMULMAX1-RV64-NEXT: vsrl.vx v27, v25, a7
+; LMULMAX1-RV64-NEXT: vsrl.vx v28, v25, t0
+; LMULMAX1-RV64-NEXT: vand.vx v28, v28, t2
+; LMULMAX1-RV64-NEXT: vor.vv v27, v28, v27
+; LMULMAX1-RV64-NEXT: vsrl.vi v28, v25, 24
+; LMULMAX1-RV64-NEXT: vand.vx v28, v28, a6
+; LMULMAX1-RV64-NEXT: vsrl.vi v29, v25, 8
+; LMULMAX1-RV64-NEXT: vand.vx v29, v29, t3
+; LMULMAX1-RV64-NEXT: vor.vv v28, v29, v28
+; LMULMAX1-RV64-NEXT: vor.vv v27, v28, v27
+; LMULMAX1-RV64-NEXT: vsll.vi v28, v25, 8
+; LMULMAX1-RV64-NEXT: vand.vx v28, v28, t4
+; LMULMAX1-RV64-NEXT: vsll.vi v29, v25, 24
+; LMULMAX1-RV64-NEXT: vand.vx v29, v29, t5
+; LMULMAX1-RV64-NEXT: vor.vv v28, v29, v28
+; LMULMAX1-RV64-NEXT: vsll.vx v29, v25, a7
+; LMULMAX1-RV64-NEXT: vsll.vx v25, v25, t0
+; LMULMAX1-RV64-NEXT: vand.vx v25, v25, t6
+; LMULMAX1-RV64-NEXT: vor.vv v25, v29, v25
+; LMULMAX1-RV64-NEXT: vor.vv v25, v25, v28
+; LMULMAX1-RV64-NEXT: vor.vv v25, v25, v27
+; LMULMAX1-RV64-NEXT: vand.vx v27, v25, a4
+; LMULMAX1-RV64-NEXT: vsll.vi v27, v27, 4
+; LMULMAX1-RV64-NEXT: vand.vx v25, v25, a5
+; LMULMAX1-RV64-NEXT: vsrl.vi v25, v25, 4
+; LMULMAX1-RV64-NEXT: vor.vv v25, v25, v27
+; LMULMAX1-RV64-NEXT: vand.vx v27, v25, a2
+; LMULMAX1-RV64-NEXT: vsll.vi v27, v27, 2
+; LMULMAX1-RV64-NEXT: vand.vx v25, v25, a3
+; LMULMAX1-RV64-NEXT: vsrl.vi v25, v25, 2
+; LMULMAX1-RV64-NEXT: vor.vv v25, v25, v27
+; LMULMAX1-RV64-NEXT: vand.vx v27, v25, a1
+; LMULMAX1-RV64-NEXT: vsll.vi v27, v27, 1
+; LMULMAX1-RV64-NEXT: vand.vx v25, v25, s0
+; LMULMAX1-RV64-NEXT: vsrl.vi v25, v25, 1
+; LMULMAX1-RV64-NEXT: vor.vv v25, v25, v27
+; LMULMAX1-RV64-NEXT: vse64.v v25, (a0)
+; LMULMAX1-RV64-NEXT: vse64.v v26, (t1)
+; LMULMAX1-RV64-NEXT: ld s0, 8(sp) # 8-byte Folded Reload
+; LMULMAX1-RV64-NEXT: addi sp, sp, 16
+; LMULMAX1-RV64-NEXT: ret
%a = load <4 x i64>, <4 x i64>* %x
%b = load <4 x i64>, <4 x i64>* %y
%c = call <4 x i64> @llvm.bitreverse.v4i64(<4 x i64> %a)
diff --git a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-bswap.ll b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-bswap.ll
index b9c999daa4d7..1694d683b024 100644
--- a/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-bswap.ll
+++ b/llvm/test/CodeGen/RISCV/rvv/fixed-vectors-bswap.ll
@@ -1,71 +1,281 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc -mtriple=riscv32 -mattr=+experimental-v -verify-machineinstrs -riscv-v-vector-bits-min=128 -riscv-v-fixed-length-vector-lmul-max=2 -verify-machineinstrs < %s | FileCheck %s --check-prefixes=CHECK,LMULMAX2-RV32
-; RUN: llc -mtriple=riscv64 -mattr=+experimental-v -verify-machineinstrs -riscv-v-vector-bits-min=128 -riscv-v-fixed-length-vector-lmul-max=2 -verify-machineinstrs < %s | FileCheck %s --check-prefixes=CHECK,LMULMAX2-RV64
-; RUN: llc -mtriple=riscv32 -mattr=+experimental-v -verify-machineinstrs -riscv-v-vector-bits-min=128 -riscv-v-fixed-length-vector-lmul-max=1 -verify-machineinstrs < %s | FileCheck %s --check-prefixes=CHECK,LMULMAX1
-; RUN: llc -mtriple=riscv64 -mattr=+experimental-v -verify-machineinstrs -riscv-v-vector-bits-min=128 -riscv-v-fixed-length-vector-lmul-max=1 -verify-machineinstrs < %s | FileCheck %s --check-prefixes=CHECK,LMULMAX1
+; RUN: llc -mtriple=riscv32 -mattr=+experimental-v -verify-machineinstrs -riscv-v-vector-bits-min=128 -riscv-v-fixed-length-vector-lmul-max=2 -verify-machineinstrs < %s | FileCheck %s --check-prefixes=LMULMAX2-RV32
+; RUN: llc -mtriple=riscv64 -mattr=+experimental-v -verify-machineinstrs -riscv-v-vector-bits-min=128 -riscv-v-fixed-length-vector-lmul-max=2 -verify-machineinstrs < %s | FileCheck %s --check-prefixes=LMULMAX2-RV64
+; RUN: llc -mtriple=riscv32 -mattr=+experimental-v -verify-machineinstrs -riscv-v-vector-bits-min=128 -riscv-v-fixed-length-vector-lmul-max=1 -verify-machineinstrs < %s | FileCheck %s --check-prefixes=LMULMAX1-RV32
+; RUN: llc -mtriple=riscv64 -mattr=+experimental-v -verify-machineinstrs -riscv-v-vector-bits-min=128 -riscv-v-fixed-length-vector-lmul-max=1 -verify-machineinstrs < %s | FileCheck %s --check-prefixes=LMULMAX1-RV64
define void @bswap_v8i16(<8 x i16>* %x, <8 x i16>* %y) {
-; CHECK-LABEL: bswap_v8i16:
-; CHECK: # %bb.0:
-; CHECK-NEXT: addi sp, sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
-; CHECK-NEXT: vsetivli a1, 8, e16,m1,ta,mu
-; CHECK-NEXT: vle16.v v25, (a0)
-; CHECK-NEXT: vsetvli zero, zero, e8,m1,ta,mu
-; CHECK-NEXT: vmv.x.s a1, v25
-; CHECK-NEXT: sb a1, 1(sp)
-; CHECK-NEXT: vsetivli a1, 1, e8,m1,ta,mu
-; CHECK-NEXT: vslidedown.vi v26, v25, 14
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 15(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 15
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 14(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 12
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 13(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 13
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 12(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 10
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 11(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 11
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 10(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 8
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 9(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 9
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 8(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 6
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 7(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 7
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 6(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 4
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 5(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 5
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 4(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 2
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 3(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 3
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 2(sp)
-; CHECK-NEXT: vslidedown.vi v25, v25, 1
-; CHECK-NEXT: vmv.x.s a1, v25
-; CHECK-NEXT: sb a1, 0(sp)
-; CHECK-NEXT: vsetivli a1, 16, e8,m1,ta,mu
-; CHECK-NEXT: vle8.v v25, (sp)
-; CHECK-NEXT: vsetivli a1, 8, e16,m1,ta,mu
-; CHECK-NEXT: vse16.v v25, (a0)
-; CHECK-NEXT: addi sp, sp, 16
-; CHECK-NEXT: ret
+; LMULMAX2-RV32-LABEL: bswap_v8i16:
+; LMULMAX2-RV32: # %bb.0:
+; LMULMAX2-RV32-NEXT: addi sp, sp, -16
+; LMULMAX2-RV32-NEXT: .cfi_def_cfa_offset 16
+; LMULMAX2-RV32-NEXT: vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vle16.v v25, (a0)
+; LMULMAX2-RV32-NEXT: vmv.x.s a1, v25
+; LMULMAX2-RV32-NEXT: slli a2, a1, 8
+; LMULMAX2-RV32-NEXT: slli a1, a1, 16
+; LMULMAX2-RV32-NEXT: srli a1, a1, 24
+; LMULMAX2-RV32-NEXT: or a1, a2, a1
+; LMULMAX2-RV32-NEXT: sh a1, 0(sp)
+; LMULMAX2-RV32-NEXT: vsetivli a1, 1, e16,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vslidedown.vi v26, v25, 7
+; LMULMAX2-RV32-NEXT: vmv.x.s a1, v26
+; LMULMAX2-RV32-NEXT: slli a2, a1, 8
+; LMULMAX2-RV32-NEXT: slli a1, a1, 16
+; LMULMAX2-RV32-NEXT: srli a1, a1, 24
+; LMULMAX2-RV32-NEXT: or a1, a2, a1
+; LMULMAX2-RV32-NEXT: sh a1, 14(sp)
+; LMULMAX2-RV32-NEXT: vslidedown.vi v26, v25, 6
+; LMULMAX2-RV32-NEXT: vmv.x.s a1, v26
+; LMULMAX2-RV32-NEXT: slli a2, a1, 8
+; LMULMAX2-RV32-NEXT: slli a1, a1, 16
+; LMULMAX2-RV32-NEXT: srli a1, a1, 24
+; LMULMAX2-RV32-NEXT: or a1, a2, a1
+; LMULMAX2-RV32-NEXT: sh a1, 12(sp)
+; LMULMAX2-RV32-NEXT: vslidedown.vi v26, v25, 5
+; LMULMAX2-RV32-NEXT: vmv.x.s a1, v26
+; LMULMAX2-RV32-NEXT: slli a2, a1, 8
+; LMULMAX2-RV32-NEXT: slli a1, a1, 16
+; LMULMAX2-RV32-NEXT: srli a1, a1, 24
+; LMULMAX2-RV32-NEXT: or a1, a2, a1
+; LMULMAX2-RV32-NEXT: sh a1, 10(sp)
+; LMULMAX2-RV32-NEXT: vslidedown.vi v26, v25, 4
+; LMULMAX2-RV32-NEXT: vmv.x.s a1, v26
+; LMULMAX2-RV32-NEXT: slli a2, a1, 8
+; LMULMAX2-RV32-NEXT: slli a1, a1, 16
+; LMULMAX2-RV32-NEXT: srli a1, a1, 24
+; LMULMAX2-RV32-NEXT: or a1, a2, a1
+; LMULMAX2-RV32-NEXT: sh a1, 8(sp)
+; LMULMAX2-RV32-NEXT: vslidedown.vi v26, v25, 3
+; LMULMAX2-RV32-NEXT: vmv.x.s a1, v26
+; LMULMAX2-RV32-NEXT: slli a2, a1, 8
+; LMULMAX2-RV32-NEXT: slli a1, a1, 16
+; LMULMAX2-RV32-NEXT: srli a1, a1, 24
+; LMULMAX2-RV32-NEXT: or a1, a2, a1
+; LMULMAX2-RV32-NEXT: sh a1, 6(sp)
+; LMULMAX2-RV32-NEXT: vslidedown.vi v26, v25, 2
+; LMULMAX2-RV32-NEXT: vmv.x.s a1, v26
+; LMULMAX2-RV32-NEXT: slli a2, a1, 8
+; LMULMAX2-RV32-NEXT: slli a1, a1, 16
+; LMULMAX2-RV32-NEXT: srli a1, a1, 24
+; LMULMAX2-RV32-NEXT: or a1, a2, a1
+; LMULMAX2-RV32-NEXT: sh a1, 4(sp)
+; LMULMAX2-RV32-NEXT: vslidedown.vi v25, v25, 1
+; LMULMAX2-RV32-NEXT: vmv.x.s a1, v25
+; LMULMAX2-RV32-NEXT: slli a2, a1, 8
+; LMULMAX2-RV32-NEXT: slli a1, a1, 16
+; LMULMAX2-RV32-NEXT: srli a1, a1, 24
+; LMULMAX2-RV32-NEXT: or a1, a2, a1
+; LMULMAX2-RV32-NEXT: sh a1, 2(sp)
+; LMULMAX2-RV32-NEXT: vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vle16.v v25, (sp)
+; LMULMAX2-RV32-NEXT: vse16.v v25, (a0)
+; LMULMAX2-RV32-NEXT: addi sp, sp, 16
+; LMULMAX2-RV32-NEXT: ret
+;
+; LMULMAX2-RV64-LABEL: bswap_v8i16:
+; LMULMAX2-RV64: # %bb.0:
+; LMULMAX2-RV64-NEXT: addi sp, sp, -16
+; LMULMAX2-RV64-NEXT: .cfi_def_cfa_offset 16
+; LMULMAX2-RV64-NEXT: vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX2-RV64-NEXT: vle16.v v25, (a0)
+; LMULMAX2-RV64-NEXT: vmv.x.s a1, v25
+; LMULMAX2-RV64-NEXT: slli a2, a1, 8
+; LMULMAX2-RV64-NEXT: slliw a1, a1, 16
+; LMULMAX2-RV64-NEXT: srliw a1, a1, 24
+; LMULMAX2-RV64-NEXT: or a1, a2, a1
+; LMULMAX2-RV64-NEXT: sh a1, 0(sp)
+; LMULMAX2-RV64-NEXT: vsetivli a1, 1, e16,m1,ta,mu
+; LMULMAX2-RV64-NEXT: vslidedown.vi v26, v25, 7
+; LMULMAX2-RV64-NEXT: vmv.x.s a1, v26
+; LMULMAX2-RV64-NEXT: slli a2, a1, 8
+; LMULMAX2-RV64-NEXT: slliw a1, a1, 16
+; LMULMAX2-RV64-NEXT: srliw a1, a1, 24
+; LMULMAX2-RV64-NEXT: or a1, a2, a1
+; LMULMAX2-RV64-NEXT: sh a1, 14(sp)
+; LMULMAX2-RV64-NEXT: vslidedown.vi v26, v25, 6
+; LMULMAX2-RV64-NEXT: vmv.x.s a1, v26
+; LMULMAX2-RV64-NEXT: slli a2, a1, 8
+; LMULMAX2-RV64-NEXT: slliw a1, a1, 16
+; LMULMAX2-RV64-NEXT: srliw a1, a1, 24
+; LMULMAX2-RV64-NEXT: or a1, a2, a1
+; LMULMAX2-RV64-NEXT: sh a1, 12(sp)
+; LMULMAX2-RV64-NEXT: vslidedown.vi v26, v25, 5
+; LMULMAX2-RV64-NEXT: vmv.x.s a1, v26
+; LMULMAX2-RV64-NEXT: slli a2, a1, 8
+; LMULMAX2-RV64-NEXT: slliw a1, a1, 16
+; LMULMAX2-RV64-NEXT: srliw a1, a1, 24
+; LMULMAX2-RV64-NEXT: or a1, a2, a1
+; LMULMAX2-RV64-NEXT: sh a1, 10(sp)
+; LMULMAX2-RV64-NEXT: vslidedown.vi v26, v25, 4
+; LMULMAX2-RV64-NEXT: vmv.x.s a1, v26
+; LMULMAX2-RV64-NEXT: slli a2, a1, 8
+; LMULMAX2-RV64-NEXT: slliw a1, a1, 16
+; LMULMAX2-RV64-NEXT: srliw a1, a1, 24
+; LMULMAX2-RV64-NEXT: or a1, a2, a1
+; LMULMAX2-RV64-NEXT: sh a1, 8(sp)
+; LMULMAX2-RV64-NEXT: vslidedown.vi v26, v25, 3
+; LMULMAX2-RV64-NEXT: vmv.x.s a1, v26
+; LMULMAX2-RV64-NEXT: slli a2, a1, 8
+; LMULMAX2-RV64-NEXT: slliw a1, a1, 16
+; LMULMAX2-RV64-NEXT: srliw a1, a1, 24
+; LMULMAX2-RV64-NEXT: or a1, a2, a1
+; LMULMAX2-RV64-NEXT: sh a1, 6(sp)
+; LMULMAX2-RV64-NEXT: vslidedown.vi v26, v25, 2
+; LMULMAX2-RV64-NEXT: vmv.x.s a1, v26
+; LMULMAX2-RV64-NEXT: slli a2, a1, 8
+; LMULMAX2-RV64-NEXT: slliw a1, a1, 16
+; LMULMAX2-RV64-NEXT: srliw a1, a1, 24
+; LMULMAX2-RV64-NEXT: or a1, a2, a1
+; LMULMAX2-RV64-NEXT: sh a1, 4(sp)
+; LMULMAX2-RV64-NEXT: vslidedown.vi v25, v25, 1
+; LMULMAX2-RV64-NEXT: vmv.x.s a1, v25
+; LMULMAX2-RV64-NEXT: slli a2, a1, 8
+; LMULMAX2-RV64-NEXT: slliw a1, a1, 16
+; LMULMAX2-RV64-NEXT: srliw a1, a1, 24
+; LMULMAX2-RV64-NEXT: or a1, a2, a1
+; LMULMAX2-RV64-NEXT: sh a1, 2(sp)
+; LMULMAX2-RV64-NEXT: vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX2-RV64-NEXT: vle16.v v25, (sp)
+; LMULMAX2-RV64-NEXT: vse16.v v25, (a0)
+; LMULMAX2-RV64-NEXT: addi sp, sp, 16
+; LMULMAX2-RV64-NEXT: ret
+;
+; LMULMAX1-RV32-LABEL: bswap_v8i16:
+; LMULMAX1-RV32: # %bb.0:
+; LMULMAX1-RV32-NEXT: addi sp, sp, -16
+; LMULMAX1-RV32-NEXT: .cfi_def_cfa_offset 16
+; LMULMAX1-RV32-NEXT: vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vle16.v v25, (a0)
+; LMULMAX1-RV32-NEXT: vmv.x.s a1, v25
+; LMULMAX1-RV32-NEXT: slli a2, a1, 8
+; LMULMAX1-RV32-NEXT: slli a1, a1, 16
+; LMULMAX1-RV32-NEXT: srli a1, a1, 24
+; LMULMAX1-RV32-NEXT: or a1, a2, a1
+; LMULMAX1-RV32-NEXT: sh a1, 0(sp)
+; LMULMAX1-RV32-NEXT: vsetivli a1, 1, e16,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vslidedown.vi v26, v25, 7
+; LMULMAX1-RV32-NEXT: vmv.x.s a1, v26
+; LMULMAX1-RV32-NEXT: slli a2, a1, 8
+; LMULMAX1-RV32-NEXT: slli a1, a1, 16
+; LMULMAX1-RV32-NEXT: srli a1, a1, 24
+; LMULMAX1-RV32-NEXT: or a1, a2, a1
+; LMULMAX1-RV32-NEXT: sh a1, 14(sp)
+; LMULMAX1-RV32-NEXT: vslidedown.vi v26, v25, 6
+; LMULMAX1-RV32-NEXT: vmv.x.s a1, v26
+; LMULMAX1-RV32-NEXT: slli a2, a1, 8
+; LMULMAX1-RV32-NEXT: slli a1, a1, 16
+; LMULMAX1-RV32-NEXT: srli a1, a1, 24
+; LMULMAX1-RV32-NEXT: or a1, a2, a1
+; LMULMAX1-RV32-NEXT: sh a1, 12(sp)
+; LMULMAX1-RV32-NEXT: vslidedown.vi v26, v25, 5
+; LMULMAX1-RV32-NEXT: vmv.x.s a1, v26
+; LMULMAX1-RV32-NEXT: slli a2, a1, 8
+; LMULMAX1-RV32-NEXT: slli a1, a1, 16
+; LMULMAX1-RV32-NEXT: srli a1, a1, 24
+; LMULMAX1-RV32-NEXT: or a1, a2, a1
+; LMULMAX1-RV32-NEXT: sh a1, 10(sp)
+; LMULMAX1-RV32-NEXT: vslidedown.vi v26, v25, 4
+; LMULMAX1-RV32-NEXT: vmv.x.s a1, v26
+; LMULMAX1-RV32-NEXT: slli a2, a1, 8
+; LMULMAX1-RV32-NEXT: slli a1, a1, 16
+; LMULMAX1-RV32-NEXT: srli a1, a1, 24
+; LMULMAX1-RV32-NEXT: or a1, a2, a1
+; LMULMAX1-RV32-NEXT: sh a1, 8(sp)
+; LMULMAX1-RV32-NEXT: vslidedown.vi v26, v25, 3
+; LMULMAX1-RV32-NEXT: vmv.x.s a1, v26
+; LMULMAX1-RV32-NEXT: slli a2, a1, 8
+; LMULMAX1-RV32-NEXT: slli a1, a1, 16
+; LMULMAX1-RV32-NEXT: srli a1, a1, 24
+; LMULMAX1-RV32-NEXT: or a1, a2, a1
+; LMULMAX1-RV32-NEXT: sh a1, 6(sp)
+; LMULMAX1-RV32-NEXT: vslidedown.vi v26, v25, 2
+; LMULMAX1-RV32-NEXT: vmv.x.s a1, v26
+; LMULMAX1-RV32-NEXT: slli a2, a1, 8
+; LMULMAX1-RV32-NEXT: slli a1, a1, 16
+; LMULMAX1-RV32-NEXT: srli a1, a1, 24
+; LMULMAX1-RV32-NEXT: or a1, a2, a1
+; LMULMAX1-RV32-NEXT: sh a1, 4(sp)
+; LMULMAX1-RV32-NEXT: vslidedown.vi v25, v25, 1
+; LMULMAX1-RV32-NEXT: vmv.x.s a1, v25
+; LMULMAX1-RV32-NEXT: slli a2, a1, 8
+; LMULMAX1-RV32-NEXT: slli a1, a1, 16
+; LMULMAX1-RV32-NEXT: srli a1, a1, 24
+; LMULMAX1-RV32-NEXT: or a1, a2, a1
+; LMULMAX1-RV32-NEXT: sh a1, 2(sp)
+; LMULMAX1-RV32-NEXT: vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vle16.v v25, (sp)
+; LMULMAX1-RV32-NEXT: vse16.v v25, (a0)
+; LMULMAX1-RV32-NEXT: addi sp, sp, 16
+; LMULMAX1-RV32-NEXT: ret
+;
+; LMULMAX1-RV64-LABEL: bswap_v8i16:
+; LMULMAX1-RV64: # %bb.0:
+; LMULMAX1-RV64-NEXT: addi sp, sp, -16
+; LMULMAX1-RV64-NEXT: .cfi_def_cfa_offset 16
+; LMULMAX1-RV64-NEXT: vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX1-RV64-NEXT: vle16.v v25, (a0)
+; LMULMAX1-RV64-NEXT: vmv.x.s a1, v25
+; LMULMAX1-RV64-NEXT: slli a2, a1, 8
+; LMULMAX1-RV64-NEXT: slliw a1, a1, 16
+; LMULMAX1-RV64-NEXT: srliw a1, a1, 24
+; LMULMAX1-RV64-NEXT: or a1, a2, a1
+; LMULMAX1-RV64-NEXT: sh a1, 0(sp)
+; LMULMAX1-RV64-NEXT: vsetivli a1, 1, e16,m1,ta,mu
+; LMULMAX1-RV64-NEXT: vslidedown.vi v26, v25, 7
+; LMULMAX1-RV64-NEXT: vmv.x.s a1, v26
+; LMULMAX1-RV64-NEXT: slli a2, a1, 8
+; LMULMAX1-RV64-NEXT: slliw a1, a1, 16
+; LMULMAX1-RV64-NEXT: srliw a1, a1, 24
+; LMULMAX1-RV64-NEXT: or a1, a2, a1
+; LMULMAX1-RV64-NEXT: sh a1, 14(sp)
+; LMULMAX1-RV64-NEXT: vslidedown.vi v26, v25, 6
+; LMULMAX1-RV64-NEXT: vmv.x.s a1, v26
+; LMULMAX1-RV64-NEXT: slli a2, a1, 8
+; LMULMAX1-RV64-NEXT: slliw a1, a1, 16
+; LMULMAX1-RV64-NEXT: srliw a1, a1, 24
+; LMULMAX1-RV64-NEXT: or a1, a2, a1
+; LMULMAX1-RV64-NEXT: sh a1, 12(sp)
+; LMULMAX1-RV64-NEXT: vslidedown.vi v26, v25, 5
+; LMULMAX1-RV64-NEXT: vmv.x.s a1, v26
+; LMULMAX1-RV64-NEXT: slli a2, a1, 8
+; LMULMAX1-RV64-NEXT: slliw a1, a1, 16
+; LMULMAX1-RV64-NEXT: srliw a1, a1, 24
+; LMULMAX1-RV64-NEXT: or a1, a2, a1
+; LMULMAX1-RV64-NEXT: sh a1, 10(sp)
+; LMULMAX1-RV64-NEXT: vslidedown.vi v26, v25, 4
+; LMULMAX1-RV64-NEXT: vmv.x.s a1, v26
+; LMULMAX1-RV64-NEXT: slli a2, a1, 8
+; LMULMAX1-RV64-NEXT: slliw a1, a1, 16
+; LMULMAX1-RV64-NEXT: srliw a1, a1, 24
+; LMULMAX1-RV64-NEXT: or a1, a2, a1
+; LMULMAX1-RV64-NEXT: sh a1, 8(sp)
+; LMULMAX1-RV64-NEXT: vslidedown.vi v26, v25, 3
+; LMULMAX1-RV64-NEXT: vmv.x.s a1, v26
+; LMULMAX1-RV64-NEXT: slli a2, a1, 8
+; LMULMAX1-RV64-NEXT: slliw a1, a1, 16
+; LMULMAX1-RV64-NEXT: srliw a1, a1, 24
+; LMULMAX1-RV64-NEXT: or a1, a2, a1
+; LMULMAX1-RV64-NEXT: sh a1, 6(sp)
+; LMULMAX1-RV64-NEXT: vslidedown.vi v26, v25, 2
+; LMULMAX1-RV64-NEXT: vmv.x.s a1, v26
+; LMULMAX1-RV64-NEXT: slli a2, a1, 8
+; LMULMAX1-RV64-NEXT: slliw a1, a1, 16
+; LMULMAX1-RV64-NEXT: srliw a1, a1, 24
