--- /dev/null
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt < %s -mtriple=x86_64-unknown -basicaa -slp-vectorizer -S | FileCheck %s --check-prefix=CHECK --check-prefix=SSE
+; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=corei7-avx -basicaa -slp-vectorizer -S | FileCheck %s --check-prefix=CHECK --check-prefix=AVX
+; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=core-avx2 -basicaa -slp-vectorizer -S | FileCheck %s --check-prefix=CHECK --check-prefix=AVX
+; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=bdver2 -basicaa -slp-vectorizer -S | FileCheck %s --check-prefix=CHECK --check-prefix=AVX
+; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=bdver4 -basicaa -slp-vectorizer -S | FileCheck %s --check-prefix=CHECK --check-prefix=AVX
+
+target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
+
+@src64 = common global [4 x i64] zeroinitializer, align 32
+@dst64 = common global [4 x i64] zeroinitializer, align 32
+@src32 = common global [8 x i32] zeroinitializer, align 32
+@dst32 = common global [8 x i32] zeroinitializer, align 32
+@src16 = common global [16 x i16] zeroinitializer, align 32
+@dst16 = common global [16 x i16] zeroinitializer, align 32
+@src8 = common global [32 x i8] zeroinitializer, align 32
+@dst8 = common global [32 x i8] zeroinitializer, align 32
+
+declare i64 @llvm.bitreverse.i64(i64)
+declare i32 @llvm.bitreverse.i32(i32)
+declare i16 @llvm.bitreverse.i16(i16)
+declare i8 @llvm.bitreverse.i8(i8)
+
+define void @bitreverse_2i64() #0 {
+; CHECK-LABEL: @bitreverse_2i64(
+; CHECK-NEXT: [[LD0:%.*]] = load i64, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @src64, i32 0, i64 0), align 8
+; CHECK-NEXT: [[LD1:%.*]] = load i64, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @src64, i32 0, i64 1), align 8
+; CHECK-NEXT: [[BITREVERSE0:%.*]] = call i64 @llvm.bitreverse.i64(i64 [[LD0]])
+; CHECK-NEXT: [[BITREVERSE1:%.*]] = call i64 @llvm.bitreverse.i64(i64 [[LD1]])
+; CHECK-NEXT: store i64 [[BITREVERSE0]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i32 0, i64 0), align 8
+; CHECK-NEXT: store i64 [[BITREVERSE1]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i32 0, i64 1), align 8
+; CHECK-NEXT: ret void
+;
+ %ld0 = load i64, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @src64, i32 0, i64 0), align 8
+ %ld1 = load i64, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @src64, i32 0, i64 1), align 8
+ %bitreverse0 = call i64 @llvm.bitreverse.i64(i64 %ld0)
+ %bitreverse1 = call i64 @llvm.bitreverse.i64(i64 %ld1)
+ store i64 %bitreverse0, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i32 0, i64 0), align 8
+ store i64 %bitreverse1, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i32 0, i64 1), align 8
+ ret void
+}
+
+define void @bitreverse_4i64() #0 {
+; CHECK-LABEL: @bitreverse_4i64(
+; CHECK-NEXT: [[LD0:%.*]] = load i64, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @src64, i64 0, i64 0), align 4
+; CHECK-NEXT: [[LD1:%.*]] = load i64, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @src64, i64 0, i64 1), align 4
+; CHECK-NEXT: [[LD2:%.*]] = load i64, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @src64, i64 0, i64 2), align 4
+; CHECK-NEXT: [[LD3:%.*]] = load i64, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @src64, i64 0, i64 3), align 4
+; CHECK-NEXT: [[BITREVERSE0:%.*]] = call i64 @llvm.bitreverse.i64(i64 [[LD0]])
+; CHECK-NEXT: [[BITREVERSE1:%.*]] = call i64 @llvm.bitreverse.i64(i64 [[LD1]])
+; CHECK-NEXT: [[BITREVERSE2:%.*]] = call i64 @llvm.bitreverse.i64(i64 [[LD2]])
+; CHECK-NEXT: [[BITREVERSE3:%.*]] = call i64 @llvm.bitreverse.i64(i64 [[LD3]])
+; CHECK-NEXT: store i64 [[BITREVERSE0]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 0), align 4
+; CHECK-NEXT: store i64 [[BITREVERSE1]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 1), align 4
+; CHECK-NEXT: store i64 [[BITREVERSE2]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 2), align 4
+; CHECK-NEXT: store i64 [[BITREVERSE3]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 3), align 4
+; CHECK-NEXT: ret void
+;
+ %ld0 = load i64, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @src64, i64 0, i64 0), align 4
+ %ld1 = load i64, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @src64, i64 0, i64 1), align 4
+ %ld2 = load i64, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @src64, i64 0, i64 2), align 4
+ %ld3 = load i64, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @src64, i64 0, i64 3), align 4
+ %bitreverse0 = call i64 @llvm.bitreverse.i64(i64 %ld0)
+ %bitreverse1 = call i64 @llvm.bitreverse.i64(i64 %ld1)
+ %bitreverse2 = call i64 @llvm.bitreverse.i64(i64 %ld2)
+ %bitreverse3 = call i64 @llvm.bitreverse.i64(i64 %ld3)
+ store i64 %bitreverse0, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 0), align 4
+ store i64 %bitreverse1, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 1), align 4
+ store i64 %bitreverse2, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 2), align 4
+ store i64 %bitreverse3, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 3), align 4
+ ret void
+}
+
+define void @bitreverse_4i32() #0 {
+; CHECK-LABEL: @bitreverse_4i32(