+; LMULMAX1-RV64-NEXT: or a1, a2, a1
+; LMULMAX1-RV64-NEXT: sh a1, 4(sp)
+; LMULMAX1-RV64-NEXT: vslidedown.vi v25, v25, 1
+; LMULMAX1-RV64-NEXT: vmv.x.s a1, v25
+; LMULMAX1-RV64-NEXT: slli a2, a1, 8
+; LMULMAX1-RV64-NEXT: slliw a1, a1, 16
+; LMULMAX1-RV64-NEXT: srliw a1, a1, 24
+; LMULMAX1-RV64-NEXT: or a1, a2, a1
+; LMULMAX1-RV64-NEXT: sh a1, 2(sp)
+; LMULMAX1-RV64-NEXT: vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX1-RV64-NEXT: vle16.v v25, (sp)
+; LMULMAX1-RV64-NEXT: vse16.v v25, (a0)
+; LMULMAX1-RV64-NEXT: addi sp, sp, 16
+; LMULMAX1-RV64-NEXT: ret
%a = load <8 x i16>, <8 x i16>* %x
%b = load <8 x i16>, <8 x i16>* %y
%c = call <8 x i16> @llvm.bswap.v8i16(<8 x i16> %a)
@@ -75,67 +285,257 @@ define void @bswap_v8i16(<8 x i16>* %x, <8 x i16>* %y) {
declare <8 x i16> @llvm.bswap.v8i16(<8 x i16>)
define void @bswap_v4i32(<4 x i32>* %x, <4 x i32>* %y) {
-; CHECK-LABEL: bswap_v4i32:
-; CHECK: # %bb.0:
-; CHECK-NEXT: addi sp, sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
-; CHECK-NEXT: vsetivli a1, 4, e32,m1,ta,mu
-; CHECK-NEXT: vle32.v v25, (a0)
-; CHECK-NEXT: vsetvli zero, zero, e8,m1,ta,mu
-; CHECK-NEXT: vmv.x.s a1, v25
-; CHECK-NEXT: sb a1, 3(sp)
-; CHECK-NEXT: vsetivli a1, 1, e8,m1,ta,mu
-; CHECK-NEXT: vslidedown.vi v26, v25, 12
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 15(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 13
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 14(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 14
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 13(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 15
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 12(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 8
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 11(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 9
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 10(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 10
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 9(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 11
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 8(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 4
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 7(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 5
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 6(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 6
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 5(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 7
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 4(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 1
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 2(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 2
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 1(sp)
-; CHECK-NEXT: vslidedown.vi v25, v25, 3
-; CHECK-NEXT: vmv.x.s a1, v25
-; CHECK-NEXT: sb a1, 0(sp)
-; CHECK-NEXT: vsetivli a1, 16, e8,m1,ta,mu
-; CHECK-NEXT: vle8.v v25, (sp)
-; CHECK-NEXT: vsetivli a1, 4, e32,m1,ta,mu
-; CHECK-NEXT: vse32.v v25, (a0)
-; CHECK-NEXT: addi sp, sp, 16
-; CHECK-NEXT: ret
+; LMULMAX2-RV32-LABEL: bswap_v4i32:
+; LMULMAX2-RV32: # %bb.0:
+; LMULMAX2-RV32-NEXT: addi sp, sp, -16
+; LMULMAX2-RV32-NEXT: .cfi_def_cfa_offset 16
+; LMULMAX2-RV32-NEXT: vsetivli a1, 4, e32,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vle32.v v25, (a0)
+; LMULMAX2-RV32-NEXT: vmv.x.s a1, v25
+; LMULMAX2-RV32-NEXT: srli a2, a1, 8
+; LMULMAX2-RV32-NEXT: lui a3, 16
+; LMULMAX2-RV32-NEXT: addi a3, a3, -256
+; LMULMAX2-RV32-NEXT: and a2, a2, a3
+; LMULMAX2-RV32-NEXT: srli a4, a1, 24
+; LMULMAX2-RV32-NEXT: or a2, a2, a4
+; LMULMAX2-RV32-NEXT: slli a4, a1, 8
+; LMULMAX2-RV32-NEXT: lui a5, 4080
+; LMULMAX2-RV32-NEXT: and a4, a4, a5
+; LMULMAX2-RV32-NEXT: slli a1, a1, 24
+; LMULMAX2-RV32-NEXT: or a1, a1, a4
+; LMULMAX2-RV32-NEXT: or a1, a1, a2
+; LMULMAX2-RV32-NEXT: sw a1, 0(sp)
+; LMULMAX2-RV32-NEXT: vsetivli a1, 1, e32,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vslidedown.vi v26, v25, 3
+; LMULMAX2-RV32-NEXT: vmv.x.s a1, v26
+; LMULMAX2-RV32-NEXT: srli a2, a1, 8
+; LMULMAX2-RV32-NEXT: and a2, a2, a3
+; LMULMAX2-RV32-NEXT: srli a4, a1, 24
+; LMULMAX2-RV32-NEXT: or a2, a2, a4
+; LMULMAX2-RV32-NEXT: slli a4, a1, 8
+; LMULMAX2-RV32-NEXT: and a4, a4, a5
+; LMULMAX2-RV32-NEXT: slli a1, a1, 24
+; LMULMAX2-RV32-NEXT: or a1, a1, a4
+; LMULMAX2-RV32-NEXT: or a1, a1, a2
+; LMULMAX2-RV32-NEXT: sw a1, 12(sp)
+; LMULMAX2-RV32-NEXT: vslidedown.vi v26, v25, 2
+; LMULMAX2-RV32-NEXT: vmv.x.s a1, v26
+; LMULMAX2-RV32-NEXT: srli a2, a1, 8
+; LMULMAX2-RV32-NEXT: and a2, a2, a3
+; LMULMAX2-RV32-NEXT: srli a4, a1, 24
+; LMULMAX2-RV32-NEXT: or a2, a2, a4
+; LMULMAX2-RV32-NEXT: slli a4, a1, 8
+; LMULMAX2-RV32-NEXT: and a4, a4, a5
+; LMULMAX2-RV32-NEXT: slli a1, a1, 24
+; LMULMAX2-RV32-NEXT: or a1, a1, a4
+; LMULMAX2-RV32-NEXT: or a1, a1, a2
+; LMULMAX2-RV32-NEXT: sw a1, 8(sp)
+; LMULMAX2-RV32-NEXT: vslidedown.vi v25, v25, 1
+; LMULMAX2-RV32-NEXT: vmv.x.s a1, v25
+; LMULMAX2-RV32-NEXT: srli a2, a1, 8
+; LMULMAX2-RV32-NEXT: and a2, a2, a3
+; LMULMAX2-RV32-NEXT: srli a3, a1, 24
+; LMULMAX2-RV32-NEXT: or a2, a2, a3
+; LMULMAX2-RV32-NEXT: slli a3, a1, 8
+; LMULMAX2-RV32-NEXT: and a3, a3, a5
+; LMULMAX2-RV32-NEXT: slli a1, a1, 24
+; LMULMAX2-RV32-NEXT: or a1, a1, a3
+; LMULMAX2-RV32-NEXT: or a1, a1, a2
+; LMULMAX2-RV32-NEXT: sw a1, 4(sp)
+; LMULMAX2-RV32-NEXT: vsetivli a1, 4, e32,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vle32.v v25, (sp)
+; LMULMAX2-RV32-NEXT: vse32.v v25, (a0)
+; LMULMAX2-RV32-NEXT: addi sp, sp, 16
+; LMULMAX2-RV32-NEXT: ret
+;
+; LMULMAX2-RV64-LABEL: bswap_v4i32:
+; LMULMAX2-RV64: # %bb.0:
+; LMULMAX2-RV64-NEXT: addi sp, sp, -16
+; LMULMAX2-RV64-NEXT: .cfi_def_cfa_offset 16
+; LMULMAX2-RV64-NEXT: vsetivli a1, 4, e32,m1,ta,mu
+; LMULMAX2-RV64-NEXT: vle32.v v25, (a0)
+; LMULMAX2-RV64-NEXT: vmv.x.s a1, v25
+; LMULMAX2-RV64-NEXT: srliw a2, a1, 8
+; LMULMAX2-RV64-NEXT: lui a3, 16
+; LMULMAX2-RV64-NEXT: addiw a3, a3, -256
+; LMULMAX2-RV64-NEXT: and a2, a2, a3
+; LMULMAX2-RV64-NEXT: srliw a4, a1, 24
+; LMULMAX2-RV64-NEXT: or a2, a2, a4
+; LMULMAX2-RV64-NEXT: slli a4, a1, 8
+; LMULMAX2-RV64-NEXT: lui a5, 4080
+; LMULMAX2-RV64-NEXT: and a4, a4, a5
+; LMULMAX2-RV64-NEXT: slli a1, a1, 24
+; LMULMAX2-RV64-NEXT: or a1, a1, a4
+; LMULMAX2-RV64-NEXT: or a1, a1, a2
+; LMULMAX2-RV64-NEXT: sw a1, 0(sp)
+; LMULMAX2-RV64-NEXT: vsetivli a1, 1, e32,m1,ta,mu
+; LMULMAX2-RV64-NEXT: vslidedown.vi v26, v25, 3
+; LMULMAX2-RV64-NEXT: vmv.x.s a1, v26
+; LMULMAX2-RV64-NEXT: srliw a2, a1, 8
+; LMULMAX2-RV64-NEXT: and a2, a2, a3
+; LMULMAX2-RV64-NEXT: srliw a4, a1, 24
+; LMULMAX2-RV64-NEXT: or a2, a2, a4
+; LMULMAX2-RV64-NEXT: slli a4, a1, 8
+; LMULMAX2-RV64-NEXT: and a4, a4, a5
+; LMULMAX2-RV64-NEXT: slli a1, a1, 24
+; LMULMAX2-RV64-NEXT: or a1, a1, a4
+; LMULMAX2-RV64-NEXT: or a1, a1, a2
+; LMULMAX2-RV64-NEXT: sw a1, 12(sp)
+; LMULMAX2-RV64-NEXT: vslidedown.vi v26, v25, 2
+; LMULMAX2-RV64-NEXT: vmv.x.s a1, v26
+; LMULMAX2-RV64-NEXT: srliw a2, a1, 8
+; LMULMAX2-RV64-NEXT: and a2, a2, a3
+; LMULMAX2-RV64-NEXT: srliw a4, a1, 24
+; LMULMAX2-RV64-NEXT: or a2, a2, a4
+; LMULMAX2-RV64-NEXT: slli a4, a1, 8
+; LMULMAX2-RV64-NEXT: and a4, a4, a5
+; LMULMAX2-RV64-NEXT: slli a1, a1, 24
+; LMULMAX2-RV64-NEXT: or a1, a1, a4
+; LMULMAX2-RV64-NEXT: or a1, a1, a2
+; LMULMAX2-RV64-NEXT: sw a1, 8(sp)
+; LMULMAX2-RV64-NEXT: vslidedown.vi v25, v25, 1
+; LMULMAX2-RV64-NEXT: vmv.x.s a1, v25
+; LMULMAX2-RV64-NEXT: srliw a2, a1, 8
+; LMULMAX2-RV64-NEXT: and a2, a2, a3
+; LMULMAX2-RV64-NEXT: srliw a3, a1, 24
+; LMULMAX2-RV64-NEXT: or a2, a2, a3
+; LMULMAX2-RV64-NEXT: slli a3, a1, 8
+; LMULMAX2-RV64-NEXT: and a3, a3, a5
+; LMULMAX2-RV64-NEXT: slli a1, a1, 24
+; LMULMAX2-RV64-NEXT: or a1, a1, a3
+; LMULMAX2-RV64-NEXT: or a1, a1, a2
+; LMULMAX2-RV64-NEXT: sw a1, 4(sp)
+; LMULMAX2-RV64-NEXT: vsetivli a1, 4, e32,m1,ta,mu
+; LMULMAX2-RV64-NEXT: vle32.v v25, (sp)
+; LMULMAX2-RV64-NEXT: vse32.v v25, (a0)
+; LMULMAX2-RV64-NEXT: addi sp, sp, 16
+; LMULMAX2-RV64-NEXT: ret
+;
+; LMULMAX1-RV32-LABEL: bswap_v4i32:
+; LMULMAX1-RV32: # %bb.0:
+; LMULMAX1-RV32-NEXT: addi sp, sp, -16
+; LMULMAX1-RV32-NEXT: .cfi_def_cfa_offset 16
+; LMULMAX1-RV32-NEXT: vsetivli a1, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vle32.v v25, (a0)
+; LMULMAX1-RV32-NEXT: vmv.x.s a1, v25
+; LMULMAX1-RV32-NEXT: srli a2, a1, 8
+; LMULMAX1-RV32-NEXT: lui a3, 16
+; LMULMAX1-RV32-NEXT: addi a3, a3, -256
+; LMULMAX1-RV32-NEXT: and a2, a2, a3
+; LMULMAX1-RV32-NEXT: srli a4, a1, 24
+; LMULMAX1-RV32-NEXT: or a2, a2, a4
+; LMULMAX1-RV32-NEXT: slli a4, a1, 8
+; LMULMAX1-RV32-NEXT: lui a5, 4080
+; LMULMAX1-RV32-NEXT: and a4, a4, a5
+; LMULMAX1-RV32-NEXT: slli a1, a1, 24
+; LMULMAX1-RV32-NEXT: or a1, a1, a4
+; LMULMAX1-RV32-NEXT: or a1, a1, a2
+; LMULMAX1-RV32-NEXT: sw a1, 0(sp)
+; LMULMAX1-RV32-NEXT: vsetivli a1, 1, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vslidedown.vi v26, v25, 3
+; LMULMAX1-RV32-NEXT: vmv.x.s a1, v26
+; LMULMAX1-RV32-NEXT: srli a2, a1, 8
+; LMULMAX1-RV32-NEXT: and a2, a2, a3
+; LMULMAX1-RV32-NEXT: srli a4, a1, 24
+; LMULMAX1-RV32-NEXT: or a2, a2, a4
+; LMULMAX1-RV32-NEXT: slli a4, a1, 8
+; LMULMAX1-RV32-NEXT: and a4, a4, a5
+; LMULMAX1-RV32-NEXT: slli a1, a1, 24
+; LMULMAX1-RV32-NEXT: or a1, a1, a4
+; LMULMAX1-RV32-NEXT: or a1, a1, a2
+; LMULMAX1-RV32-NEXT: sw a1, 12(sp)
+; LMULMAX1-RV32-NEXT: vslidedown.vi v26, v25, 2
+; LMULMAX1-RV32-NEXT: vmv.x.s a1, v26
+; LMULMAX1-RV32-NEXT: srli a2, a1, 8
+; LMULMAX1-RV32-NEXT: and a2, a2, a3
+; LMULMAX1-RV32-NEXT: srli a4, a1, 24
+; LMULMAX1-RV32-NEXT: or a2, a2, a4
+; LMULMAX1-RV32-NEXT: slli a4, a1, 8
+; LMULMAX1-RV32-NEXT: and a4, a4, a5
+; LMULMAX1-RV32-NEXT: slli a1, a1, 24
+; LMULMAX1-RV32-NEXT: or a1, a1, a4
+; LMULMAX1-RV32-NEXT: or a1, a1, a2
+; LMULMAX1-RV32-NEXT: sw a1, 8(sp)
+; LMULMAX1-RV32-NEXT: vslidedown.vi v25, v25, 1
+; LMULMAX1-RV32-NEXT: vmv.x.s a1, v25
+; LMULMAX1-RV32-NEXT: srli a2, a1, 8
+; LMULMAX1-RV32-NEXT: and a2, a2, a3
+; LMULMAX1-RV32-NEXT: srli a3, a1, 24
+; LMULMAX1-RV32-NEXT: or a2, a2, a3
+; LMULMAX1-RV32-NEXT: slli a3, a1, 8
+; LMULMAX1-RV32-NEXT: and a3, a3, a5
+; LMULMAX1-RV32-NEXT: slli a1, a1, 24
+; LMULMAX1-RV32-NEXT: or a1, a1, a3
+; LMULMAX1-RV32-NEXT: or a1, a1, a2
+; LMULMAX1-RV32-NEXT: sw a1, 4(sp)
+; LMULMAX1-RV32-NEXT: vsetivli a1, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vle32.v v25, (sp)
+; LMULMAX1-RV32-NEXT: vse32.v v25, (a0)
+; LMULMAX1-RV32-NEXT: addi sp, sp, 16
+; LMULMAX1-RV32-NEXT: ret
+;
+; LMULMAX1-RV64-LABEL: bswap_v4i32:
+; LMULMAX1-RV64: # %bb.0:
+; LMULMAX1-RV64-NEXT: addi sp, sp, -16
+; LMULMAX1-RV64-NEXT: .cfi_def_cfa_offset 16
+; LMULMAX1-RV64-NEXT: vsetivli a1, 4, e32,m1,ta,mu
+; LMULMAX1-RV64-NEXT: vle32.v v25, (a0)
+; LMULMAX1-RV64-NEXT: vmv.x.s a1, v25
+; LMULMAX1-RV64-NEXT: srliw a2, a1, 8
+; LMULMAX1-RV64-NEXT: lui a3, 16
+; LMULMAX1-RV64-NEXT: addiw a3, a3, -256
+; LMULMAX1-RV64-NEXT: and a2, a2, a3
+; LMULMAX1-RV64-NEXT: srliw a4, a1, 24
+; LMULMAX1-RV64-NEXT: or a2, a2, a4
+; LMULMAX1-RV64-NEXT: slli a4, a1, 8
+; LMULMAX1-RV64-NEXT: lui a5, 4080
+; LMULMAX1-RV64-NEXT: and a4, a4, a5
+; LMULMAX1-RV64-NEXT: slli a1, a1, 24
+; LMULMAX1-RV64-NEXT: or a1, a1, a4
+; LMULMAX1-RV64-NEXT: or a1, a1, a2
+; LMULMAX1-RV64-NEXT: sw a1, 0(sp)
+; LMULMAX1-RV64-NEXT: vsetivli a1, 1, e32,m1,ta,mu
+; LMULMAX1-RV64-NEXT: vslidedown.vi v26, v25, 3
+; LMULMAX1-RV64-NEXT: vmv.x.s a1, v26
+; LMULMAX1-RV64-NEXT: srliw a2, a1, 8
+; LMULMAX1-RV64-NEXT: and a2, a2, a3
+; LMULMAX1-RV64-NEXT: srliw a4, a1, 24
+; LMULMAX1-RV64-NEXT: or a2, a2, a4
+; LMULMAX1-RV64-NEXT: slli a4, a1, 8
+; LMULMAX1-RV64-NEXT: and a4, a4, a5
+; LMULMAX1-RV64-NEXT: slli a1, a1, 24
+; LMULMAX1-RV64-NEXT: or a1, a1, a4
+; LMULMAX1-RV64-NEXT: or a1, a1, a2
+; LMULMAX1-RV64-NEXT: sw a1, 12(sp)
+; LMULMAX1-RV64-NEXT: vslidedown.vi v26, v25, 2
+; LMULMAX1-RV64-NEXT: vmv.x.s a1, v26
+; LMULMAX1-RV64-NEXT: srliw a2, a1, 8
+; LMULMAX1-RV64-NEXT: and a2, a2, a3
+; LMULMAX1-RV64-NEXT: srliw a4, a1, 24
+; LMULMAX1-RV64-NEXT: or a2, a2, a4
+; LMULMAX1-RV64-NEXT: slli a4, a1, 8
+; LMULMAX1-RV64-NEXT: and a4, a4, a5
+; LMULMAX1-RV64-NEXT: slli a1, a1, 24
+; LMULMAX1-RV64-NEXT: or a1, a1, a4
+; LMULMAX1-RV64-NEXT: or a1, a1, a2
+; LMULMAX1-RV64-NEXT: sw a1, 8(sp)
+; LMULMAX1-RV64-NEXT: vslidedown.vi v25, v25, 1
+; LMULMAX1-RV64-NEXT: vmv.x.s a1, v25
+; LMULMAX1-RV64-NEXT: srliw a2, a1, 8
+; LMULMAX1-RV64-NEXT: and a2, a2, a3
+; LMULMAX1-RV64-NEXT: srliw a3, a1, 24
+; LMULMAX1-RV64-NEXT: or a2, a2, a3
+; LMULMAX1-RV64-NEXT: slli a3, a1, 8
+; LMULMAX1-RV64-NEXT: and a3, a3, a5
+; LMULMAX1-RV64-NEXT: slli a1, a1, 24
+; LMULMAX1-RV64-NEXT: or a1, a1, a3
+; LMULMAX1-RV64-NEXT: or a1, a1, a2
+; LMULMAX1-RV64-NEXT: sw a1, 4(sp)
+; LMULMAX1-RV64-NEXT: vsetivli a1, 4, e32,m1,ta,mu
+; LMULMAX1-RV64-NEXT: vle32.v v25, (sp)
+; LMULMAX1-RV64-NEXT: vse32.v v25, (a0)
+; LMULMAX1-RV64-NEXT: addi sp, sp, 16
+; LMULMAX1-RV64-NEXT: ret
%a = load <4 x i32>, <4 x i32>* %x
%b = load <4 x i32>, <4 x i32>* %y
%c = call <4 x i32> @llvm.bswap.v4i32(<4 x i32> %a)
@@ -145,67 +545,271 @@ define void @bswap_v4i32(<4 x i32>* %x, <4 x i32>* %y) {
declare <4 x i32> @llvm.bswap.v4i32(<4 x i32>)
define void @bswap_v2i64(<2 x i64>* %x, <2 x i64>* %y) {
-; CHECK-LABEL: bswap_v2i64:
-; CHECK: # %bb.0:
-; CHECK-NEXT: addi sp, sp, -16
-; CHECK-NEXT: .cfi_def_cfa_offset 16
-; CHECK-NEXT: vsetivli a1, 2, e64,m1,ta,mu
-; CHECK-NEXT: vle64.v v25, (a0)
-; CHECK-NEXT: vsetvli zero, zero, e8,m1,ta,mu
-; CHECK-NEXT: vmv.x.s a1, v25
-; CHECK-NEXT: sb a1, 7(sp)
-; CHECK-NEXT: vsetivli a1, 1, e8,m1,ta,mu
-; CHECK-NEXT: vslidedown.vi v26, v25, 8
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 15(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 9
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 14(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 10
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 13(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 11
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 12(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 12
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 11(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 13
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 10(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 14
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 9(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 15
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 8(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 1
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 6(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 2
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 5(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 3
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 4(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 4
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 3(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 5
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 2(sp)
-; CHECK-NEXT: vslidedown.vi v26, v25, 6
-; CHECK-NEXT: vmv.x.s a1, v26
-; CHECK-NEXT: sb a1, 1(sp)
-; CHECK-NEXT: vslidedown.vi v25, v25, 7
-; CHECK-NEXT: vmv.x.s a1, v25
-; CHECK-NEXT: sb a1, 0(sp)
-; CHECK-NEXT: vsetivli a1, 16, e8,m1,ta,mu
-; CHECK-NEXT: vle8.v v25, (sp)
-; CHECK-NEXT: vsetivli a1, 2, e64,m1,ta,mu
-; CHECK-NEXT: vse64.v v25, (a0)
-; CHECK-NEXT: addi sp, sp, 16
-; CHECK-NEXT: ret
+; LMULMAX2-RV32-LABEL: bswap_v2i64:
+; LMULMAX2-RV32: # %bb.0:
+; LMULMAX2-RV32-NEXT: addi sp, sp, -16
+; LMULMAX2-RV32-NEXT: .cfi_def_cfa_offset 16
+; LMULMAX2-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vle64.v v25, (a0)
+; LMULMAX2-RV32-NEXT: vmv.x.s a1, v25
+; LMULMAX2-RV32-NEXT: srli a2, a1, 8
+; LMULMAX2-RV32-NEXT: lui a3, 16
+; LMULMAX2-RV32-NEXT: addi a3, a3, -256
+; LMULMAX2-RV32-NEXT: and a2, a2, a3
+; LMULMAX2-RV32-NEXT: srli a4, a1, 24
+; LMULMAX2-RV32-NEXT: or a2, a2, a4
+; LMULMAX2-RV32-NEXT: slli a4, a1, 8
+; LMULMAX2-RV32-NEXT: lui a6, 4080
+; LMULMAX2-RV32-NEXT: and a4, a4, a6
+; LMULMAX2-RV32-NEXT: slli a1, a1, 24
+; LMULMAX2-RV32-NEXT: or a1, a1, a4
+; LMULMAX2-RV32-NEXT: or a1, a1, a2
+; LMULMAX2-RV32-NEXT: sw a1, 4(sp)
+; LMULMAX2-RV32-NEXT: vsetivli a1, 1, e64,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vslidedown.vi v26, v25, 1
+; LMULMAX2-RV32-NEXT: vmv.x.s a1, v26
+; LMULMAX2-RV32-NEXT: srli a2, a1, 8
+; LMULMAX2-RV32-NEXT: and a2, a2, a3
+; LMULMAX2-RV32-NEXT: srli a4, a1, 24
+; LMULMAX2-RV32-NEXT: or a2, a2, a4
+; LMULMAX2-RV32-NEXT: slli a4, a1, 8
+; LMULMAX2-RV32-NEXT: and a4, a4, a6
+; LMULMAX2-RV32-NEXT: slli a1, a1, 24
+; LMULMAX2-RV32-NEXT: or a1, a1, a4
+; LMULMAX2-RV32-NEXT: or a1, a1, a2
+; LMULMAX2-RV32-NEXT: sw a1, 12(sp)
+; LMULMAX2-RV32-NEXT: addi a1, zero, 32
+; LMULMAX2-RV32-NEXT: vsrl.vx v25, v25, a1
+; LMULMAX2-RV32-NEXT: vmv.x.s a2, v25
+; LMULMAX2-RV32-NEXT: srli a4, a2, 8
+; LMULMAX2-RV32-NEXT: and a4, a4, a3
+; LMULMAX2-RV32-NEXT: srli a5, a2, 24
+; LMULMAX2-RV32-NEXT: or a4, a4, a5
+; LMULMAX2-RV32-NEXT: slli a5, a2, 8
+; LMULMAX2-RV32-NEXT: and a5, a5, a6
+; LMULMAX2-RV32-NEXT: slli a2, a2, 24
+; LMULMAX2-RV32-NEXT: or a2, a2, a5
+; LMULMAX2-RV32-NEXT: or a2, a2, a4
+; LMULMAX2-RV32-NEXT: sw a2, 0(sp)
+; LMULMAX2-RV32-NEXT: vsrl.vx v25, v26, a1
+; LMULMAX2-RV32-NEXT: vmv.x.s a1, v25
+; LMULMAX2-RV32-NEXT: srli a2, a1, 8
+; LMULMAX2-RV32-NEXT: and a2, a2, a3
+; LMULMAX2-RV32-NEXT: srli a3, a1, 24
+; LMULMAX2-RV32-NEXT: or a2, a2, a3
+; LMULMAX2-RV32-NEXT: slli a3, a1, 8
+; LMULMAX2-RV32-NEXT: and a3, a3, a6
+; LMULMAX2-RV32-NEXT: slli a1, a1, 24
+; LMULMAX2-RV32-NEXT: or a1, a1, a3
+; LMULMAX2-RV32-NEXT: or a1, a1, a2
+; LMULMAX2-RV32-NEXT: sw a1, 8(sp)
+; LMULMAX2-RV32-NEXT: vsetivli a1, 4, e32,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vle32.v v25, (sp)
+; LMULMAX2-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX2-RV32-NEXT: vse64.v v25, (a0)
+; LMULMAX2-RV32-NEXT: addi sp, sp, 16
+; LMULMAX2-RV32-NEXT: ret
+;
+; LMULMAX2-RV64-LABEL: bswap_v2i64:
+; LMULMAX2-RV64: # %bb.0:
+; LMULMAX2-RV64-NEXT: addi sp, sp, -16
+; LMULMAX2-RV64-NEXT: .cfi_def_cfa_offset 16
+; LMULMAX2-RV64-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX2-RV64-NEXT: vle64.v v25, (a0)
+; LMULMAX2-RV64-NEXT: vmv.x.s a1, v25
+; LMULMAX2-RV64-NEXT: srli a2, a1, 40
+; LMULMAX2-RV64-NEXT: lui a3, 16
+; LMULMAX2-RV64-NEXT: addiw a7, a3, -256
+; LMULMAX2-RV64-NEXT: and a2, a2, a7
+; LMULMAX2-RV64-NEXT: srli a4, a1, 56
+; LMULMAX2-RV64-NEXT: or a2, a2, a4
+; LMULMAX2-RV64-NEXT: srli a4, a1, 24
+; LMULMAX2-RV64-NEXT: lui a6, 4080
+; LMULMAX2-RV64-NEXT: and a4, a4, a6
+; LMULMAX2-RV64-NEXT: srli a5, a1, 8
+; LMULMAX2-RV64-NEXT: addi a3, zero, 255
+; LMULMAX2-RV64-NEXT: slli t0, a3, 24
+; LMULMAX2-RV64-NEXT: and a5, a5, t0
+; LMULMAX2-RV64-NEXT: or a4, a5, a4
+; LMULMAX2-RV64-NEXT: or t1, a4, a2
+; LMULMAX2-RV64-NEXT: slli a4, a1, 8
+; LMULMAX2-RV64-NEXT: slli t2, a3, 32
+; LMULMAX2-RV64-NEXT: and a4, a4, t2
+; LMULMAX2-RV64-NEXT: slli a2, a1, 24
+; LMULMAX2-RV64-NEXT: slli t3, a3, 40
+; LMULMAX2-RV64-NEXT: and a2, a2, t3
+; LMULMAX2-RV64-NEXT: or a2, a2, a4
+; LMULMAX2-RV64-NEXT: slli a4, a1, 40
+; LMULMAX2-RV64-NEXT: slli a3, a3, 48
+; LMULMAX2-RV64-NEXT: and a4, a4, a3
+; LMULMAX2-RV64-NEXT: slli a1, a1, 56
+; LMULMAX2-RV64-NEXT: or a1, a1, a4
+; LMULMAX2-RV64-NEXT: or a1, a1, a2
+; LMULMAX2-RV64-NEXT: or a1, a1, t1
+; LMULMAX2-RV64-NEXT: sd a1, 0(sp)
+; LMULMAX2-RV64-NEXT: vsetivli a1, 1, e64,m1,ta,mu
+; LMULMAX2-RV64-NEXT: vslidedown.vi v25, v25, 1
+; LMULMAX2-RV64-NEXT: vmv.x.s a1, v25
+; LMULMAX2-RV64-NEXT: srli a2, a1, 40
+; LMULMAX2-RV64-NEXT: and a2, a2, a7
+; LMULMAX2-RV64-NEXT: srli a4, a1, 56
+; LMULMAX2-RV64-NEXT: or a2, a2, a4
+; LMULMAX2-RV64-NEXT: srli a4, a1, 24
+; LMULMAX2-RV64-NEXT: and a4, a4, a6
+; LMULMAX2-RV64-NEXT: srli a5, a1, 8
+; LMULMAX2-RV64-NEXT: and a5, a5, t0
+; LMULMAX2-RV64-NEXT: or a4, a5, a4
+; LMULMAX2-RV64-NEXT: or a2, a4, a2
+; LMULMAX2-RV64-NEXT: slli a4, a1, 8
+; LMULMAX2-RV64-NEXT: and a4, a4, t2
+; LMULMAX2-RV64-NEXT: slli a5, a1, 24
+; LMULMAX2-RV64-NEXT: and a5, a5, t3
+; LMULMAX2-RV64-NEXT: or a4, a5, a4
+; LMULMAX2-RV64-NEXT: slli a5, a1, 40
+; LMULMAX2-RV64-NEXT: and a3, a5, a3
+; LMULMAX2-RV64-NEXT: slli a1, a1, 56
+; LMULMAX2-RV64-NEXT: or a1, a1, a3
+; LMULMAX2-RV64-NEXT: or a1, a1, a4
+; LMULMAX2-RV64-NEXT: or a1, a1, a2
+; LMULMAX2-RV64-NEXT: sd a1, 8(sp)
+; LMULMAX2-RV64-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX2-RV64-NEXT: vle64.v v25, (sp)
+; LMULMAX2-RV64-NEXT: vse64.v v25, (a0)
+; LMULMAX2-RV64-NEXT: addi sp, sp, 16
+; LMULMAX2-RV64-NEXT: ret
+;
+; LMULMAX1-RV32-LABEL: bswap_v2i64:
+; LMULMAX1-RV32: # %bb.0:
+; LMULMAX1-RV32-NEXT: addi sp, sp, -16
+; LMULMAX1-RV32-NEXT: .cfi_def_cfa_offset 16
+; LMULMAX1-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vle64.v v25, (a0)
+; LMULMAX1-RV32-NEXT: vmv.x.s a1, v25
+; LMULMAX1-RV32-NEXT: srli a2, a1, 8
+; LMULMAX1-RV32-NEXT: lui a3, 16
+; LMULMAX1-RV32-NEXT: addi a3, a3, -256
+; LMULMAX1-RV32-NEXT: and a2, a2, a3
+; LMULMAX1-RV32-NEXT: srli a4, a1, 24
+; LMULMAX1-RV32-NEXT: or a2, a2, a4
+; LMULMAX1-RV32-NEXT: slli a4, a1, 8
+; LMULMAX1-RV32-NEXT: lui a6, 4080
+; LMULMAX1-RV32-NEXT: and a4, a4, a6
+; LMULMAX1-RV32-NEXT: slli a1, a1, 24
+; LMULMAX1-RV32-NEXT: or a1, a1, a4
+; LMULMAX1-RV32-NEXT: or a1, a1, a2
+; LMULMAX1-RV32-NEXT: sw a1, 4(sp)
+; LMULMAX1-RV32-NEXT: vsetivli a1, 1, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vslidedown.vi v26, v25, 1
+; LMULMAX1-RV32-NEXT: vmv.x.s a1, v26
+; LMULMAX1-RV32-NEXT: srli a2, a1, 8
+; LMULMAX1-RV32-NEXT: and a2, a2, a3
+; LMULMAX1-RV32-NEXT: srli a4, a1, 24
+; LMULMAX1-RV32-NEXT: or a2, a2, a4
+; LMULMAX1-RV32-NEXT: slli a4, a1, 8
+; LMULMAX1-RV32-NEXT: and a4, a4, a6
+; LMULMAX1-RV32-NEXT: slli a1, a1, 24
+; LMULMAX1-RV32-NEXT: or a1, a1, a4
+; LMULMAX1-RV32-NEXT: or a1, a1, a2
+; LMULMAX1-RV32-NEXT: sw a1, 12(sp)
+; LMULMAX1-RV32-NEXT: addi a1, zero, 32
+; LMULMAX1-RV32-NEXT: vsrl.vx v25, v25, a1
+; LMULMAX1-RV32-NEXT: vmv.x.s a2, v25
+; LMULMAX1-RV32-NEXT: srli a4, a2, 8
+; LMULMAX1-RV32-NEXT: and a4, a4, a3
+; LMULMAX1-RV32-NEXT: srli a5, a2, 24
+; LMULMAX1-RV32-NEXT: or a4, a4, a5
+; LMULMAX1-RV32-NEXT: slli a5, a2, 8
+; LMULMAX1-RV32-NEXT: and a5, a5, a6
+; LMULMAX1-RV32-NEXT: slli a2, a2, 24
+; LMULMAX1-RV32-NEXT: or a2, a2, a5
+; LMULMAX1-RV32-NEXT: or a2, a2, a4
+; LMULMAX1-RV32-NEXT: sw a2, 0(sp)
+; LMULMAX1-RV32-NEXT: vsrl.vx v25, v26, a1
+; LMULMAX1-RV32-NEXT: vmv.x.s a1, v25
+; LMULMAX1-RV32-NEXT: srli a2, a1, 8
+; LMULMAX1-RV32-NEXT: and a2, a2, a3
+; LMULMAX1-RV32-NEXT: srli a3, a1, 24
+; LMULMAX1-RV32-NEXT: or a2, a2, a3
+; LMULMAX1-RV32-NEXT: slli a3, a1, 8
+; LMULMAX1-RV32-NEXT: and a3, a3, a6
+; LMULMAX1-RV32-NEXT: slli a1, a1, 24
+; LMULMAX1-RV32-NEXT: or a1, a1, a3
+; LMULMAX1-RV32-NEXT: or a1, a1, a2
+; LMULMAX1-RV32-NEXT: sw a1, 8(sp)
+; LMULMAX1-RV32-NEXT: vsetivli a1, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vle32.v v25, (sp)
+; LMULMAX1-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vse64.v v25, (a0)
+; LMULMAX1-RV32-NEXT: addi sp, sp, 16
+; LMULMAX1-RV32-NEXT: ret
+;
+; LMULMAX1-RV64-LABEL: bswap_v2i64:
+; LMULMAX1-RV64: # %bb.0:
+; LMULMAX1-RV64-NEXT: addi sp, sp, -16
+; LMULMAX1-RV64-NEXT: .cfi_def_cfa_offset 16
+; LMULMAX1-RV64-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX1-RV64-NEXT: vle64.v v25, (a0)
+; LMULMAX1-RV64-NEXT: vmv.x.s a1, v25
+; LMULMAX1-RV64-NEXT: srli a2, a1, 40
+; LMULMAX1-RV64-NEXT: lui a3, 16
+; LMULMAX1-RV64-NEXT: addiw a7, a3, -256
+; LMULMAX1-RV64-NEXT: and a2, a2, a7
+; LMULMAX1-RV64-NEXT: srli a4, a1, 56
+; LMULMAX1-RV64-NEXT: or a2, a2, a4
+; LMULMAX1-RV64-NEXT: srli a4, a1, 24
+; LMULMAX1-RV64-NEXT: lui a6, 4080
+; LMULMAX1-RV64-NEXT: and a4, a4, a6
+; LMULMAX1-RV64-NEXT: srli a5, a1, 8
+; LMULMAX1-RV64-NEXT: addi a3, zero, 255
+; LMULMAX1-RV64-NEXT: slli t0, a3, 24
+; LMULMAX1-RV64-NEXT: and a5, a5, t0
+; LMULMAX1-RV64-NEXT: or a4, a5, a4
+; LMULMAX1-RV64-NEXT: or t1, a4, a2
+; LMULMAX1-RV64-NEXT: slli a4, a1, 8
+; LMULMAX1-RV64-NEXT: slli t2, a3, 32
+; LMULMAX1-RV64-NEXT: and a4, a4, t2
+; LMULMAX1-RV64-NEXT: slli a2, a1, 24
+; LMULMAX1-RV64-NEXT: slli t3, a3, 40
+; LMULMAX1-RV64-NEXT: and a2, a2, t3
+; LMULMAX1-RV64-NEXT: or a2, a2, a4
+; LMULMAX1-RV64-NEXT: slli a4, a1, 40
+; LMULMAX1-RV64-NEXT: slli a3, a3, 48
+; LMULMAX1-RV64-NEXT: and a4, a4, a3
+; LMULMAX1-RV64-NEXT: slli a1, a1, 56
+; LMULMAX1-RV64-NEXT: or a1, a1, a4
+; LMULMAX1-RV64-NEXT: or a1, a1, a2
+; LMULMAX1-RV64-NEXT: or a1, a1, t1
+; LMULMAX1-RV64-NEXT: sd a1, 0(sp)
+; LMULMAX1-RV64-NEXT: vsetivli a1, 1, e64,m1,ta,mu
+; LMULMAX1-RV64-NEXT: vslidedown.vi v25, v25, 1
+; LMULMAX1-RV64-NEXT: vmv.x.s a1, v25
+; LMULMAX1-RV64-NEXT: srli a2, a1, 40
+; LMULMAX1-RV64-NEXT: and a2, a2, a7
+; LMULMAX1-RV64-NEXT: srli a4, a1, 56
+; LMULMAX1-RV64-NEXT: or a2, a2, a4
+; LMULMAX1-RV64-NEXT: srli a4, a1, 24
+; LMULMAX1-RV64-NEXT: and a4, a4, a6
+; LMULMAX1-RV64-NEXT: srli a5, a1, 8
+; LMULMAX1-RV64-NEXT: and a5, a5, t0
+; LMULMAX1-RV64-NEXT: or a4, a5, a4
+; LMULMAX1-RV64-NEXT: or a2, a4, a2
+; LMULMAX1-RV64-NEXT: slli a4, a1, 8
+; LMULMAX1-RV64-NEXT: and a4, a4, t2
+; LMULMAX1-RV64-NEXT: slli a5, a1, 24
+; LMULMAX1-RV64-NEXT: and a5, a5, t3
+; LMULMAX1-RV64-NEXT: or a4, a5, a4
+; LMULMAX1-RV64-NEXT: slli a5, a1, 40
+; LMULMAX1-RV64-NEXT: and a3, a5, a3
+; LMULMAX1-RV64-NEXT: slli a1, a1, 56
+; LMULMAX1-RV64-NEXT: or a1, a1, a3
+; LMULMAX1-RV64-NEXT: or a1, a1, a4
+; LMULMAX1-RV64-NEXT: or a1, a1, a2
+; LMULMAX1-RV64-NEXT: sd a1, 8(sp)
+; LMULMAX1-RV64-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX1-RV64-NEXT: vle64.v v25, (sp)
+; LMULMAX1-RV64-NEXT: vse64.v v25, (a0)
+; LMULMAX1-RV64-NEXT: addi sp, sp, 16
+; LMULMAX1-RV64-NEXT: ret
%a = load <2 x i64>, <2 x i64>* %x
%b = load <2 x i64>, <2 x i64>* %y
%c = call <2 x i64> @llvm.bswap.v2i64(<2 x i64> %a)
@@ -228,107 +832,120 @@ define void @bswap_v16i16(<16 x i16>* %x, <16 x i16>* %y) {
; LMULMAX2-RV32-NEXT: andi sp, sp, -32
; LMULMAX2-RV32-NEXT: vsetivli a1, 16, e16,m2,ta,mu
; LMULMAX2-RV32-NEXT: vle16.v v26, (a0)
-; LMULMAX2-RV32-NEXT: vsetvli zero, zero, e8,m2,ta,mu
; LMULMAX2-RV32-NEXT: vmv.x.s a1, v26
-; LMULMAX2-RV32-NEXT: sb a1, 1(sp)
-; LMULMAX2-RV32-NEXT: vsetivli a1, 1, e8,m2,ta,mu
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 30
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 31(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 31
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 30(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 28
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 29(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 29
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 28(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 26
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 27(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 27
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 26(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 24
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 25(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 25
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 24(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 22
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 23(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 23
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 22(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 20
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 21(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 21
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 20(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 18
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 19(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 19
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 18(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 16
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 17(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 17
+; LMULMAX2-RV32-NEXT: slli a2, a1, 8
+; LMULMAX2-RV32-NEXT: slli a1, a1, 16
+; LMULMAX2-RV32-NEXT: srli a1, a1, 24
+; LMULMAX2-RV32-NEXT: or a1, a2, a1
+; LMULMAX2-RV32-NEXT: sh a1, 0(sp)
+; LMULMAX2-RV32-NEXT: vsetivli a1, 1, e16,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 15
; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 16(sp)
+; LMULMAX2-RV32-NEXT: slli a2, a1, 8
+; LMULMAX2-RV32-NEXT: slli a1, a1, 16
+; LMULMAX2-RV32-NEXT: srli a1, a1, 24
+; LMULMAX2-RV32-NEXT: or a1, a2, a1
+; LMULMAX2-RV32-NEXT: sh a1, 30(sp)
; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 14
; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 15(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 15
+; LMULMAX2-RV32-NEXT: slli a2, a1, 8
+; LMULMAX2-RV32-NEXT: slli a1, a1, 16
+; LMULMAX2-RV32-NEXT: srli a1, a1, 24
+; LMULMAX2-RV32-NEXT: or a1, a2, a1
+; LMULMAX2-RV32-NEXT: sh a1, 28(sp)
+; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 13
; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 14(sp)
+; LMULMAX2-RV32-NEXT: slli a2, a1, 8
+; LMULMAX2-RV32-NEXT: slli a1, a1, 16
+; LMULMAX2-RV32-NEXT: srli a1, a1, 24
+; LMULMAX2-RV32-NEXT: or a1, a2, a1
+; LMULMAX2-RV32-NEXT: sh a1, 26(sp)
; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 12
; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 13(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 13
+; LMULMAX2-RV32-NEXT: slli a2, a1, 8
+; LMULMAX2-RV32-NEXT: slli a1, a1, 16
+; LMULMAX2-RV32-NEXT: srli a1, a1, 24
+; LMULMAX2-RV32-NEXT: or a1, a2, a1
+; LMULMAX2-RV32-NEXT: sh a1, 24(sp)
+; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 11
; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 12(sp)
+; LMULMAX2-RV32-NEXT: slli a2, a1, 8
+; LMULMAX2-RV32-NEXT: slli a1, a1, 16
+; LMULMAX2-RV32-NEXT: srli a1, a1, 24
+; LMULMAX2-RV32-NEXT: or a1, a2, a1
+; LMULMAX2-RV32-NEXT: sh a1, 22(sp)
; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 10
; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 11(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 11
+; LMULMAX2-RV32-NEXT: slli a2, a1, 8
+; LMULMAX2-RV32-NEXT: slli a1, a1, 16
+; LMULMAX2-RV32-NEXT: srli a1, a1, 24
+; LMULMAX2-RV32-NEXT: or a1, a2, a1
+; LMULMAX2-RV32-NEXT: sh a1, 20(sp)
+; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 9
; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 10(sp)
+; LMULMAX2-RV32-NEXT: slli a2, a1, 8
+; LMULMAX2-RV32-NEXT: slli a1, a1, 16
+; LMULMAX2-RV32-NEXT: srli a1, a1, 24
+; LMULMAX2-RV32-NEXT: or a1, a2, a1
+; LMULMAX2-RV32-NEXT: sh a1, 18(sp)
; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 8
; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 9(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 9
+; LMULMAX2-RV32-NEXT: slli a2, a1, 8
+; LMULMAX2-RV32-NEXT: slli a1, a1, 16
+; LMULMAX2-RV32-NEXT: srli a1, a1, 24
+; LMULMAX2-RV32-NEXT: or a1, a2, a1
+; LMULMAX2-RV32-NEXT: sh a1, 16(sp)
+; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 7
; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 8(sp)
+; LMULMAX2-RV32-NEXT: slli a2, a1, 8
+; LMULMAX2-RV32-NEXT: slli a1, a1, 16
+; LMULMAX2-RV32-NEXT: srli a1, a1, 24
+; LMULMAX2-RV32-NEXT: or a1, a2, a1
+; LMULMAX2-RV32-NEXT: sh a1, 14(sp)
; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 6
; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 7(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 7
+; LMULMAX2-RV32-NEXT: slli a2, a1, 8
+; LMULMAX2-RV32-NEXT: slli a1, a1, 16
+; LMULMAX2-RV32-NEXT: srli a1, a1, 24
+; LMULMAX2-RV32-NEXT: or a1, a2, a1
+; LMULMAX2-RV32-NEXT: sh a1, 12(sp)
+; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 5
; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 6(sp)
+; LMULMAX2-RV32-NEXT: slli a2, a1, 8
+; LMULMAX2-RV32-NEXT: slli a1, a1, 16
+; LMULMAX2-RV32-NEXT: srli a1, a1, 24
+; LMULMAX2-RV32-NEXT: or a1, a2, a1
+; LMULMAX2-RV32-NEXT: sh a1, 10(sp)
; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 4
; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 5(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 5
+; LMULMAX2-RV32-NEXT: slli a2, a1, 8
+; LMULMAX2-RV32-NEXT: slli a1, a1, 16
+; LMULMAX2-RV32-NEXT: srli a1, a1, 24
+; LMULMAX2-RV32-NEXT: or a1, a2, a1
+; LMULMAX2-RV32-NEXT: sh a1, 8(sp)
+; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 3
; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 4(sp)
+; LMULMAX2-RV32-NEXT: slli a2, a1, 8
+; LMULMAX2-RV32-NEXT: slli a1, a1, 16
+; LMULMAX2-RV32-NEXT: srli a1, a1, 24
+; LMULMAX2-RV32-NEXT: or a1, a2, a1
+; LMULMAX2-RV32-NEXT: sh a1, 6(sp)
; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 2
; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 3(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 3
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 2(sp)
+; LMULMAX2-RV32-NEXT: slli a2, a1, 8
+; LMULMAX2-RV32-NEXT: slli a1, a1, 16
+; LMULMAX2-RV32-NEXT: srli a1, a1, 24
+; LMULMAX2-RV32-NEXT: or a1, a2, a1
+; LMULMAX2-RV32-NEXT: sh a1, 4(sp)
; LMULMAX2-RV32-NEXT: vslidedown.vi v26, v26, 1
; LMULMAX2-RV32-NEXT: vmv.x.s a1, v26
-; LMULMAX2-RV32-NEXT: sb a1, 0(sp)
-; LMULMAX2-RV32-NEXT: addi a1, zero, 32
-; LMULMAX2-RV32-NEXT: vsetvli a1, a1, e8,m2,ta,mu
-; LMULMAX2-RV32-NEXT: vle8.v v26, (sp)
+; LMULMAX2-RV32-NEXT: slli a2, a1, 8
+; LMULMAX2-RV32-NEXT: slli a1, a1, 16
+; LMULMAX2-RV32-NEXT: srli a1, a1, 24
+; LMULMAX2-RV32-NEXT: or a1, a2, a1
+; LMULMAX2-RV32-NEXT: sh a1, 2(sp)
; LMULMAX2-RV32-NEXT: vsetivli a1, 16, e16,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vle16.v v26, (sp)
; LMULMAX2-RV32-NEXT: vse16.v v26, (a0)
; LMULMAX2-RV32-NEXT: addi sp, s0, -64
; LMULMAX2-RV32-NEXT: lw s0, 56(sp) # 4-byte Folded Reload
@@ -349,107 +966,120 @@ define void @bswap_v16i16(<16 x i16>* %x, <16 x i16>* %y) {
; LMULMAX2-RV64-NEXT: andi sp, sp, -32
; LMULMAX2-RV64-NEXT: vsetivli a1, 16, e16,m2,ta,mu
; LMULMAX2-RV64-NEXT: vle16.v v26, (a0)
-; LMULMAX2-RV64-NEXT: vsetvli zero, zero, e8,m2,ta,mu
; LMULMAX2-RV64-NEXT: vmv.x.s a1, v26
-; LMULMAX2-RV64-NEXT: sb a1, 1(sp)
-; LMULMAX2-RV64-NEXT: vsetivli a1, 1, e8,m2,ta,mu
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 30
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 31(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 31
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 30(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 28
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 29(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 29
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 28(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 26
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 27(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 27
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 26(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 24
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 25(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 25
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 24(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 22
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 23(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 23
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 22(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 20
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 21(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 21
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 20(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 18
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 19(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 19
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 18(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 16
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 17(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 17
+; LMULMAX2-RV64-NEXT: slli a2, a1, 8
+; LMULMAX2-RV64-NEXT: slliw a1, a1, 16
+; LMULMAX2-RV64-NEXT: srliw a1, a1, 24
+; LMULMAX2-RV64-NEXT: or a1, a2, a1
+; LMULMAX2-RV64-NEXT: sh a1, 0(sp)
+; LMULMAX2-RV64-NEXT: vsetivli a1, 1, e16,m2,ta,mu
+; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 15
; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 16(sp)
+; LMULMAX2-RV64-NEXT: slli a2, a1, 8
+; LMULMAX2-RV64-NEXT: slliw a1, a1, 16
+; LMULMAX2-RV64-NEXT: srliw a1, a1, 24
+; LMULMAX2-RV64-NEXT: or a1, a2, a1
+; LMULMAX2-RV64-NEXT: sh a1, 30(sp)
; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 14
; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 15(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 15
+; LMULMAX2-RV64-NEXT: slli a2, a1, 8
+; LMULMAX2-RV64-NEXT: slliw a1, a1, 16
+; LMULMAX2-RV64-NEXT: srliw a1, a1, 24
+; LMULMAX2-RV64-NEXT: or a1, a2, a1
+; LMULMAX2-RV64-NEXT: sh a1, 28(sp)
+; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 13
; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 14(sp)
+; LMULMAX2-RV64-NEXT: slli a2, a1, 8
+; LMULMAX2-RV64-NEXT: slliw a1, a1, 16
+; LMULMAX2-RV64-NEXT: srliw a1, a1, 24
+; LMULMAX2-RV64-NEXT: or a1, a2, a1
+; LMULMAX2-RV64-NEXT: sh a1, 26(sp)
; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 12
; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 13(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 13
+; LMULMAX2-RV64-NEXT: slli a2, a1, 8
+; LMULMAX2-RV64-NEXT: slliw a1, a1, 16
+; LMULMAX2-RV64-NEXT: srliw a1, a1, 24
+; LMULMAX2-RV64-NEXT: or a1, a2, a1
+; LMULMAX2-RV64-NEXT: sh a1, 24(sp)
+; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 11
; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 12(sp)
+; LMULMAX2-RV64-NEXT: slli a2, a1, 8
+; LMULMAX2-RV64-NEXT: slliw a1, a1, 16
+; LMULMAX2-RV64-NEXT: srliw a1, a1, 24
+; LMULMAX2-RV64-NEXT: or a1, a2, a1
+; LMULMAX2-RV64-NEXT: sh a1, 22(sp)
; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 10
; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 11(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 11
+; LMULMAX2-RV64-NEXT: slli a2, a1, 8
+; LMULMAX2-RV64-NEXT: slliw a1, a1, 16
+; LMULMAX2-RV64-NEXT: srliw a1, a1, 24
+; LMULMAX2-RV64-NEXT: or a1, a2, a1
+; LMULMAX2-RV64-NEXT: sh a1, 20(sp)
+; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 9
; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 10(sp)
+; LMULMAX2-RV64-NEXT: slli a2, a1, 8
+; LMULMAX2-RV64-NEXT: slliw a1, a1, 16
+; LMULMAX2-RV64-NEXT: srliw a1, a1, 24
+; LMULMAX2-RV64-NEXT: or a1, a2, a1
+; LMULMAX2-RV64-NEXT: sh a1, 18(sp)
; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 8
; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 9(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 9
+; LMULMAX2-RV64-NEXT: slli a2, a1, 8
+; LMULMAX2-RV64-NEXT: slliw a1, a1, 16
+; LMULMAX2-RV64-NEXT: srliw a1, a1, 24
+; LMULMAX2-RV64-NEXT: or a1, a2, a1
+; LMULMAX2-RV64-NEXT: sh a1, 16(sp)
+; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 7
; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 8(sp)
+; LMULMAX2-RV64-NEXT: slli a2, a1, 8
+; LMULMAX2-RV64-NEXT: slliw a1, a1, 16
+; LMULMAX2-RV64-NEXT: srliw a1, a1, 24
+; LMULMAX2-RV64-NEXT: or a1, a2, a1
+; LMULMAX2-RV64-NEXT: sh a1, 14(sp)
; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 6
; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 7(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 7
+; LMULMAX2-RV64-NEXT: slli a2, a1, 8
+; LMULMAX2-RV64-NEXT: slliw a1, a1, 16
+; LMULMAX2-RV64-NEXT: srliw a1, a1, 24
+; LMULMAX2-RV64-NEXT: or a1, a2, a1
+; LMULMAX2-RV64-NEXT: sh a1, 12(sp)
+; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 5
; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 6(sp)
+; LMULMAX2-RV64-NEXT: slli a2, a1, 8
+; LMULMAX2-RV64-NEXT: slliw a1, a1, 16
+; LMULMAX2-RV64-NEXT: srliw a1, a1, 24
+; LMULMAX2-RV64-NEXT: or a1, a2, a1
+; LMULMAX2-RV64-NEXT: sh a1, 10(sp)
; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 4
; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 5(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 5
+; LMULMAX2-RV64-NEXT: slli a2, a1, 8
+; LMULMAX2-RV64-NEXT: slliw a1, a1, 16
+; LMULMAX2-RV64-NEXT: srliw a1, a1, 24
+; LMULMAX2-RV64-NEXT: or a1, a2, a1
+; LMULMAX2-RV64-NEXT: sh a1, 8(sp)
+; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 3
; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 4(sp)
+; LMULMAX2-RV64-NEXT: slli a2, a1, 8
+; LMULMAX2-RV64-NEXT: slliw a1, a1, 16
+; LMULMAX2-RV64-NEXT: srliw a1, a1, 24
+; LMULMAX2-RV64-NEXT: or a1, a2, a1
+; LMULMAX2-RV64-NEXT: sh a1, 6(sp)
; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 2
; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 3(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 3
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 2(sp)
+; LMULMAX2-RV64-NEXT: slli a2, a1, 8
+; LMULMAX2-RV64-NEXT: slliw a1, a1, 16
+; LMULMAX2-RV64-NEXT: srliw a1, a1, 24
+; LMULMAX2-RV64-NEXT: or a1, a2, a1
+; LMULMAX2-RV64-NEXT: sh a1, 4(sp)
; LMULMAX2-RV64-NEXT: vslidedown.vi v26, v26, 1
; LMULMAX2-RV64-NEXT: vmv.x.s a1, v26
-; LMULMAX2-RV64-NEXT: sb a1, 0(sp)
-; LMULMAX2-RV64-NEXT: addi a1, zero, 32
-; LMULMAX2-RV64-NEXT: vsetvli a1, a1, e8,m2,ta,mu
-; LMULMAX2-RV64-NEXT: vle8.v v26, (sp)
+; LMULMAX2-RV64-NEXT: slli a2, a1, 8
+; LMULMAX2-RV64-NEXT: slliw a1, a1, 16
+; LMULMAX2-RV64-NEXT: srliw a1, a1, 24
+; LMULMAX2-RV64-NEXT: or a1, a2, a1
+; LMULMAX2-RV64-NEXT: sh a1, 2(sp)
; LMULMAX2-RV64-NEXT: vsetivli a1, 16, e16,m2,ta,mu
+; LMULMAX2-RV64-NEXT: vle16.v v26, (sp)
; LMULMAX2-RV64-NEXT: vse16.v v26, (a0)
; LMULMAX2-RV64-NEXT: addi sp, s0, -64
; LMULMAX2-RV64-NEXT: ld s0, 48(sp) # 8-byte Folded Reload
@@ -457,119 +1087,261 @@ define void @bswap_v16i16(<16 x i16>* %x, <16 x i16>* %y) {
; LMULMAX2-RV64-NEXT: addi sp, sp, 64
; LMULMAX2-RV64-NEXT: ret
;
-; LMULMAX1-LABEL: bswap_v16i16:
-; LMULMAX1: # %bb.0:
-; LMULMAX1-NEXT: addi sp, sp, -32
-; LMULMAX1-NEXT: .cfi_def_cfa_offset 32
-; LMULMAX1-NEXT: vsetivli a1, 8, e16,m1,ta,mu
-; LMULMAX1-NEXT: addi a1, a0, 16
-; LMULMAX1-NEXT: vle16.v v26, (a1)
-; LMULMAX1-NEXT: vle16.v v25, (a0)
-; LMULMAX1-NEXT: vsetvli zero, zero, e8,m1,ta,mu
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 1(sp)
-; LMULMAX1-NEXT: vsetivli a2, 1, e8,m1,ta,mu
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 14
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 15(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 15
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 14(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 12
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 13(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 13
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 12(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 10
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 11(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 11
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 10(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 8
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 9(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 9
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 8(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 6
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 7(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 7
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 6(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 4
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 5(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 5
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 4(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 2
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 3(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 3
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 2(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v26, 1
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 0(sp)
-; LMULMAX1-NEXT: vmv.x.s a2, v25
-; LMULMAX1-NEXT: sb a2, 17(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 14
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 31(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 15
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 30(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 12
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 29(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 13
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 28(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 10
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 27(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 11
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 26(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 8
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 25(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 9
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 24(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 6
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 23(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 7
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 22(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 4
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 21(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 5
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 20(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 2
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 19(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 3
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 18(sp)
-; LMULMAX1-NEXT: vslidedown.vi v25, v25, 1
-; LMULMAX1-NEXT: vmv.x.s a2, v25
-; LMULMAX1-NEXT: sb a2, 16(sp)
-; LMULMAX1-NEXT: vsetivli a2, 16, e8,m1,ta,mu
-; LMULMAX1-NEXT: addi a2, sp, 16
-; LMULMAX1-NEXT: vle8.v v25, (a2)
-; LMULMAX1-NEXT: vle8.v v26, (sp)
-; LMULMAX1-NEXT: vsetivli a2, 8, e16,m1,ta,mu
-; LMULMAX1-NEXT: vse16.v v25, (a0)
-; LMULMAX1-NEXT: vse16.v v26, (a1)
-; LMULMAX1-NEXT: addi sp, sp, 32
-; LMULMAX1-NEXT: ret
+; LMULMAX1-RV32-LABEL: bswap_v16i16:
+; LMULMAX1-RV32: # %bb.0:
+; LMULMAX1-RV32-NEXT: addi sp, sp, -32
+; LMULMAX1-RV32-NEXT: .cfi_def_cfa_offset 32
+; LMULMAX1-RV32-NEXT: vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX1-RV32-NEXT: addi a1, a0, 16
+; LMULMAX1-RV32-NEXT: vle16.v v26, (a1)
+; LMULMAX1-RV32-NEXT: vle16.v v25, (a0)
+; LMULMAX1-RV32-NEXT: vmv.x.s a2, v26
+; LMULMAX1-RV32-NEXT: slli a3, a2, 8
+; LMULMAX1-RV32-NEXT: slli a2, a2, 16
+; LMULMAX1-RV32-NEXT: srli a2, a2, 24
+; LMULMAX1-RV32-NEXT: or a2, a3, a2
+; LMULMAX1-RV32-NEXT: sh a2, 16(sp)
+; LMULMAX1-RV32-NEXT: vsetivli a2, 1, e16,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vslidedown.vi v27, v26, 7
+; LMULMAX1-RV32-NEXT: vmv.x.s a2, v27
+; LMULMAX1-RV32-NEXT: slli a3, a2, 8
+; LMULMAX1-RV32-NEXT: slli a2, a2, 16
+; LMULMAX1-RV32-NEXT: srli a2, a2, 24
+; LMULMAX1-RV32-NEXT: or a2, a3, a2
+; LMULMAX1-RV32-NEXT: sh a2, 30(sp)
+; LMULMAX1-RV32-NEXT: vslidedown.vi v27, v26, 6
+; LMULMAX1-RV32-NEXT: vmv.x.s a2, v27
+; LMULMAX1-RV32-NEXT: slli a3, a2, 8
+; LMULMAX1-RV32-NEXT: slli a2, a2, 16
+; LMULMAX1-RV32-NEXT: srli a2, a2, 24
+; LMULMAX1-RV32-NEXT: or a2, a3, a2
+; LMULMAX1-RV32-NEXT: sh a2, 28(sp)
+; LMULMAX1-RV32-NEXT: vslidedown.vi v27, v26, 5
+; LMULMAX1-RV32-NEXT: vmv.x.s a2, v27
+; LMULMAX1-RV32-NEXT: slli a3, a2, 8
+; LMULMAX1-RV32-NEXT: slli a2, a2, 16
+; LMULMAX1-RV32-NEXT: srli a2, a2, 24
+; LMULMAX1-RV32-NEXT: or a2, a3, a2
+; LMULMAX1-RV32-NEXT: sh a2, 26(sp)
+; LMULMAX1-RV32-NEXT: vslidedown.vi v27, v26, 4
+; LMULMAX1-RV32-NEXT: vmv.x.s a2, v27
+; LMULMAX1-RV32-NEXT: slli a3, a2, 8
+; LMULMAX1-RV32-NEXT: slli a2, a2, 16
+; LMULMAX1-RV32-NEXT: srli a2, a2, 24
+; LMULMAX1-RV32-NEXT: or a2, a3, a2
+; LMULMAX1-RV32-NEXT: sh a2, 24(sp)
+; LMULMAX1-RV32-NEXT: vslidedown.vi v27, v26, 3
+; LMULMAX1-RV32-NEXT: vmv.x.s a2, v27
+; LMULMAX1-RV32-NEXT: slli a3, a2, 8
+; LMULMAX1-RV32-NEXT: slli a2, a2, 16
+; LMULMAX1-RV32-NEXT: srli a2, a2, 24
+; LMULMAX1-RV32-NEXT: or a2, a3, a2
+; LMULMAX1-RV32-NEXT: sh a2, 22(sp)
+; LMULMAX1-RV32-NEXT: vslidedown.vi v27, v26, 2
+; LMULMAX1-RV32-NEXT: vmv.x.s a2, v27
+; LMULMAX1-RV32-NEXT: slli a3, a2, 8
+; LMULMAX1-RV32-NEXT: slli a2, a2, 16
+; LMULMAX1-RV32-NEXT: srli a2, a2, 24
+; LMULMAX1-RV32-NEXT: or a2, a3, a2
+; LMULMAX1-RV32-NEXT: sh a2, 20(sp)
+; LMULMAX1-RV32-NEXT: vslidedown.vi v26, v26, 1
+; LMULMAX1-RV32-NEXT: vmv.x.s a2, v26
+; LMULMAX1-RV32-NEXT: slli a3, a2, 8
+; LMULMAX1-RV32-NEXT: slli a2, a2, 16
+; LMULMAX1-RV32-NEXT: srli a2, a2, 24
+; LMULMAX1-RV32-NEXT: or a2, a3, a2
+; LMULMAX1-RV32-NEXT: sh a2, 18(sp)
+; LMULMAX1-RV32-NEXT: vmv.x.s a2, v25
+; LMULMAX1-RV32-NEXT: slli a3, a2, 8
+; LMULMAX1-RV32-NEXT: slli a2, a2, 16
+; LMULMAX1-RV32-NEXT: srli a2, a2, 24
+; LMULMAX1-RV32-NEXT: or a2, a3, a2
+; LMULMAX1-RV32-NEXT: sh a2, 0(sp)
+; LMULMAX1-RV32-NEXT: vslidedown.vi v26, v25, 7
+; LMULMAX1-RV32-NEXT: vmv.x.s a2, v26
+; LMULMAX1-RV32-NEXT: slli a3, a2, 8
+; LMULMAX1-RV32-NEXT: slli a2, a2, 16
+; LMULMAX1-RV32-NEXT: srli a2, a2, 24
+; LMULMAX1-RV32-NEXT: or a2, a3, a2
+; LMULMAX1-RV32-NEXT: sh a2, 14(sp)
+; LMULMAX1-RV32-NEXT: vslidedown.vi v26, v25, 6
+; LMULMAX1-RV32-NEXT: vmv.x.s a2, v26
+; LMULMAX1-RV32-NEXT: slli a3, a2, 8
+; LMULMAX1-RV32-NEXT: slli a2, a2, 16
+; LMULMAX1-RV32-NEXT: srli a2, a2, 24
+; LMULMAX1-RV32-NEXT: or a2, a3, a2
+; LMULMAX1-RV32-NEXT: sh a2, 12(sp)
+; LMULMAX1-RV32-NEXT: vslidedown.vi v26, v25, 5
+; LMULMAX1-RV32-NEXT: vmv.x.s a2, v26
+; LMULMAX1-RV32-NEXT: slli a3, a2, 8
+; LMULMAX1-RV32-NEXT: slli a2, a2, 16
+; LMULMAX1-RV32-NEXT: srli a2, a2, 24
+; LMULMAX1-RV32-NEXT: or a2, a3, a2
+; LMULMAX1-RV32-NEXT: sh a2, 10(sp)
+; LMULMAX1-RV32-NEXT: vslidedown.vi v26, v25, 4
+; LMULMAX1-RV32-NEXT: vmv.x.s a2, v26
+; LMULMAX1-RV32-NEXT: slli a3, a2, 8
+; LMULMAX1-RV32-NEXT: slli a2, a2, 16
+; LMULMAX1-RV32-NEXT: srli a2, a2, 24
+; LMULMAX1-RV32-NEXT: or a2, a3, a2
+; LMULMAX1-RV32-NEXT: sh a2, 8(sp)
+; LMULMAX1-RV32-NEXT: vslidedown.vi v26, v25, 3
+; LMULMAX1-RV32-NEXT: vmv.x.s a2, v26
+; LMULMAX1-RV32-NEXT: slli a3, a2, 8
+; LMULMAX1-RV32-NEXT: slli a2, a2, 16
+; LMULMAX1-RV32-NEXT: srli a2, a2, 24
+; LMULMAX1-RV32-NEXT: or a2, a3, a2
+; LMULMAX1-RV32-NEXT: sh a2, 6(sp)
+; LMULMAX1-RV32-NEXT: vslidedown.vi v26, v25, 2
+; LMULMAX1-RV32-NEXT: vmv.x.s a2, v26
+; LMULMAX1-RV32-NEXT: slli a3, a2, 8
+; LMULMAX1-RV32-NEXT: slli a2, a2, 16
+; LMULMAX1-RV32-NEXT: srli a2, a2, 24
+; LMULMAX1-RV32-NEXT: or a2, a3, a2
+; LMULMAX1-RV32-NEXT: sh a2, 4(sp)
+; LMULMAX1-RV32-NEXT: vslidedown.vi v25, v25, 1
+; LMULMAX1-RV32-NEXT: vmv.x.s a2, v25
+; LMULMAX1-RV32-NEXT: slli a3, a2, 8
+; LMULMAX1-RV32-NEXT: slli a2, a2, 16
+; LMULMAX1-RV32-NEXT: srli a2, a2, 24
+; LMULMAX1-RV32-NEXT: or a2, a3, a2
+; LMULMAX1-RV32-NEXT: sh a2, 2(sp)
+; LMULMAX1-RV32-NEXT: vsetivli a2, 8, e16,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vle16.v v25, (sp)
+; LMULMAX1-RV32-NEXT: addi a2, sp, 16
+; LMULMAX1-RV32-NEXT: vle16.v v26, (a2)
+; LMULMAX1-RV32-NEXT: vse16.v v25, (a0)
+; LMULMAX1-RV32-NEXT: vse16.v v26, (a1)
+; LMULMAX1-RV32-NEXT: addi sp, sp, 32
+; LMULMAX1-RV32-NEXT: ret
+;
+; LMULMAX1-RV64-LABEL: bswap_v16i16:
+; LMULMAX1-RV64: # %bb.0:
+; LMULMAX1-RV64-NEXT: addi sp, sp, -32
+; LMULMAX1-RV64-NEXT: .cfi_def_cfa_offset 32
+; LMULMAX1-RV64-NEXT: vsetivli a1, 8, e16,m1,ta,mu
+; LMULMAX1-RV64-NEXT: addi a1, a0, 16
+; LMULMAX1-RV64-NEXT: vle16.v v26, (a1)
+; LMULMAX1-RV64-NEXT: vle16.v v25, (a0)
+; LMULMAX1-RV64-NEXT: vmv.x.s a2, v26
+; LMULMAX1-RV64-NEXT: slli a3, a2, 8
+; LMULMAX1-RV64-NEXT: slliw a2, a2, 16
+; LMULMAX1-RV64-NEXT: srliw a2, a2, 24
+; LMULMAX1-RV64-NEXT: or a2, a3, a2
+; LMULMAX1-RV64-NEXT: sh a2, 16(sp)
+; LMULMAX1-RV64-NEXT: vsetivli a2, 1, e16,m1,ta,mu
+; LMULMAX1-RV64-NEXT: vslidedown.vi v27, v26, 7
+; LMULMAX1-RV64-NEXT: vmv.x.s a2, v27
+; LMULMAX1-RV64-NEXT: slli a3, a2, 8
+; LMULMAX1-RV64-NEXT: slliw a2, a2, 16
+; LMULMAX1-RV64-NEXT: srliw a2, a2, 24
+; LMULMAX1-RV64-NEXT: or a2, a3, a2
+; LMULMAX1-RV64-NEXT: sh a2, 30(sp)
+; LMULMAX1-RV64-NEXT: vslidedown.vi v27, v26, 6
+; LMULMAX1-RV64-NEXT: vmv.x.s a2, v27
+; LMULMAX1-RV64-NEXT: slli a3, a2, 8
+; LMULMAX1-RV64-NEXT: slliw a2, a2, 16
+; LMULMAX1-RV64-NEXT: srliw a2, a2, 24
+; LMULMAX1-RV64-NEXT: or a2, a3, a2
+; LMULMAX1-RV64-NEXT: sh a2, 28(sp)
+; LMULMAX1-RV64-NEXT: vslidedown.vi v27, v26, 5
+; LMULMAX1-RV64-NEXT: vmv.x.s a2, v27
+; LMULMAX1-RV64-NEXT: slli a3, a2, 8
+; LMULMAX1-RV64-NEXT: slliw a2, a2, 16
+; LMULMAX1-RV64-NEXT: srliw a2, a2, 24
+; LMULMAX1-RV64-NEXT: or a2, a3, a2
+; LMULMAX1-RV64-NEXT: sh a2, 26(sp)
+; LMULMAX1-RV64-NEXT: vslidedown.vi v27, v26, 4
+; LMULMAX1-RV64-NEXT: vmv.x.s a2, v27
+; LMULMAX1-RV64-NEXT: slli a3, a2, 8
+; LMULMAX1-RV64-NEXT: slliw a2, a2, 16
+; LMULMAX1-RV64-NEXT: srliw a2, a2, 24
+; LMULMAX1-RV64-NEXT: or a2, a3, a2
+; LMULMAX1-RV64-NEXT: sh a2, 24(sp)
+; LMULMAX1-RV64-NEXT: vslidedown.vi v27, v26, 3
+; LMULMAX1-RV64-NEXT: vmv.x.s a2, v27
+; LMULMAX1-RV64-NEXT: slli a3, a2, 8
+; LMULMAX1-RV64-NEXT: slliw a2, a2, 16
+; LMULMAX1-RV64-NEXT: srliw a2, a2, 24
+; LMULMAX1-RV64-NEXT: or a2, a3, a2
+; LMULMAX1-RV64-NEXT: sh a2, 22(sp)
+; LMULMAX1-RV64-NEXT: vslidedown.vi v27, v26, 2
+; LMULMAX1-RV64-NEXT: vmv.x.s a2, v27
+; LMULMAX1-RV64-NEXT: slli a3, a2, 8
+; LMULMAX1-RV64-NEXT: slliw a2, a2, 16
+; LMULMAX1-RV64-NEXT: srliw a2, a2, 24
+; LMULMAX1-RV64-NEXT: or a2, a3, a2
+; LMULMAX1-RV64-NEXT: sh a2, 20(sp)
+; LMULMAX1-RV64-NEXT: vslidedown.vi v26, v26, 1
+; LMULMAX1-RV64-NEXT: vmv.x.s a2, v26
+; LMULMAX1-RV64-NEXT: slli a3, a2, 8
+; LMULMAX1-RV64-NEXT: slliw a2, a2, 16
+; LMULMAX1-RV64-NEXT: srliw a2, a2, 24
+; LMULMAX1-RV64-NEXT: or a2, a3, a2
+; LMULMAX1-RV64-NEXT: sh a2, 18(sp)
+; LMULMAX1-RV64-NEXT: vmv.x.s a2, v25
+; LMULMAX1-RV64-NEXT: slli a3, a2, 8
+; LMULMAX1-RV64-NEXT: slliw a2, a2, 16
+; LMULMAX1-RV64-NEXT: srliw a2, a2, 24
+; LMULMAX1-RV64-NEXT: or a2, a3, a2
+; LMULMAX1-RV64-NEXT: sh a2, 0(sp)
+; LMULMAX1-RV64-NEXT: vslidedown.vi v26, v25, 7
+; LMULMAX1-RV64-NEXT: vmv.x.s a2, v26
+; LMULMAX1-RV64-NEXT: slli a3, a2, 8
+; LMULMAX1-RV64-NEXT: slliw a2, a2, 16
+; LMULMAX1-RV64-NEXT: srliw a2, a2, 24
+; LMULMAX1-RV64-NEXT: or a2, a3, a2
+; LMULMAX1-RV64-NEXT: sh a2, 14(sp)
+; LMULMAX1-RV64-NEXT: vslidedown.vi v26, v25, 6
+; LMULMAX1-RV64-NEXT: vmv.x.s a2, v26
+; LMULMAX1-RV64-NEXT: slli a3, a2, 8
+; LMULMAX1-RV64-NEXT: slliw a2, a2, 16
+; LMULMAX1-RV64-NEXT: srliw a2, a2, 24
+; LMULMAX1-RV64-NEXT: or a2, a3, a2
+; LMULMAX1-RV64-NEXT: sh a2, 12(sp)
+; LMULMAX1-RV64-NEXT: vslidedown.vi v26, v25, 5
+; LMULMAX1-RV64-NEXT: vmv.x.s a2, v26
+; LMULMAX1-RV64-NEXT: slli a3, a2, 8
+; LMULMAX1-RV64-NEXT: slliw a2, a2, 16
+; LMULMAX1-RV64-NEXT: srliw a2, a2, 24
+; LMULMAX1-RV64-NEXT: or a2, a3, a2
+; LMULMAX1-RV64-NEXT: sh a2, 10(sp)
+; LMULMAX1-RV64-NEXT: vslidedown.vi v26, v25, 4
+; LMULMAX1-RV64-NEXT: vmv.x.s a2, v26
+; LMULMAX1-RV64-NEXT: slli a3, a2, 8
+; LMULMAX1-RV64-NEXT: slliw a2, a2, 16
+; LMULMAX1-RV64-NEXT: srliw a2, a2, 24
+; LMULMAX1-RV64-NEXT: or a2, a3, a2
+; LMULMAX1-RV64-NEXT: sh a2, 8(sp)
+; LMULMAX1-RV64-NEXT: vslidedown.vi v26, v25, 3
+; LMULMAX1-RV64-NEXT: vmv.x.s a2, v26
+; LMULMAX1-RV64-NEXT: slli a3, a2, 8
+; LMULMAX1-RV64-NEXT: slliw a2, a2, 16
+; LMULMAX1-RV64-NEXT: srliw a2, a2, 24
+; LMULMAX1-RV64-NEXT: or a2, a3, a2
+; LMULMAX1-RV64-NEXT: sh a2, 6(sp)
+; LMULMAX1-RV64-NEXT: vslidedown.vi v26, v25, 2
+; LMULMAX1-RV64-NEXT: vmv.x.s a2, v26
+; LMULMAX1-RV64-NEXT: slli a3, a2, 8
+; LMULMAX1-RV64-NEXT: slliw a2, a2, 16
+; LMULMAX1-RV64-NEXT: srliw a2, a2, 24
+; LMULMAX1-RV64-NEXT: or a2, a3, a2
+; LMULMAX1-RV64-NEXT: sh a2, 4(sp)
+; LMULMAX1-RV64-NEXT: vslidedown.vi v25, v25, 1
+; LMULMAX1-RV64-NEXT: vmv.x.s a2, v25
+; LMULMAX1-RV64-NEXT: slli a3, a2, 8
+; LMULMAX1-RV64-NEXT: slliw a2, a2, 16
+; LMULMAX1-RV64-NEXT: srliw a2, a2, 24
+; LMULMAX1-RV64-NEXT: or a2, a3, a2
+; LMULMAX1-RV64-NEXT: sh a2, 2(sp)
+; LMULMAX1-RV64-NEXT: vsetivli a2, 8, e16,m1,ta,mu
+; LMULMAX1-RV64-NEXT: vle16.v v25, (sp)
+; LMULMAX1-RV64-NEXT: addi a2, sp, 16
+; LMULMAX1-RV64-NEXT: vle16.v v26, (a2)
+; LMULMAX1-RV64-NEXT: vse16.v v25, (a0)
+; LMULMAX1-RV64-NEXT: vse16.v v26, (a1)
+; LMULMAX1-RV64-NEXT: addi sp, sp, 32
+; LMULMAX1-RV64-NEXT: ret
%a = load <16 x i16>, <16 x i16>* %x
%b = load <16 x i16>, <16 x i16>* %y
%c = call <16 x i16> @llvm.bswap.v16i16(<16 x i16> %a)
@@ -592,107 +1364,107 @@ define void @bswap_v8i32(<8 x i32>* %x, <8 x i32>* %y) {
; LMULMAX2-RV32-NEXT: andi sp, sp, -32
; LMULMAX2-RV32-NEXT: vsetivli a1, 8, e32,m2,ta,mu
; LMULMAX2-RV32-NEXT: vle32.v v26, (a0)
-; LMULMAX2-RV32-NEXT: vsetvli zero, zero, e8,m2,ta,mu
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v26
-; LMULMAX2-RV32-NEXT: sb a1, 3(sp)
-; LMULMAX2-RV32-NEXT: vsetivli a1, 1, e8,m2,ta,mu
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 28
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 31(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 29
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 30(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 30
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 29(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 31
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 28(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 24
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 27(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 25
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 26(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 26
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 25(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 27
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 24(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 20
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 23(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 21
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 22(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 22
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 21(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 23
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 20(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 16
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 19(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 17
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 18(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 18
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 17(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 19
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 16(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 12
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 15(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 13
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 14(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 14
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 13(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 15
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 12(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 8
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 11(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 9
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 10(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 10
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 9(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 11
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 8(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 4
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 7(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 5
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 6(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 6
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 5(sp)
+; LMULMAX2-RV32-NEXT: vmv.x.s a3, v26
+; LMULMAX2-RV32-NEXT: srli a2, a3, 8
+; LMULMAX2-RV32-NEXT: lui a1, 16
+; LMULMAX2-RV32-NEXT: addi a1, a1, -256
+; LMULMAX2-RV32-NEXT: and a2, a2, a1
+; LMULMAX2-RV32-NEXT: srli a4, a3, 24
+; LMULMAX2-RV32-NEXT: or a4, a2, a4
+; LMULMAX2-RV32-NEXT: slli a5, a3, 8
+; LMULMAX2-RV32-NEXT: lui a2, 4080
+; LMULMAX2-RV32-NEXT: and a5, a5, a2
+; LMULMAX2-RV32-NEXT: slli a3, a3, 24
+; LMULMAX2-RV32-NEXT: or a3, a3, a5
+; LMULMAX2-RV32-NEXT: or a3, a3, a4
+; LMULMAX2-RV32-NEXT: sw a3, 0(sp)
+; LMULMAX2-RV32-NEXT: vsetivli a3, 1, e32,m2,ta,mu
; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 7
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 4(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 1
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 2(sp)
+; LMULMAX2-RV32-NEXT: vmv.x.s a3, v28
+; LMULMAX2-RV32-NEXT: srli a4, a3, 8
+; LMULMAX2-RV32-NEXT: and a4, a4, a1
+; LMULMAX2-RV32-NEXT: srli a5, a3, 24
+; LMULMAX2-RV32-NEXT: or a4, a4, a5
+; LMULMAX2-RV32-NEXT: slli a5, a3, 8
+; LMULMAX2-RV32-NEXT: and a5, a5, a2
+; LMULMAX2-RV32-NEXT: slli a3, a3, 24
+; LMULMAX2-RV32-NEXT: or a3, a3, a5
+; LMULMAX2-RV32-NEXT: or a3, a3, a4
+; LMULMAX2-RV32-NEXT: sw a3, 28(sp)
+; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 6
+; LMULMAX2-RV32-NEXT: vmv.x.s a3, v28
+; LMULMAX2-RV32-NEXT: srli a4, a3, 8
+; LMULMAX2-RV32-NEXT: and a4, a4, a1
+; LMULMAX2-RV32-NEXT: srli a5, a3, 24
+; LMULMAX2-RV32-NEXT: or a4, a4, a5
+; LMULMAX2-RV32-NEXT: slli a5, a3, 8
+; LMULMAX2-RV32-NEXT: and a5, a5, a2
+; LMULMAX2-RV32-NEXT: slli a3, a3, 24
+; LMULMAX2-RV32-NEXT: or a3, a3, a5
+; LMULMAX2-RV32-NEXT: or a3, a3, a4
+; LMULMAX2-RV32-NEXT: sw a3, 24(sp)
+; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 5
+; LMULMAX2-RV32-NEXT: vmv.x.s a3, v28
+; LMULMAX2-RV32-NEXT: srli a4, a3, 8
+; LMULMAX2-RV32-NEXT: and a4, a4, a1
+; LMULMAX2-RV32-NEXT: srli a5, a3, 24
+; LMULMAX2-RV32-NEXT: or a4, a4, a5
+; LMULMAX2-RV32-NEXT: slli a5, a3, 8
+; LMULMAX2-RV32-NEXT: and a5, a5, a2
+; LMULMAX2-RV32-NEXT: slli a3, a3, 24
+; LMULMAX2-RV32-NEXT: or a3, a3, a5
+; LMULMAX2-RV32-NEXT: or a3, a3, a4
+; LMULMAX2-RV32-NEXT: sw a3, 20(sp)
+; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 4
+; LMULMAX2-RV32-NEXT: vmv.x.s a3, v28
+; LMULMAX2-RV32-NEXT: srli a4, a3, 8
+; LMULMAX2-RV32-NEXT: and a4, a4, a1
+; LMULMAX2-RV32-NEXT: srli a5, a3, 24
+; LMULMAX2-RV32-NEXT: or a4, a4, a5
+; LMULMAX2-RV32-NEXT: slli a5, a3, 8
+; LMULMAX2-RV32-NEXT: and a5, a5, a2
+; LMULMAX2-RV32-NEXT: slli a3, a3, 24
+; LMULMAX2-RV32-NEXT: or a3, a3, a5
+; LMULMAX2-RV32-NEXT: or a3, a3, a4
+; LMULMAX2-RV32-NEXT: sw a3, 16(sp)
+; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 3
+; LMULMAX2-RV32-NEXT: vmv.x.s a3, v28
+; LMULMAX2-RV32-NEXT: srli a4, a3, 8
+; LMULMAX2-RV32-NEXT: and a4, a4, a1
+; LMULMAX2-RV32-NEXT: srli a5, a3, 24
+; LMULMAX2-RV32-NEXT: or a4, a4, a5
+; LMULMAX2-RV32-NEXT: slli a5, a3, 8
+; LMULMAX2-RV32-NEXT: and a5, a5, a2
+; LMULMAX2-RV32-NEXT: slli a3, a3, 24
+; LMULMAX2-RV32-NEXT: or a3, a3, a5
+; LMULMAX2-RV32-NEXT: or a3, a3, a4
+; LMULMAX2-RV32-NEXT: sw a3, 12(sp)
; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 2
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 1(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v26, v26, 3
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v26
-; LMULMAX2-RV32-NEXT: sb a1, 0(sp)
-; LMULMAX2-RV32-NEXT: addi a1, zero, 32
-; LMULMAX2-RV32-NEXT: vsetvli a1, a1, e8,m2,ta,mu
-; LMULMAX2-RV32-NEXT: vle8.v v26, (sp)
+; LMULMAX2-RV32-NEXT: vmv.x.s a3, v28
+; LMULMAX2-RV32-NEXT: srli a4, a3, 8
+; LMULMAX2-RV32-NEXT: and a4, a4, a1
+; LMULMAX2-RV32-NEXT: srli a5, a3, 24
+; LMULMAX2-RV32-NEXT: or a4, a4, a5
+; LMULMAX2-RV32-NEXT: slli a5, a3, 8
+; LMULMAX2-RV32-NEXT: and a5, a5, a2
+; LMULMAX2-RV32-NEXT: slli a3, a3, 24
+; LMULMAX2-RV32-NEXT: or a3, a3, a5
+; LMULMAX2-RV32-NEXT: or a3, a3, a4
+; LMULMAX2-RV32-NEXT: sw a3, 8(sp)
+; LMULMAX2-RV32-NEXT: vslidedown.vi v26, v26, 1
+; LMULMAX2-RV32-NEXT: vmv.x.s a3, v26
+; LMULMAX2-RV32-NEXT: srli a4, a3, 8
+; LMULMAX2-RV32-NEXT: and a1, a4, a1
+; LMULMAX2-RV32-NEXT: srli a4, a3, 24
+; LMULMAX2-RV32-NEXT: or a1, a1, a4
+; LMULMAX2-RV32-NEXT: slli a4, a3, 8
+; LMULMAX2-RV32-NEXT: and a2, a4, a2
+; LMULMAX2-RV32-NEXT: slli a3, a3, 24
+; LMULMAX2-RV32-NEXT: or a2, a3, a2
+; LMULMAX2-RV32-NEXT: or a1, a2, a1
+; LMULMAX2-RV32-NEXT: sw a1, 4(sp)
; LMULMAX2-RV32-NEXT: vsetivli a1, 8, e32,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vle32.v v26, (sp)
; LMULMAX2-RV32-NEXT: vse32.v v26, (a0)
; LMULMAX2-RV32-NEXT: addi sp, s0, -64
; LMULMAX2-RV32-NEXT: lw s0, 56(sp) # 4-byte Folded Reload
@@ -713,107 +1485,107 @@ define void @bswap_v8i32(<8 x i32>* %x, <8 x i32>* %y) {
; LMULMAX2-RV64-NEXT: andi sp, sp, -32
; LMULMAX2-RV64-NEXT: vsetivli a1, 8, e32,m2,ta,mu
; LMULMAX2-RV64-NEXT: vle32.v v26, (a0)
-; LMULMAX2-RV64-NEXT: vsetvli zero, zero, e8,m2,ta,mu
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v26
-; LMULMAX2-RV64-NEXT: sb a1, 3(sp)
-; LMULMAX2-RV64-NEXT: vsetivli a1, 1, e8,m2,ta,mu
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 28
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 31(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 29
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 30(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 30
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 29(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 31
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 28(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 24
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 27(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 25
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 26(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 26
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 25(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 27
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 24(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 20
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 23(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 21
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 22(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 22
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 21(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 23
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 20(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 16
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 19(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 17
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 18(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 18
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 17(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 19
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 16(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 12
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 15(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 13
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 14(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 14
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 13(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 15
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 12(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 8
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 11(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 9
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 10(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 10
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 9(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 11
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 8(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 4
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 7(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 5
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 6(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 6
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 5(sp)
+; LMULMAX2-RV64-NEXT: vmv.x.s a3, v26
+; LMULMAX2-RV64-NEXT: srliw a2, a3, 8
+; LMULMAX2-RV64-NEXT: lui a1, 16
+; LMULMAX2-RV64-NEXT: addiw a1, a1, -256
+; LMULMAX2-RV64-NEXT: and a2, a2, a1
+; LMULMAX2-RV64-NEXT: srliw a4, a3, 24
+; LMULMAX2-RV64-NEXT: or a4, a2, a4
+; LMULMAX2-RV64-NEXT: slli a5, a3, 8
+; LMULMAX2-RV64-NEXT: lui a2, 4080
+; LMULMAX2-RV64-NEXT: and a5, a5, a2
+; LMULMAX2-RV64-NEXT: slli a3, a3, 24
+; LMULMAX2-RV64-NEXT: or a3, a3, a5
+; LMULMAX2-RV64-NEXT: or a3, a3, a4
+; LMULMAX2-RV64-NEXT: sw a3, 0(sp)
+; LMULMAX2-RV64-NEXT: vsetivli a3, 1, e32,m2,ta,mu
; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 7
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 4(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 1
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 2(sp)
+; LMULMAX2-RV64-NEXT: vmv.x.s a3, v28
+; LMULMAX2-RV64-NEXT: srliw a4, a3, 8
+; LMULMAX2-RV64-NEXT: and a4, a4, a1
+; LMULMAX2-RV64-NEXT: srliw a5, a3, 24
+; LMULMAX2-RV64-NEXT: or a4, a4, a5
+; LMULMAX2-RV64-NEXT: slli a5, a3, 8
+; LMULMAX2-RV64-NEXT: and a5, a5, a2
+; LMULMAX2-RV64-NEXT: slli a3, a3, 24
+; LMULMAX2-RV64-NEXT: or a3, a3, a5
+; LMULMAX2-RV64-NEXT: or a3, a3, a4
+; LMULMAX2-RV64-NEXT: sw a3, 28(sp)
+; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 6
+; LMULMAX2-RV64-NEXT: vmv.x.s a3, v28
+; LMULMAX2-RV64-NEXT: srliw a4, a3, 8
+; LMULMAX2-RV64-NEXT: and a4, a4, a1
+; LMULMAX2-RV64-NEXT: srliw a5, a3, 24
+; LMULMAX2-RV64-NEXT: or a4, a4, a5
+; LMULMAX2-RV64-NEXT: slli a5, a3, 8
+; LMULMAX2-RV64-NEXT: and a5, a5, a2
+; LMULMAX2-RV64-NEXT: slli a3, a3, 24
+; LMULMAX2-RV64-NEXT: or a3, a3, a5
+; LMULMAX2-RV64-NEXT: or a3, a3, a4
+; LMULMAX2-RV64-NEXT: sw a3, 24(sp)
+; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 5
+; LMULMAX2-RV64-NEXT: vmv.x.s a3, v28
+; LMULMAX2-RV64-NEXT: srliw a4, a3, 8
+; LMULMAX2-RV64-NEXT: and a4, a4, a1
+; LMULMAX2-RV64-NEXT: srliw a5, a3, 24
+; LMULMAX2-RV64-NEXT: or a4, a4, a5
+; LMULMAX2-RV64-NEXT: slli a5, a3, 8
+; LMULMAX2-RV64-NEXT: and a5, a5, a2
+; LMULMAX2-RV64-NEXT: slli a3, a3, 24
+; LMULMAX2-RV64-NEXT: or a3, a3, a5
+; LMULMAX2-RV64-NEXT: or a3, a3, a4
+; LMULMAX2-RV64-NEXT: sw a3, 20(sp)
+; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 4
+; LMULMAX2-RV64-NEXT: vmv.x.s a3, v28
+; LMULMAX2-RV64-NEXT: srliw a4, a3, 8
+; LMULMAX2-RV64-NEXT: and a4, a4, a1
+; LMULMAX2-RV64-NEXT: srliw a5, a3, 24
+; LMULMAX2-RV64-NEXT: or a4, a4, a5
+; LMULMAX2-RV64-NEXT: slli a5, a3, 8
+; LMULMAX2-RV64-NEXT: and a5, a5, a2
+; LMULMAX2-RV64-NEXT: slli a3, a3, 24
+; LMULMAX2-RV64-NEXT: or a3, a3, a5
+; LMULMAX2-RV64-NEXT: or a3, a3, a4
+; LMULMAX2-RV64-NEXT: sw a3, 16(sp)
+; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 3
+; LMULMAX2-RV64-NEXT: vmv.x.s a3, v28
+; LMULMAX2-RV64-NEXT: srliw a4, a3, 8
+; LMULMAX2-RV64-NEXT: and a4, a4, a1
+; LMULMAX2-RV64-NEXT: srliw a5, a3, 24
+; LMULMAX2-RV64-NEXT: or a4, a4, a5
+; LMULMAX2-RV64-NEXT: slli a5, a3, 8
+; LMULMAX2-RV64-NEXT: and a5, a5, a2
+; LMULMAX2-RV64-NEXT: slli a3, a3, 24
+; LMULMAX2-RV64-NEXT: or a3, a3, a5
+; LMULMAX2-RV64-NEXT: or a3, a3, a4
+; LMULMAX2-RV64-NEXT: sw a3, 12(sp)
; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 2
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 1(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v26, v26, 3
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v26
-; LMULMAX2-RV64-NEXT: sb a1, 0(sp)
-; LMULMAX2-RV64-NEXT: addi a1, zero, 32
-; LMULMAX2-RV64-NEXT: vsetvli a1, a1, e8,m2,ta,mu
-; LMULMAX2-RV64-NEXT: vle8.v v26, (sp)
+; LMULMAX2-RV64-NEXT: vmv.x.s a3, v28
+; LMULMAX2-RV64-NEXT: srliw a4, a3, 8
+; LMULMAX2-RV64-NEXT: and a4, a4, a1
+; LMULMAX2-RV64-NEXT: srliw a5, a3, 24
+; LMULMAX2-RV64-NEXT: or a4, a4, a5
+; LMULMAX2-RV64-NEXT: slli a5, a3, 8
+; LMULMAX2-RV64-NEXT: and a5, a5, a2
+; LMULMAX2-RV64-NEXT: slli a3, a3, 24
+; LMULMAX2-RV64-NEXT: or a3, a3, a5
+; LMULMAX2-RV64-NEXT: or a3, a3, a4
+; LMULMAX2-RV64-NEXT: sw a3, 8(sp)
+; LMULMAX2-RV64-NEXT: vslidedown.vi v26, v26, 1
+; LMULMAX2-RV64-NEXT: vmv.x.s a3, v26
+; LMULMAX2-RV64-NEXT: srliw a4, a3, 8
+; LMULMAX2-RV64-NEXT: and a1, a4, a1
+; LMULMAX2-RV64-NEXT: srliw a4, a3, 24
+; LMULMAX2-RV64-NEXT: or a1, a1, a4
+; LMULMAX2-RV64-NEXT: slli a4, a3, 8
+; LMULMAX2-RV64-NEXT: and a2, a4, a2
+; LMULMAX2-RV64-NEXT: slli a3, a3, 24
+; LMULMAX2-RV64-NEXT: or a2, a3, a2
+; LMULMAX2-RV64-NEXT: or a1, a2, a1
+; LMULMAX2-RV64-NEXT: sw a1, 4(sp)
; LMULMAX2-RV64-NEXT: vsetivli a1, 8, e32,m2,ta,mu
+; LMULMAX2-RV64-NEXT: vle32.v v26, (sp)
; LMULMAX2-RV64-NEXT: vse32.v v26, (a0)
; LMULMAX2-RV64-NEXT: addi sp, s0, -64
; LMULMAX2-RV64-NEXT: ld s0, 48(sp) # 8-byte Folded Reload
@@ -821,119 +1593,235 @@ define void @bswap_v8i32(<8 x i32>* %x, <8 x i32>* %y) {
; LMULMAX2-RV64-NEXT: addi sp, sp, 64
; LMULMAX2-RV64-NEXT: ret
;
-; LMULMAX1-LABEL: bswap_v8i32:
-; LMULMAX1: # %bb.0:
-; LMULMAX1-NEXT: addi sp, sp, -32
-; LMULMAX1-NEXT: .cfi_def_cfa_offset 32
-; LMULMAX1-NEXT: vsetivli a1, 4, e32,m1,ta,mu
-; LMULMAX1-NEXT: addi a1, a0, 16
-; LMULMAX1-NEXT: vle32.v v26, (a1)
-; LMULMAX1-NEXT: vle32.v v25, (a0)
-; LMULMAX1-NEXT: vsetvli zero, zero, e8,m1,ta,mu
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 3(sp)
-; LMULMAX1-NEXT: vsetivli a2, 1, e8,m1,ta,mu
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 12
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 15(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 13
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 14(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 14
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 13(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 15
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 12(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 8
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 11(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 9
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 10(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 10
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 9(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 11
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 8(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 4
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 7(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 5
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 6(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 6
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 5(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 7
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 4(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 1
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 2(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 2
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 1(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v26, 3
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 0(sp)
-; LMULMAX1-NEXT: vmv.x.s a2, v25
-; LMULMAX1-NEXT: sb a2, 19(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 12
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 31(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 13
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 30(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 14
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 29(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 15
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 28(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 8
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 27(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 9
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 26(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 10
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 25(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 11
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 24(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 4
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 23(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 5
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 22(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 6
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 21(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 7
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 20(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 1
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 18(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 2
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 17(sp)
-; LMULMAX1-NEXT: vslidedown.vi v25, v25, 3
-; LMULMAX1-NEXT: vmv.x.s a2, v25
-; LMULMAX1-NEXT: sb a2, 16(sp)
-; LMULMAX1-NEXT: vsetivli a2, 16, e8,m1,ta,mu
-; LMULMAX1-NEXT: addi a2, sp, 16
-; LMULMAX1-NEXT: vle8.v v25, (a2)
-; LMULMAX1-NEXT: vle8.v v26, (sp)
-; LMULMAX1-NEXT: vsetivli a2, 4, e32,m1,ta,mu
-; LMULMAX1-NEXT: vse32.v v25, (a0)
-; LMULMAX1-NEXT: vse32.v v26, (a1)
-; LMULMAX1-NEXT: addi sp, sp, 32
-; LMULMAX1-NEXT: ret
+; LMULMAX1-RV32-LABEL: bswap_v8i32:
+; LMULMAX1-RV32: # %bb.0:
+; LMULMAX1-RV32-NEXT: addi sp, sp, -32
+; LMULMAX1-RV32-NEXT: .cfi_def_cfa_offset 32
+; LMULMAX1-RV32-NEXT: vsetivli a1, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: addi a6, a0, 16
+; LMULMAX1-RV32-NEXT: vle32.v v26, (a6)
+; LMULMAX1-RV32-NEXT: vle32.v v25, (a0)
+; LMULMAX1-RV32-NEXT: vmv.x.s a4, v26
+; LMULMAX1-RV32-NEXT: srli a3, a4, 8
+; LMULMAX1-RV32-NEXT: lui a2, 16
+; LMULMAX1-RV32-NEXT: addi a2, a2, -256
+; LMULMAX1-RV32-NEXT: and a3, a3, a2
+; LMULMAX1-RV32-NEXT: srli a5, a4, 24
+; LMULMAX1-RV32-NEXT: or a5, a3, a5
+; LMULMAX1-RV32-NEXT: slli a1, a4, 8
+; LMULMAX1-RV32-NEXT: lui a3, 4080
+; LMULMAX1-RV32-NEXT: and a1, a1, a3
+; LMULMAX1-RV32-NEXT: slli a4, a4, 24
+; LMULMAX1-RV32-NEXT: or a1, a4, a1
+; LMULMAX1-RV32-NEXT: or a1, a1, a5
+; LMULMAX1-RV32-NEXT: sw a1, 16(sp)
+; LMULMAX1-RV32-NEXT: vsetivli a1, 1, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vslidedown.vi v27, v26, 3
+; LMULMAX1-RV32-NEXT: vmv.x.s a1, v27
+; LMULMAX1-RV32-NEXT: srli a4, a1, 8
+; LMULMAX1-RV32-NEXT: and a4, a4, a2
+; LMULMAX1-RV32-NEXT: srli a5, a1, 24
+; LMULMAX1-RV32-NEXT: or a4, a4, a5
+; LMULMAX1-RV32-NEXT: slli a5, a1, 8
+; LMULMAX1-RV32-NEXT: and a5, a5, a3
+; LMULMAX1-RV32-NEXT: slli a1, a1, 24
+; LMULMAX1-RV32-NEXT: or a1, a1, a5
+; LMULMAX1-RV32-NEXT: or a1, a1, a4
+; LMULMAX1-RV32-NEXT: sw a1, 28(sp)
+; LMULMAX1-RV32-NEXT: vslidedown.vi v27, v26, 2
+; LMULMAX1-RV32-NEXT: vmv.x.s a1, v27
+; LMULMAX1-RV32-NEXT: srli a4, a1, 8
+; LMULMAX1-RV32-NEXT: and a4, a4, a2
+; LMULMAX1-RV32-NEXT: srli a5, a1, 24
+; LMULMAX1-RV32-NEXT: or a4, a4, a5
+; LMULMAX1-RV32-NEXT: slli a5, a1, 8
+; LMULMAX1-RV32-NEXT: and a5, a5, a3
+; LMULMAX1-RV32-NEXT: slli a1, a1, 24
+; LMULMAX1-RV32-NEXT: or a1, a1, a5
+; LMULMAX1-RV32-NEXT: or a1, a1, a4
+; LMULMAX1-RV32-NEXT: sw a1, 24(sp)
+; LMULMAX1-RV32-NEXT: vslidedown.vi v26, v26, 1
+; LMULMAX1-RV32-NEXT: vmv.x.s a1, v26
+; LMULMAX1-RV32-NEXT: srli a4, a1, 8
+; LMULMAX1-RV32-NEXT: and a4, a4, a2
+; LMULMAX1-RV32-NEXT: srli a5, a1, 24
+; LMULMAX1-RV32-NEXT: or a4, a4, a5
+; LMULMAX1-RV32-NEXT: slli a5, a1, 8
+; LMULMAX1-RV32-NEXT: and a5, a5, a3
+; LMULMAX1-RV32-NEXT: slli a1, a1, 24
+; LMULMAX1-RV32-NEXT: or a1, a1, a5
+; LMULMAX1-RV32-NEXT: or a1, a1, a4
+; LMULMAX1-RV32-NEXT: sw a1, 20(sp)
+; LMULMAX1-RV32-NEXT: vmv.x.s a1, v25
+; LMULMAX1-RV32-NEXT: srli a4, a1, 8
+; LMULMAX1-RV32-NEXT: and a4, a4, a2
+; LMULMAX1-RV32-NEXT: srli a5, a1, 24
+; LMULMAX1-RV32-NEXT: or a4, a4, a5
+; LMULMAX1-RV32-NEXT: slli a5, a1, 8
+; LMULMAX1-RV32-NEXT: and a5, a5, a3
+; LMULMAX1-RV32-NEXT: slli a1, a1, 24
+; LMULMAX1-RV32-NEXT: or a1, a1, a5
+; LMULMAX1-RV32-NEXT: or a1, a1, a4
+; LMULMAX1-RV32-NEXT: sw a1, 0(sp)
+; LMULMAX1-RV32-NEXT: vslidedown.vi v26, v25, 3
+; LMULMAX1-RV32-NEXT: vmv.x.s a1, v26
+; LMULMAX1-RV32-NEXT: srli a4, a1, 8
+; LMULMAX1-RV32-NEXT: and a4, a4, a2
+; LMULMAX1-RV32-NEXT: srli a5, a1, 24
+; LMULMAX1-RV32-NEXT: or a4, a4, a5
+; LMULMAX1-RV32-NEXT: slli a5, a1, 8
+; LMULMAX1-RV32-NEXT: and a5, a5, a3
+; LMULMAX1-RV32-NEXT: slli a1, a1, 24
+; LMULMAX1-RV32-NEXT: or a1, a1, a5
+; LMULMAX1-RV32-NEXT: or a1, a1, a4
+; LMULMAX1-RV32-NEXT: sw a1, 12(sp)
+; LMULMAX1-RV32-NEXT: vslidedown.vi v26, v25, 2
+; LMULMAX1-RV32-NEXT: vmv.x.s a1, v26
+; LMULMAX1-RV32-NEXT: srli a4, a1, 8
+; LMULMAX1-RV32-NEXT: and a4, a4, a2
+; LMULMAX1-RV32-NEXT: srli a5, a1, 24
+; LMULMAX1-RV32-NEXT: or a4, a4, a5
+; LMULMAX1-RV32-NEXT: slli a5, a1, 8
+; LMULMAX1-RV32-NEXT: and a5, a5, a3
+; LMULMAX1-RV32-NEXT: slli a1, a1, 24
+; LMULMAX1-RV32-NEXT: or a1, a1, a5
+; LMULMAX1-RV32-NEXT: or a1, a1, a4
+; LMULMAX1-RV32-NEXT: sw a1, 8(sp)
+; LMULMAX1-RV32-NEXT: vslidedown.vi v25, v25, 1
+; LMULMAX1-RV32-NEXT: vmv.x.s a1, v25
+; LMULMAX1-RV32-NEXT: srli a4, a1, 8
+; LMULMAX1-RV32-NEXT: and a2, a4, a2
+; LMULMAX1-RV32-NEXT: srli a4, a1, 24
+; LMULMAX1-RV32-NEXT: or a2, a2, a4
+; LMULMAX1-RV32-NEXT: slli a4, a1, 8
+; LMULMAX1-RV32-NEXT: and a3, a4, a3
+; LMULMAX1-RV32-NEXT: slli a1, a1, 24
+; LMULMAX1-RV32-NEXT: or a1, a1, a3
+; LMULMAX1-RV32-NEXT: or a1, a1, a2
+; LMULMAX1-RV32-NEXT: sw a1, 4(sp)
+; LMULMAX1-RV32-NEXT: vsetivli a1, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vle32.v v25, (sp)
+; LMULMAX1-RV32-NEXT: addi a1, sp, 16
+; LMULMAX1-RV32-NEXT: vle32.v v26, (a1)
+; LMULMAX1-RV32-NEXT: vse32.v v25, (a0)
+; LMULMAX1-RV32-NEXT: vse32.v v26, (a6)
+; LMULMAX1-RV32-NEXT: addi sp, sp, 32
+; LMULMAX1-RV32-NEXT: ret
+;
+; LMULMAX1-RV64-LABEL: bswap_v8i32:
+; LMULMAX1-RV64: # %bb.0:
+; LMULMAX1-RV64-NEXT: addi sp, sp, -32
+; LMULMAX1-RV64-NEXT: .cfi_def_cfa_offset 32
+; LMULMAX1-RV64-NEXT: vsetivli a1, 4, e32,m1,ta,mu
+; LMULMAX1-RV64-NEXT: addi a6, a0, 16
+; LMULMAX1-RV64-NEXT: vle32.v v26, (a6)
+; LMULMAX1-RV64-NEXT: vle32.v v25, (a0)
+; LMULMAX1-RV64-NEXT: vmv.x.s a4, v26
+; LMULMAX1-RV64-NEXT: srliw a3, a4, 8
+; LMULMAX1-RV64-NEXT: lui a2, 16
+; LMULMAX1-RV64-NEXT: addiw a2, a2, -256
+; LMULMAX1-RV64-NEXT: and a3, a3, a2
+; LMULMAX1-RV64-NEXT: srliw a5, a4, 24
+; LMULMAX1-RV64-NEXT: or a5, a3, a5
+; LMULMAX1-RV64-NEXT: slli a1, a4, 8
+; LMULMAX1-RV64-NEXT: lui a3, 4080
+; LMULMAX1-RV64-NEXT: and a1, a1, a3
+; LMULMAX1-RV64-NEXT: slli a4, a4, 24
+; LMULMAX1-RV64-NEXT: or a1, a4, a1
+; LMULMAX1-RV64-NEXT: or a1, a1, a5
+; LMULMAX1-RV64-NEXT: sw a1, 16(sp)
+; LMULMAX1-RV64-NEXT: vsetivli a1, 1, e32,m1,ta,mu
+; LMULMAX1-RV64-NEXT: vslidedown.vi v27, v26, 3
+; LMULMAX1-RV64-NEXT: vmv.x.s a1, v27
+; LMULMAX1-RV64-NEXT: srliw a4, a1, 8
+; LMULMAX1-RV64-NEXT: and a4, a4, a2
+; LMULMAX1-RV64-NEXT: srliw a5, a1, 24
+; LMULMAX1-RV64-NEXT: or a4, a4, a5
+; LMULMAX1-RV64-NEXT: slli a5, a1, 8
+; LMULMAX1-RV64-NEXT: and a5, a5, a3
+; LMULMAX1-RV64-NEXT: slli a1, a1, 24
+; LMULMAX1-RV64-NEXT: or a1, a1, a5
+; LMULMAX1-RV64-NEXT: or a1, a1, a4
+; LMULMAX1-RV64-NEXT: sw a1, 28(sp)
+; LMULMAX1-RV64-NEXT: vslidedown.vi v27, v26, 2
+; LMULMAX1-RV64-NEXT: vmv.x.s a1, v27
+; LMULMAX1-RV64-NEXT: srliw a4, a1, 8
+; LMULMAX1-RV64-NEXT: and a4, a4, a2
+; LMULMAX1-RV64-NEXT: srliw a5, a1, 24
+; LMULMAX1-RV64-NEXT: or a4, a4, a5
+; LMULMAX1-RV64-NEXT: slli a5, a1, 8
+; LMULMAX1-RV64-NEXT: and a5, a5, a3
+; LMULMAX1-RV64-NEXT: slli a1, a1, 24
+; LMULMAX1-RV64-NEXT: or a1, a1, a5
+; LMULMAX1-RV64-NEXT: or a1, a1, a4
+; LMULMAX1-RV64-NEXT: sw a1, 24(sp)
+; LMULMAX1-RV64-NEXT: vslidedown.vi v26, v26, 1
+; LMULMAX1-RV64-NEXT: vmv.x.s a1, v26
+; LMULMAX1-RV64-NEXT: srliw a4, a1, 8
+; LMULMAX1-RV64-NEXT: and a4, a4, a2
+; LMULMAX1-RV64-NEXT: srliw a5, a1, 24
+; LMULMAX1-RV64-NEXT: or a4, a4, a5
+; LMULMAX1-RV64-NEXT: slli a5, a1, 8
+; LMULMAX1-RV64-NEXT: and a5, a5, a3
+; LMULMAX1-RV64-NEXT: slli a1, a1, 24
+; LMULMAX1-RV64-NEXT: or a1, a1, a5
+; LMULMAX1-RV64-NEXT: or a1, a1, a4
+; LMULMAX1-RV64-NEXT: sw a1, 20(sp)
+; LMULMAX1-RV64-NEXT: vmv.x.s a1, v25
+; LMULMAX1-RV64-NEXT: srliw a4, a1, 8
+; LMULMAX1-RV64-NEXT: and a4, a4, a2
+; LMULMAX1-RV64-NEXT: srliw a5, a1, 24
+; LMULMAX1-RV64-NEXT: or a4, a4, a5
+; LMULMAX1-RV64-NEXT: slli a5, a1, 8
+; LMULMAX1-RV64-NEXT: and a5, a5, a3
+; LMULMAX1-RV64-NEXT: slli a1, a1, 24
+; LMULMAX1-RV64-NEXT: or a1, a1, a5
+; LMULMAX1-RV64-NEXT: or a1, a1, a4
+; LMULMAX1-RV64-NEXT: sw a1, 0(sp)
+; LMULMAX1-RV64-NEXT: vslidedown.vi v26, v25, 3
+; LMULMAX1-RV64-NEXT: vmv.x.s a1, v26
+; LMULMAX1-RV64-NEXT: srliw a4, a1, 8
+; LMULMAX1-RV64-NEXT: and a4, a4, a2
+; LMULMAX1-RV64-NEXT: srliw a5, a1, 24
+; LMULMAX1-RV64-NEXT: or a4, a4, a5
+; LMULMAX1-RV64-NEXT: slli a5, a1, 8
+; LMULMAX1-RV64-NEXT: and a5, a5, a3
+; LMULMAX1-RV64-NEXT: slli a1, a1, 24
+; LMULMAX1-RV64-NEXT: or a1, a1, a5
+; LMULMAX1-RV64-NEXT: or a1, a1, a4
+; LMULMAX1-RV64-NEXT: sw a1, 12(sp)
+; LMULMAX1-RV64-NEXT: vslidedown.vi v26, v25, 2
+; LMULMAX1-RV64-NEXT: vmv.x.s a1, v26
+; LMULMAX1-RV64-NEXT: srliw a4, a1, 8
+; LMULMAX1-RV64-NEXT: and a4, a4, a2
+; LMULMAX1-RV64-NEXT: srliw a5, a1, 24
+; LMULMAX1-RV64-NEXT: or a4, a4, a5
+; LMULMAX1-RV64-NEXT: slli a5, a1, 8
+; LMULMAX1-RV64-NEXT: and a5, a5, a3
+; LMULMAX1-RV64-NEXT: slli a1, a1, 24
+; LMULMAX1-RV64-NEXT: or a1, a1, a5
+; LMULMAX1-RV64-NEXT: or a1, a1, a4
+; LMULMAX1-RV64-NEXT: sw a1, 8(sp)
+; LMULMAX1-RV64-NEXT: vslidedown.vi v25, v25, 1
+; LMULMAX1-RV64-NEXT: vmv.x.s a1, v25
+; LMULMAX1-RV64-NEXT: srliw a4, a1, 8
+; LMULMAX1-RV64-NEXT: and a2, a4, a2
+; LMULMAX1-RV64-NEXT: srliw a4, a1, 24
+; LMULMAX1-RV64-NEXT: or a2, a2, a4
+; LMULMAX1-RV64-NEXT: slli a4, a1, 8
+; LMULMAX1-RV64-NEXT: and a3, a4, a3
+; LMULMAX1-RV64-NEXT: slli a1, a1, 24
+; LMULMAX1-RV64-NEXT: or a1, a1, a3
+; LMULMAX1-RV64-NEXT: or a1, a1, a2
+; LMULMAX1-RV64-NEXT: sw a1, 4(sp)
+; LMULMAX1-RV64-NEXT: vsetivli a1, 4, e32,m1,ta,mu
+; LMULMAX1-RV64-NEXT: vle32.v v25, (sp)
+; LMULMAX1-RV64-NEXT: addi a1, sp, 16
+; LMULMAX1-RV64-NEXT: vle32.v v26, (a1)
+; LMULMAX1-RV64-NEXT: vse32.v v25, (a0)
+; LMULMAX1-RV64-NEXT: vse32.v v26, (a6)
+; LMULMAX1-RV64-NEXT: addi sp, sp, 32
+; LMULMAX1-RV64-NEXT: ret
%a = load <8 x i32>, <8 x i32>* %x
%b = load <8 x i32>, <8 x i32>* %y
%c = call <8 x i32> @llvm.bswap.v8i32(<8 x i32> %a)
@@ -956,106 +1844,108 @@ define void @bswap_v4i64(<4 x i64>* %x, <4 x i64>* %y) {
; LMULMAX2-RV32-NEXT: andi sp, sp, -32
; LMULMAX2-RV32-NEXT: vsetivli a1, 4, e64,m2,ta,mu
; LMULMAX2-RV32-NEXT: vle64.v v26, (a0)
-; LMULMAX2-RV32-NEXT: vsetvli zero, zero, e8,m2,ta,mu
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v26
-; LMULMAX2-RV32-NEXT: sb a1, 7(sp)
-; LMULMAX2-RV32-NEXT: vsetivli a1, 1, e8,m2,ta,mu
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 24
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 31(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 25
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 30(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 26
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 29(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 27
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 28(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 28
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 27(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 29
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 26(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 30
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 25(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 31
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 24(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 16
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 23(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 17
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 22(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 18
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 21(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 19
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 20(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 20
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 19(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 21
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 18(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 22
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 17(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 23
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 16(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 8
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 15(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 9
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 14(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 10
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 13(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 11
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 12(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 12
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 11(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 13
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 10(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 14
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 9(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 15
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 8(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 1
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 6(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 2
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 5(sp)
+; LMULMAX2-RV32-NEXT: vmv.x.s a3, v26
+; LMULMAX2-RV32-NEXT: srli a2, a3, 8
+; LMULMAX2-RV32-NEXT: lui a1, 16
+; LMULMAX2-RV32-NEXT: addi a1, a1, -256
+; LMULMAX2-RV32-NEXT: and a2, a2, a1
+; LMULMAX2-RV32-NEXT: srli a4, a3, 24
+; LMULMAX2-RV32-NEXT: or a4, a2, a4
+; LMULMAX2-RV32-NEXT: slli a5, a3, 8
+; LMULMAX2-RV32-NEXT: lui a6, 4080
+; LMULMAX2-RV32-NEXT: and a5, a5, a6
+; LMULMAX2-RV32-NEXT: slli a3, a3, 24
+; LMULMAX2-RV32-NEXT: or a3, a3, a5
+; LMULMAX2-RV32-NEXT: or a3, a3, a4
+; LMULMAX2-RV32-NEXT: sw a3, 4(sp)
+; LMULMAX2-RV32-NEXT: vsetivli a3, 1, e64,m2,ta,mu
; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 3
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 4(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 4
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 3(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 5
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 2(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v28, v26, 6
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV32-NEXT: sb a1, 1(sp)
-; LMULMAX2-RV32-NEXT: vslidedown.vi v26, v26, 7
-; LMULMAX2-RV32-NEXT: vmv.x.s a1, v26
-; LMULMAX2-RV32-NEXT: sb a1, 0(sp)
-; LMULMAX2-RV32-NEXT: addi a1, zero, 32
-; LMULMAX2-RV32-NEXT: vsetvli a1, a1, e8,m2,ta,mu
-; LMULMAX2-RV32-NEXT: vle8.v v26, (sp)
+; LMULMAX2-RV32-NEXT: vmv.x.s a3, v28
+; LMULMAX2-RV32-NEXT: srli a4, a3, 8
+; LMULMAX2-RV32-NEXT: and a4, a4, a1
+; LMULMAX2-RV32-NEXT: srli a5, a3, 24
+; LMULMAX2-RV32-NEXT: or a4, a4, a5
+; LMULMAX2-RV32-NEXT: slli a5, a3, 8
+; LMULMAX2-RV32-NEXT: and a5, a5, a6
+; LMULMAX2-RV32-NEXT: slli a3, a3, 24
+; LMULMAX2-RV32-NEXT: or a3, a3, a5
+; LMULMAX2-RV32-NEXT: or a3, a3, a4
+; LMULMAX2-RV32-NEXT: sw a3, 28(sp)
+; LMULMAX2-RV32-NEXT: vslidedown.vi v30, v26, 2
+; LMULMAX2-RV32-NEXT: vmv.x.s a3, v30
+; LMULMAX2-RV32-NEXT: srli a4, a3, 8
+; LMULMAX2-RV32-NEXT: and a4, a4, a1
+; LMULMAX2-RV32-NEXT: srli a5, a3, 24
+; LMULMAX2-RV32-NEXT: or a4, a4, a5
+; LMULMAX2-RV32-NEXT: slli a5, a3, 8
+; LMULMAX2-RV32-NEXT: and a5, a5, a6
+; LMULMAX2-RV32-NEXT: slli a3, a3, 24
+; LMULMAX2-RV32-NEXT: or a3, a3, a5
+; LMULMAX2-RV32-NEXT: or a3, a3, a4
+; LMULMAX2-RV32-NEXT: sw a3, 20(sp)
+; LMULMAX2-RV32-NEXT: vslidedown.vi v8, v26, 1
+; LMULMAX2-RV32-NEXT: vmv.x.s a3, v8
+; LMULMAX2-RV32-NEXT: srli a4, a3, 8
+; LMULMAX2-RV32-NEXT: and a4, a4, a1
+; LMULMAX2-RV32-NEXT: srli a5, a3, 24
+; LMULMAX2-RV32-NEXT: or a4, a4, a5
+; LMULMAX2-RV32-NEXT: slli a5, a3, 8
+; LMULMAX2-RV32-NEXT: and a5, a5, a6
+; LMULMAX2-RV32-NEXT: slli a3, a3, 24
+; LMULMAX2-RV32-NEXT: or a3, a3, a5
+; LMULMAX2-RV32-NEXT: or a3, a3, a4
+; LMULMAX2-RV32-NEXT: sw a3, 12(sp)
+; LMULMAX2-RV32-NEXT: addi a3, zero, 32
+; LMULMAX2-RV32-NEXT: vsrl.vx v26, v26, a3
+; LMULMAX2-RV32-NEXT: vmv.x.s a4, v26
+; LMULMAX2-RV32-NEXT: srli a5, a4, 8
+; LMULMAX2-RV32-NEXT: and a5, a5, a1
+; LMULMAX2-RV32-NEXT: srli a2, a4, 24
+; LMULMAX2-RV32-NEXT: or a2, a5, a2
+; LMULMAX2-RV32-NEXT: slli a5, a4, 8
+; LMULMAX2-RV32-NEXT: and a5, a5, a6
+; LMULMAX2-RV32-NEXT: slli a4, a4, 24
+; LMULMAX2-RV32-NEXT: or a4, a4, a5
+; LMULMAX2-RV32-NEXT: or a2, a4, a2
+; LMULMAX2-RV32-NEXT: sw a2, 0(sp)
+; LMULMAX2-RV32-NEXT: vsrl.vx v26, v28, a3
+; LMULMAX2-RV32-NEXT: vmv.x.s a2, v26
+; LMULMAX2-RV32-NEXT: srli a4, a2, 8
+; LMULMAX2-RV32-NEXT: and a4, a4, a1
+; LMULMAX2-RV32-NEXT: srli a5, a2, 24
+; LMULMAX2-RV32-NEXT: or a4, a4, a5
+; LMULMAX2-RV32-NEXT: slli a5, a2, 8
+; LMULMAX2-RV32-NEXT: and a5, a5, a6
+; LMULMAX2-RV32-NEXT: slli a2, a2, 24
+; LMULMAX2-RV32-NEXT: or a2, a2, a5
+; LMULMAX2-RV32-NEXT: or a2, a2, a4
+; LMULMAX2-RV32-NEXT: sw a2, 24(sp)
+; LMULMAX2-RV32-NEXT: vsrl.vx v26, v30, a3
+; LMULMAX2-RV32-NEXT: vmv.x.s a2, v26
+; LMULMAX2-RV32-NEXT: srli a4, a2, 8
+; LMULMAX2-RV32-NEXT: and a4, a4, a1
+; LMULMAX2-RV32-NEXT: srli a5, a2, 24
+; LMULMAX2-RV32-NEXT: or a4, a4, a5
+; LMULMAX2-RV32-NEXT: slli a5, a2, 8
+; LMULMAX2-RV32-NEXT: and a5, a5, a6
+; LMULMAX2-RV32-NEXT: slli a2, a2, 24
+; LMULMAX2-RV32-NEXT: or a2, a2, a5
+; LMULMAX2-RV32-NEXT: or a2, a2, a4
+; LMULMAX2-RV32-NEXT: sw a2, 16(sp)
+; LMULMAX2-RV32-NEXT: vsrl.vx v26, v8, a3
+; LMULMAX2-RV32-NEXT: vmv.x.s a2, v26
+; LMULMAX2-RV32-NEXT: srli a3, a2, 8
+; LMULMAX2-RV32-NEXT: and a1, a3, a1
+; LMULMAX2-RV32-NEXT: srli a3, a2, 24
+; LMULMAX2-RV32-NEXT: or a1, a1, a3
+; LMULMAX2-RV32-NEXT: slli a3, a2, 8
+; LMULMAX2-RV32-NEXT: and a3, a3, a6
+; LMULMAX2-RV32-NEXT: slli a2, a2, 24
+; LMULMAX2-RV32-NEXT: or a2, a2, a3
+; LMULMAX2-RV32-NEXT: or a1, a2, a1
+; LMULMAX2-RV32-NEXT: sw a1, 8(sp)
+; LMULMAX2-RV32-NEXT: vsetivli a1, 8, e32,m2,ta,mu
+; LMULMAX2-RV32-NEXT: vle32.v v26, (sp)
; LMULMAX2-RV32-NEXT: vsetivli a1, 4, e64,m2,ta,mu
; LMULMAX2-RV32-NEXT: vse64.v v26, (a0)
; LMULMAX2-RV32-NEXT: addi sp, s0, -64
@@ -1077,107 +1967,112 @@ define void @bswap_v4i64(<4 x i64>* %x, <4 x i64>* %y) {
; LMULMAX2-RV64-NEXT: andi sp, sp, -32
; LMULMAX2-RV64-NEXT: vsetivli a1, 4, e64,m2,ta,mu
; LMULMAX2-RV64-NEXT: vle64.v v26, (a0)
-; LMULMAX2-RV64-NEXT: vsetvli zero, zero, e8,m2,ta,mu
; LMULMAX2-RV64-NEXT: vmv.x.s a1, v26
-; LMULMAX2-RV64-NEXT: sb a1, 7(sp)
-; LMULMAX2-RV64-NEXT: vsetivli a1, 1, e8,m2,ta,mu
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 24
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 31(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 25
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 30(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 26
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 29(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 27
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 28(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 28
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 27(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 29
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 26(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 30
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 25(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 31
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 24(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 16
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 23(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 17
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 22(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 18
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 21(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 19
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 20(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 20
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 19(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 21
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 18(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 22
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 17(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 23
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 16(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 8
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 15(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 9
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 14(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 10
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 13(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 11
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 12(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 12
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 11(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 13
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 10(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 14
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 9(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 15
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 8(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 1
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 6(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 2
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 5(sp)
+; LMULMAX2-RV64-NEXT: srli a2, a1, 40
+; LMULMAX2-RV64-NEXT: lui a3, 16
+; LMULMAX2-RV64-NEXT: addiw a7, a3, -256
+; LMULMAX2-RV64-NEXT: and a2, a2, a7
+; LMULMAX2-RV64-NEXT: srli a3, a1, 56
+; LMULMAX2-RV64-NEXT: or a2, a2, a3
+; LMULMAX2-RV64-NEXT: srli a3, a1, 24
+; LMULMAX2-RV64-NEXT: lui a6, 4080
+; LMULMAX2-RV64-NEXT: and a4, a3, a6
+; LMULMAX2-RV64-NEXT: srli a5, a1, 8
+; LMULMAX2-RV64-NEXT: addi a3, zero, 255
+; LMULMAX2-RV64-NEXT: slli t0, a3, 24
+; LMULMAX2-RV64-NEXT: and a5, a5, t0
+; LMULMAX2-RV64-NEXT: or a4, a5, a4
+; LMULMAX2-RV64-NEXT: or a4, a4, a2
+; LMULMAX2-RV64-NEXT: slli a2, a1, 8
+; LMULMAX2-RV64-NEXT: slli t1, a3, 32
+; LMULMAX2-RV64-NEXT: and a2, a2, t1
+; LMULMAX2-RV64-NEXT: slli a5, a1, 24
+; LMULMAX2-RV64-NEXT: slli t2, a3, 40
+; LMULMAX2-RV64-NEXT: and a5, a5, t2
+; LMULMAX2-RV64-NEXT: or a5, a5, a2
+; LMULMAX2-RV64-NEXT: slli a2, a1, 40
+; LMULMAX2-RV64-NEXT: slli a3, a3, 48
+; LMULMAX2-RV64-NEXT: and a2, a2, a3
+; LMULMAX2-RV64-NEXT: slli a1, a1, 56
+; LMULMAX2-RV64-NEXT: or a1, a1, a2
+; LMULMAX2-RV64-NEXT: or a1, a1, a5
+; LMULMAX2-RV64-NEXT: or a1, a1, a4
+; LMULMAX2-RV64-NEXT: sd a1, 0(sp)
+; LMULMAX2-RV64-NEXT: vsetivli a1, 1, e64,m2,ta,mu
; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 3
; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 4(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 4
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 3(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 5
-; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 2(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 6
+; LMULMAX2-RV64-NEXT: srli a2, a1, 40
+; LMULMAX2-RV64-NEXT: and a2, a2, a7
+; LMULMAX2-RV64-NEXT: srli a4, a1, 56
+; LMULMAX2-RV64-NEXT: or a2, a2, a4
+; LMULMAX2-RV64-NEXT: srli a4, a1, 24
+; LMULMAX2-RV64-NEXT: and a4, a4, a6
+; LMULMAX2-RV64-NEXT: srli a5, a1, 8
+; LMULMAX2-RV64-NEXT: and a5, a5, t0
+; LMULMAX2-RV64-NEXT: or a4, a5, a4
+; LMULMAX2-RV64-NEXT: or a2, a4, a2
+; LMULMAX2-RV64-NEXT: slli a4, a1, 8
+; LMULMAX2-RV64-NEXT: and a4, a4, t1
+; LMULMAX2-RV64-NEXT: slli a5, a1, 24
+; LMULMAX2-RV64-NEXT: and a5, a5, t2
+; LMULMAX2-RV64-NEXT: or a4, a5, a4
+; LMULMAX2-RV64-NEXT: slli a5, a1, 40
+; LMULMAX2-RV64-NEXT: and a5, a5, a3
+; LMULMAX2-RV64-NEXT: slli a1, a1, 56
+; LMULMAX2-RV64-NEXT: or a1, a1, a5
+; LMULMAX2-RV64-NEXT: or a1, a1, a4
+; LMULMAX2-RV64-NEXT: or a1, a1, a2
+; LMULMAX2-RV64-NEXT: sd a1, 24(sp)
+; LMULMAX2-RV64-NEXT: vslidedown.vi v28, v26, 2
; LMULMAX2-RV64-NEXT: vmv.x.s a1, v28
-; LMULMAX2-RV64-NEXT: sb a1, 1(sp)
-; LMULMAX2-RV64-NEXT: vslidedown.vi v26, v26, 7
+; LMULMAX2-RV64-NEXT: srli a2, a1, 40
+; LMULMAX2-RV64-NEXT: and a2, a2, a7
+; LMULMAX2-RV64-NEXT: srli a4, a1, 56
+; LMULMAX2-RV64-NEXT: or a2, a2, a4
+; LMULMAX2-RV64-NEXT: srli a4, a1, 24
+; LMULMAX2-RV64-NEXT: and a4, a4, a6
+; LMULMAX2-RV64-NEXT: srli a5, a1, 8
+; LMULMAX2-RV64-NEXT: and a5, a5, t0
+; LMULMAX2-RV64-NEXT: or a4, a5, a4
+; LMULMAX2-RV64-NEXT: or a2, a4, a2
+; LMULMAX2-RV64-NEXT: slli a4, a1, 8
+; LMULMAX2-RV64-NEXT: and a4, a4, t1
+; LMULMAX2-RV64-NEXT: slli a5, a1, 24
+; LMULMAX2-RV64-NEXT: and a5, a5, t2
+; LMULMAX2-RV64-NEXT: or a4, a5, a4
+; LMULMAX2-RV64-NEXT: slli a5, a1, 40
+; LMULMAX2-RV64-NEXT: and a5, a5, a3
+; LMULMAX2-RV64-NEXT: slli a1, a1, 56
+; LMULMAX2-RV64-NEXT: or a1, a1, a5
+; LMULMAX2-RV64-NEXT: or a1, a1, a4
+; LMULMAX2-RV64-NEXT: or a1, a1, a2
+; LMULMAX2-RV64-NEXT: sd a1, 16(sp)
+; LMULMAX2-RV64-NEXT: vslidedown.vi v26, v26, 1
; LMULMAX2-RV64-NEXT: vmv.x.s a1, v26
-; LMULMAX2-RV64-NEXT: sb a1, 0(sp)
-; LMULMAX2-RV64-NEXT: addi a1, zero, 32
-; LMULMAX2-RV64-NEXT: vsetvli a1, a1, e8,m2,ta,mu
-; LMULMAX2-RV64-NEXT: vle8.v v26, (sp)
+; LMULMAX2-RV64-NEXT: srli a2, a1, 40
+; LMULMAX2-RV64-NEXT: and a2, a2, a7
+; LMULMAX2-RV64-NEXT: srli a4, a1, 56
+; LMULMAX2-RV64-NEXT: or a2, a2, a4
+; LMULMAX2-RV64-NEXT: srli a4, a1, 24
+; LMULMAX2-RV64-NEXT: and a4, a4, a6
+; LMULMAX2-RV64-NEXT: srli a5, a1, 8
+; LMULMAX2-RV64-NEXT: and a5, a5, t0
+; LMULMAX2-RV64-NEXT: or a4, a5, a4
+; LMULMAX2-RV64-NEXT: or a2, a4, a2
+; LMULMAX2-RV64-NEXT: slli a4, a1, 8
+; LMULMAX2-RV64-NEXT: and a4, a4, t1
+; LMULMAX2-RV64-NEXT: slli a5, a1, 24
+; LMULMAX2-RV64-NEXT: and a5, a5, t2
+; LMULMAX2-RV64-NEXT: or a4, a5, a4
+; LMULMAX2-RV64-NEXT: slli a5, a1, 40
+; LMULMAX2-RV64-NEXT: and a3, a5, a3
+; LMULMAX2-RV64-NEXT: slli a1, a1, 56
+; LMULMAX2-RV64-NEXT: or a1, a1, a3
+; LMULMAX2-RV64-NEXT: or a1, a1, a4
+; LMULMAX2-RV64-NEXT: or a1, a1, a2
+; LMULMAX2-RV64-NEXT: sd a1, 8(sp)
; LMULMAX2-RV64-NEXT: vsetivli a1, 4, e64,m2,ta,mu
+; LMULMAX2-RV64-NEXT: vle64.v v26, (sp)
; LMULMAX2-RV64-NEXT: vse64.v v26, (a0)
; LMULMAX2-RV64-NEXT: addi sp, s0, -64
; LMULMAX2-RV64-NEXT: ld s0, 48(sp) # 8-byte Folded Reload
@@ -1185,119 +2080,242 @@ define void @bswap_v4i64(<4 x i64>* %x, <4 x i64>* %y) {
; LMULMAX2-RV64-NEXT: addi sp, sp, 64
; LMULMAX2-RV64-NEXT: ret
;
-; LMULMAX1-LABEL: bswap_v4i64:
-; LMULMAX1: # %bb.0:
-; LMULMAX1-NEXT: addi sp, sp, -32
-; LMULMAX1-NEXT: .cfi_def_cfa_offset 32
-; LMULMAX1-NEXT: vsetivli a1, 2, e64,m1,ta,mu
-; LMULMAX1-NEXT: addi a1, a0, 16
-; LMULMAX1-NEXT: vle64.v v26, (a1)
-; LMULMAX1-NEXT: vle64.v v25, (a0)
-; LMULMAX1-NEXT: vsetvli zero, zero, e8,m1,ta,mu
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 7(sp)
-; LMULMAX1-NEXT: vsetivli a2, 1, e8,m1,ta,mu
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 8
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 15(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 9
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 14(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 10
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 13(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 11
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 12(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 12
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 11(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 13
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 10(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 14
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 9(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 15
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 8(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 1
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 6(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 2
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 5(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 3
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 4(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 4
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 3(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 5
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 2(sp)
-; LMULMAX1-NEXT: vslidedown.vi v27, v26, 6
-; LMULMAX1-NEXT: vmv.x.s a2, v27
-; LMULMAX1-NEXT: sb a2, 1(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v26, 7
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 0(sp)
-; LMULMAX1-NEXT: vmv.x.s a2, v25
-; LMULMAX1-NEXT: sb a2, 23(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 8
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 31(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 9
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 30(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 10
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 29(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 11
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 28(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 12
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 27(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 13
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 26(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 14
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 25(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 15
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 24(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 1
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 22(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 2
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 21(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 3
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 20(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 4
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 19(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 5
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 18(sp)
-; LMULMAX1-NEXT: vslidedown.vi v26, v25, 6
-; LMULMAX1-NEXT: vmv.x.s a2, v26
-; LMULMAX1-NEXT: sb a2, 17(sp)
-; LMULMAX1-NEXT: vslidedown.vi v25, v25, 7
-; LMULMAX1-NEXT: vmv.x.s a2, v25
-; LMULMAX1-NEXT: sb a2, 16(sp)
-; LMULMAX1-NEXT: vsetivli a2, 16, e8,m1,ta,mu
-; LMULMAX1-NEXT: addi a2, sp, 16
-; LMULMAX1-NEXT: vle8.v v25, (a2)
-; LMULMAX1-NEXT: vle8.v v26, (sp)
-; LMULMAX1-NEXT: vsetivli a2, 2, e64,m1,ta,mu
-; LMULMAX1-NEXT: vse64.v v25, (a0)
-; LMULMAX1-NEXT: vse64.v v26, (a1)
-; LMULMAX1-NEXT: addi sp, sp, 32
-; LMULMAX1-NEXT: ret
+; LMULMAX1-RV32-LABEL: bswap_v4i64:
+; LMULMAX1-RV32: # %bb.0:
+; LMULMAX1-RV32-NEXT: addi sp, sp, -32
+; LMULMAX1-RV32-NEXT: .cfi_def_cfa_offset 32
+; LMULMAX1-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: addi a6, a0, 16
+; LMULMAX1-RV32-NEXT: vle64.v v26, (a6)
+; LMULMAX1-RV32-NEXT: vle64.v v25, (a0)
+; LMULMAX1-RV32-NEXT: vmv.x.s a4, v26
+; LMULMAX1-RV32-NEXT: srli a3, a4, 8
+; LMULMAX1-RV32-NEXT: lui a2, 16
+; LMULMAX1-RV32-NEXT: addi a2, a2, -256
+; LMULMAX1-RV32-NEXT: and a3, a3, a2
+; LMULMAX1-RV32-NEXT: srli a5, a4, 24
+; LMULMAX1-RV32-NEXT: or a5, a3, a5
+; LMULMAX1-RV32-NEXT: slli a1, a4, 8
+; LMULMAX1-RV32-NEXT: lui a3, 4080
+; LMULMAX1-RV32-NEXT: and a1, a1, a3
+; LMULMAX1-RV32-NEXT: slli a4, a4, 24
+; LMULMAX1-RV32-NEXT: or a1, a4, a1
+; LMULMAX1-RV32-NEXT: or a1, a1, a5
+; LMULMAX1-RV32-NEXT: sw a1, 20(sp)
+; LMULMAX1-RV32-NEXT: vsetivli a1, 1, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vslidedown.vi v27, v26, 1
+; LMULMAX1-RV32-NEXT: vmv.x.s a1, v27
+; LMULMAX1-RV32-NEXT: srli a4, a1, 8
+; LMULMAX1-RV32-NEXT: and a4, a4, a2
+; LMULMAX1-RV32-NEXT: srli a5, a1, 24
+; LMULMAX1-RV32-NEXT: or a4, a4, a5
+; LMULMAX1-RV32-NEXT: slli a5, a1, 8
+; LMULMAX1-RV32-NEXT: and a5, a5, a3
+; LMULMAX1-RV32-NEXT: slli a1, a1, 24
+; LMULMAX1-RV32-NEXT: or a1, a1, a5
+; LMULMAX1-RV32-NEXT: or a1, a1, a4
+; LMULMAX1-RV32-NEXT: sw a1, 28(sp)
+; LMULMAX1-RV32-NEXT: addi a7, zero, 32
+; LMULMAX1-RV32-NEXT: vsrl.vx v26, v26, a7
+; LMULMAX1-RV32-NEXT: vmv.x.s a4, v26
+; LMULMAX1-RV32-NEXT: srli a5, a4, 8
+; LMULMAX1-RV32-NEXT: and a5, a5, a2
+; LMULMAX1-RV32-NEXT: srli a1, a4, 24
+; LMULMAX1-RV32-NEXT: or a1, a5, a1
+; LMULMAX1-RV32-NEXT: slli a5, a4, 8
+; LMULMAX1-RV32-NEXT: and a5, a5, a3
+; LMULMAX1-RV32-NEXT: slli a4, a4, 24
+; LMULMAX1-RV32-NEXT: or a4, a4, a5
+; LMULMAX1-RV32-NEXT: or a1, a4, a1
+; LMULMAX1-RV32-NEXT: sw a1, 16(sp)
+; LMULMAX1-RV32-NEXT: vsrl.vx v26, v27, a7
+; LMULMAX1-RV32-NEXT: vmv.x.s a1, v26
+; LMULMAX1-RV32-NEXT: srli a4, a1, 8
+; LMULMAX1-RV32-NEXT: and a4, a4, a2
+; LMULMAX1-RV32-NEXT: srli a5, a1, 24
+; LMULMAX1-RV32-NEXT: or a4, a4, a5
+; LMULMAX1-RV32-NEXT: slli a5, a1, 8
+; LMULMAX1-RV32-NEXT: and a5, a5, a3
+; LMULMAX1-RV32-NEXT: slli a1, a1, 24
+; LMULMAX1-RV32-NEXT: or a1, a1, a5
+; LMULMAX1-RV32-NEXT: or a1, a1, a4
+; LMULMAX1-RV32-NEXT: sw a1, 24(sp)
+; LMULMAX1-RV32-NEXT: vmv.x.s a1, v25
+; LMULMAX1-RV32-NEXT: srli a4, a1, 8
+; LMULMAX1-RV32-NEXT: and a4, a4, a2
+; LMULMAX1-RV32-NEXT: srli a5, a1, 24
+; LMULMAX1-RV32-NEXT: or a4, a4, a5
+; LMULMAX1-RV32-NEXT: slli a5, a1, 8
+; LMULMAX1-RV32-NEXT: and a5, a5, a3
+; LMULMAX1-RV32-NEXT: slli a1, a1, 24
+; LMULMAX1-RV32-NEXT: or a1, a1, a5
+; LMULMAX1-RV32-NEXT: or a1, a1, a4
+; LMULMAX1-RV32-NEXT: sw a1, 4(sp)
+; LMULMAX1-RV32-NEXT: vslidedown.vi v26, v25, 1
+; LMULMAX1-RV32-NEXT: vmv.x.s a1, v26
+; LMULMAX1-RV32-NEXT: srli a4, a1, 8
+; LMULMAX1-RV32-NEXT: and a4, a4, a2
+; LMULMAX1-RV32-NEXT: srli a5, a1, 24
+; LMULMAX1-RV32-NEXT: or a4, a4, a5
+; LMULMAX1-RV32-NEXT: slli a5, a1, 8
+; LMULMAX1-RV32-NEXT: and a5, a5, a3
+; LMULMAX1-RV32-NEXT: slli a1, a1, 24
+; LMULMAX1-RV32-NEXT: or a1, a1, a5
+; LMULMAX1-RV32-NEXT: or a1, a1, a4
+; LMULMAX1-RV32-NEXT: sw a1, 12(sp)
+; LMULMAX1-RV32-NEXT: vsrl.vx v25, v25, a7
+; LMULMAX1-RV32-NEXT: vmv.x.s a1, v25
+; LMULMAX1-RV32-NEXT: srli a4, a1, 8
+; LMULMAX1-RV32-NEXT: and a4, a4, a2
+; LMULMAX1-RV32-NEXT: srli a5, a1, 24
+; LMULMAX1-RV32-NEXT: or a4, a4, a5
+; LMULMAX1-RV32-NEXT: slli a5, a1, 8
+; LMULMAX1-RV32-NEXT: and a5, a5, a3
+; LMULMAX1-RV32-NEXT: slli a1, a1, 24
+; LMULMAX1-RV32-NEXT: or a1, a1, a5
+; LMULMAX1-RV32-NEXT: or a1, a1, a4
+; LMULMAX1-RV32-NEXT: sw a1, 0(sp)
+; LMULMAX1-RV32-NEXT: vsrl.vx v25, v26, a7
+; LMULMAX1-RV32-NEXT: vmv.x.s a1, v25
+; LMULMAX1-RV32-NEXT: srli a4, a1, 8
+; LMULMAX1-RV32-NEXT: and a2, a4, a2
+; LMULMAX1-RV32-NEXT: srli a4, a1, 24
+; LMULMAX1-RV32-NEXT: or a2, a2, a4
+; LMULMAX1-RV32-NEXT: slli a4, a1, 8
+; LMULMAX1-RV32-NEXT: and a3, a4, a3
+; LMULMAX1-RV32-NEXT: slli a1, a1, 24
+; LMULMAX1-RV32-NEXT: or a1, a1, a3
+; LMULMAX1-RV32-NEXT: or a1, a1, a2
+; LMULMAX1-RV32-NEXT: sw a1, 8(sp)
+; LMULMAX1-RV32-NEXT: vsetivli a1, 4, e32,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vle32.v v25, (sp)
+; LMULMAX1-RV32-NEXT: addi a1, sp, 16
+; LMULMAX1-RV32-NEXT: vle32.v v26, (a1)
+; LMULMAX1-RV32-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX1-RV32-NEXT: vse64.v v25, (a0)
+; LMULMAX1-RV32-NEXT: vse64.v v26, (a6)
+; LMULMAX1-RV32-NEXT: addi sp, sp, 32
+; LMULMAX1-RV32-NEXT: ret
+;
+; LMULMAX1-RV64-LABEL: bswap_v4i64:
+; LMULMAX1-RV64: # %bb.0:
+; LMULMAX1-RV64-NEXT: addi sp, sp, -32
+; LMULMAX1-RV64-NEXT: .cfi_def_cfa_offset 32
+; LMULMAX1-RV64-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX1-RV64-NEXT: addi a6, a0, 16
+; LMULMAX1-RV64-NEXT: vle64.v v26, (a6)
+; LMULMAX1-RV64-NEXT: vle64.v v25, (a0)
+; LMULMAX1-RV64-NEXT: vmv.x.s a2, v26
+; LMULMAX1-RV64-NEXT: srli a1, a2, 40
+; LMULMAX1-RV64-NEXT: lui a3, 16
+; LMULMAX1-RV64-NEXT: addiw t0, a3, -256
+; LMULMAX1-RV64-NEXT: and a1, a1, t0
+; LMULMAX1-RV64-NEXT: srli a3, a2, 56
+; LMULMAX1-RV64-NEXT: or a1, a1, a3
+; LMULMAX1-RV64-NEXT: srli a3, a2, 24
+; LMULMAX1-RV64-NEXT: lui a7, 4080
+; LMULMAX1-RV64-NEXT: and a3, a3, a7
+; LMULMAX1-RV64-NEXT: srli a5, a2, 8
+; LMULMAX1-RV64-NEXT: addi a4, zero, 255
+; LMULMAX1-RV64-NEXT: slli t1, a4, 24
+; LMULMAX1-RV64-NEXT: and a5, a5, t1
+; LMULMAX1-RV64-NEXT: or a3, a5, a3
+; LMULMAX1-RV64-NEXT: or a5, a3, a1
+; LMULMAX1-RV64-NEXT: slli a1, a2, 8
+; LMULMAX1-RV64-NEXT: slli t2, a4, 32
+; LMULMAX1-RV64-NEXT: and a3, a1, t2
+; LMULMAX1-RV64-NEXT: slli a1, a2, 24
+; LMULMAX1-RV64-NEXT: slli t3, a4, 40
+; LMULMAX1-RV64-NEXT: and a1, a1, t3
+; LMULMAX1-RV64-NEXT: or a1, a1, a3
+; LMULMAX1-RV64-NEXT: slli a3, a2, 40
+; LMULMAX1-RV64-NEXT: slli a4, a4, 48
+; LMULMAX1-RV64-NEXT: and a3, a3, a4
+; LMULMAX1-RV64-NEXT: slli a2, a2, 56
+; LMULMAX1-RV64-NEXT: or a2, a2, a3
+; LMULMAX1-RV64-NEXT: or a1, a2, a1
+; LMULMAX1-RV64-NEXT: or a1, a1, a5
+; LMULMAX1-RV64-NEXT: sd a1, 16(sp)
+; LMULMAX1-RV64-NEXT: vsetivli a1, 1, e64,m1,ta,mu
+; LMULMAX1-RV64-NEXT: vslidedown.vi v26, v26, 1
+; LMULMAX1-RV64-NEXT: vmv.x.s a1, v26
+; LMULMAX1-RV64-NEXT: srli a2, a1, 40
+; LMULMAX1-RV64-NEXT: and a2, a2, t0
+; LMULMAX1-RV64-NEXT: srli a3, a1, 56
+; LMULMAX1-RV64-NEXT: or a2, a2, a3
+; LMULMAX1-RV64-NEXT: srli a3, a1, 24
+; LMULMAX1-RV64-NEXT: and a3, a3, a7
+; LMULMAX1-RV64-NEXT: srli a5, a1, 8
+; LMULMAX1-RV64-NEXT: and a5, a5, t1
+; LMULMAX1-RV64-NEXT: or a3, a5, a3
+; LMULMAX1-RV64-NEXT: or a2, a3, a2
+; LMULMAX1-RV64-NEXT: slli a3, a1, 8
+; LMULMAX1-RV64-NEXT: and a3, a3, t2
+; LMULMAX1-RV64-NEXT: slli a5, a1, 24
+; LMULMAX1-RV64-NEXT: and a5, a5, t3
+; LMULMAX1-RV64-NEXT: or a3, a5, a3
+; LMULMAX1-RV64-NEXT: slli a5, a1, 40
+; LMULMAX1-RV64-NEXT: and a5, a5, a4
+; LMULMAX1-RV64-NEXT: slli a1, a1, 56
+; LMULMAX1-RV64-NEXT: or a1, a1, a5
+; LMULMAX1-RV64-NEXT: or a1, a1, a3
+; LMULMAX1-RV64-NEXT: or a1, a1, a2
+; LMULMAX1-RV64-NEXT: sd a1, 24(sp)
+; LMULMAX1-RV64-NEXT: vmv.x.s a1, v25
+; LMULMAX1-RV64-NEXT: srli a2, a1, 40
+; LMULMAX1-RV64-NEXT: and a2, a2, t0
+; LMULMAX1-RV64-NEXT: srli a3, a1, 56
+; LMULMAX1-RV64-NEXT: or a2, a2, a3
+; LMULMAX1-RV64-NEXT: srli a3, a1, 24
+; LMULMAX1-RV64-NEXT: and a3, a3, a7
+; LMULMAX1-RV64-NEXT: srli a5, a1, 8
+; LMULMAX1-RV64-NEXT: and a5, a5, t1
+; LMULMAX1-RV64-NEXT: or a3, a5, a3
+; LMULMAX1-RV64-NEXT: or a2, a3, a2
+; LMULMAX1-RV64-NEXT: slli a3, a1, 8
+; LMULMAX1-RV64-NEXT: and a3, a3, t2
+; LMULMAX1-RV64-NEXT: slli a5, a1, 24
+; LMULMAX1-RV64-NEXT: and a5, a5, t3
+; LMULMAX1-RV64-NEXT: or a3, a5, a3
+; LMULMAX1-RV64-NEXT: slli a5, a1, 40
+; LMULMAX1-RV64-NEXT: and a5, a5, a4
+; LMULMAX1-RV64-NEXT: slli a1, a1, 56
+; LMULMAX1-RV64-NEXT: or a1, a1, a5
+; LMULMAX1-RV64-NEXT: or a1, a1, a3
+; LMULMAX1-RV64-NEXT: or a1, a1, a2
+; LMULMAX1-RV64-NEXT: sd a1, 0(sp)
+; LMULMAX1-RV64-NEXT: vslidedown.vi v25, v25, 1
+; LMULMAX1-RV64-NEXT: vmv.x.s a1, v25
+; LMULMAX1-RV64-NEXT: srli a2, a1, 40
+; LMULMAX1-RV64-NEXT: and a2, a2, t0
+; LMULMAX1-RV64-NEXT: srli a3, a1, 56
+; LMULMAX1-RV64-NEXT: or a2, a2, a3
+; LMULMAX1-RV64-NEXT: srli a3, a1, 24
+; LMULMAX1-RV64-NEXT: and a3, a3, a7
+; LMULMAX1-RV64-NEXT: srli a5, a1, 8
+; LMULMAX1-RV64-NEXT: and a5, a5, t1
+; LMULMAX1-RV64-NEXT: or a3, a5, a3
+; LMULMAX1-RV64-NEXT: or a2, a3, a2
+; LMULMAX1-RV64-NEXT: slli a3, a1, 8
+; LMULMAX1-RV64-NEXT: and a3, a3, t2
+; LMULMAX1-RV64-NEXT: slli a5, a1, 24
+; LMULMAX1-RV64-NEXT: and a5, a5, t3
+; LMULMAX1-RV64-NEXT: or a3, a5, a3
+; LMULMAX1-RV64-NEXT: slli a5, a1, 40
+; LMULMAX1-RV64-NEXT: and a4, a5, a4
+; LMULMAX1-RV64-NEXT: slli a1, a1, 56
+; LMULMAX1-RV64-NEXT: or a1, a1, a4
+; LMULMAX1-RV64-NEXT: or a1, a1, a3
+; LMULMAX1-RV64-NEXT: or a1, a1, a2
+; LMULMAX1-RV64-NEXT: sd a1, 8(sp)
+; LMULMAX1-RV64-NEXT: vsetivli a1, 2, e64,m1,ta,mu
+; LMULMAX1-RV64-NEXT: vle64.v v25, (sp)
+; LMULMAX1-RV64-NEXT: addi a1, sp, 16
+; LMULMAX1-RV64-NEXT: vle64.v v26, (a1)
+; LMULMAX1-RV64-NEXT: vse64.v v25, (a0)
+; LMULMAX1-RV64-NEXT: vse64.v v26, (a6)
+; LMULMAX1-RV64-NEXT: addi sp, sp, 32
+; LMULMAX1-RV64-NEXT: ret
%a = load <4 x i64>, <4 x i64>* %x
%b = load <4 x i64>, <4 x i64>* %y
%c = call <4 x i64> @llvm.bswap.v4i64(<4 x i64> %a)
More information about the llvm-commits
mailing list