+; CHECK-NEXT: [[LD0:%.*]] = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 0), align 4
+; CHECK-NEXT: [[LD1:%.*]] = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 1), align 4
+; CHECK-NEXT: [[LD2:%.*]] = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 2), align 4
+; CHECK-NEXT: [[LD3:%.*]] = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 3), align 4
+; CHECK-NEXT: [[BITREVERSE0:%.*]] = call i32 @llvm.bitreverse.i32(i32 [[LD0]])
+; CHECK-NEXT: [[BITREVERSE1:%.*]] = call i32 @llvm.bitreverse.i32(i32 [[LD1]])
+; CHECK-NEXT: [[BITREVERSE2:%.*]] = call i32 @llvm.bitreverse.i32(i32 [[LD2]])
+; CHECK-NEXT: [[BITREVERSE3:%.*]] = call i32 @llvm.bitreverse.i32(i32 [[LD3]])
+; CHECK-NEXT: store i32 [[BITREVERSE0]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 0), align 4
+; CHECK-NEXT: store i32 [[BITREVERSE1]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 1), align 4
+; CHECK-NEXT: store i32 [[BITREVERSE2]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 2), align 4
+; CHECK-NEXT: store i32 [[BITREVERSE3]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 3), align 4
+; CHECK-NEXT: ret void
+;
+ %ld0 = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 0), align 4
+ %ld1 = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 1), align 4
+ %ld2 = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 2), align 4
+ %ld3 = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 3), align 4
+ %bitreverse0 = call i32 @llvm.bitreverse.i32(i32 %ld0)
+ %bitreverse1 = call i32 @llvm.bitreverse.i32(i32 %ld1)
+ %bitreverse2 = call i32 @llvm.bitreverse.i32(i32 %ld2)
+ %bitreverse3 = call i32 @llvm.bitreverse.i32(i32 %ld3)
+ store i32 %bitreverse0, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 0), align 4
+ store i32 %bitreverse1, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 1), align 4
+ store i32 %bitreverse2, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 2), align 4
+ store i32 %bitreverse3, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 3), align 4
+ ret void
+}
+
+define void @bitreverse_8i32() #0 {
+; CHECK-LABEL: @bitreverse_8i32(
+; CHECK-NEXT: [[LD0:%.*]] = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 0), align 2
+; CHECK-NEXT: [[LD1:%.*]] = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 1), align 2
+; CHECK-NEXT: [[LD2:%.*]] = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 2), align 2
+; CHECK-NEXT: [[LD3:%.*]] = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 3), align 2
+; CHECK-NEXT: [[LD4:%.*]] = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 4), align 2
+; CHECK-NEXT: [[LD5:%.*]] = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 5), align 2
+; CHECK-NEXT: [[LD6:%.*]] = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 6), align 2
+; CHECK-NEXT: [[LD7:%.*]] = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 7), align 2
+; CHECK-NEXT: [[BITREVERSE0:%.*]] = call i32 @llvm.bitreverse.i32(i32 [[LD0]])
+; CHECK-NEXT: [[BITREVERSE1:%.*]] = call i32 @llvm.bitreverse.i32(i32 [[LD1]])
+; CHECK-NEXT: [[BITREVERSE2:%.*]] = call i32 @llvm.bitreverse.i32(i32 [[LD2]])
+; CHECK-NEXT: [[BITREVERSE3:%.*]] = call i32 @llvm.bitreverse.i32(i32 [[LD3]])
+; CHECK-NEXT: [[BITREVERSE4:%.*]] = call i32 @llvm.bitreverse.i32(i32 [[LD4]])
+; CHECK-NEXT: [[BITREVERSE5:%.*]] = call i32 @llvm.bitreverse.i32(i32 [[LD5]])
+; CHECK-NEXT: [[BITREVERSE6:%.*]] = call i32 @llvm.bitreverse.i32(i32 [[LD6]])
+; CHECK-NEXT: [[BITREVERSE7:%.*]] = call i32 @llvm.bitreverse.i32(i32 [[LD7]])
+; CHECK-NEXT: store i32 [[BITREVERSE0]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 0), align 2
+; CHECK-NEXT: store i32 [[BITREVERSE1]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 1), align 2
+; CHECK-NEXT: store i32 [[BITREVERSE2]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 2), align 2
+; CHECK-NEXT: store i32 [[BITREVERSE3]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 3), align 2
+; CHECK-NEXT: store i32 [[BITREVERSE4]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 4), align 2
+; CHECK-NEXT: store i32 [[BITREVERSE5]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 5), align 2
+; CHECK-NEXT: store i32 [[BITREVERSE6]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 6), align 2
+; CHECK-NEXT: store i32 [[BITREVERSE7]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 7), align 2
+; CHECK-NEXT: ret void
+;
+ %ld0 = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 0), align 2
+ %ld1 = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 1), align 2
+ %ld2 = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 2), align 2
+ %ld3 = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 3), align 2
+ %ld4 = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 4), align 2
+ %ld5 = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 5), align 2
+ %ld6 = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 6), align 2
+ %ld7 = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 7), align 2
+ %bitreverse0 = call i32 @llvm.bitreverse.i32(i32 %ld0)
+ %bitreverse1 = call i32 @llvm.bitreverse.i32(i32 %ld1)
+ %bitreverse2 = call i32 @llvm.bitreverse.i32(i32 %ld2)
+ %bitreverse3 = call i32 @llvm.bitreverse.i32(i32 %ld3)
+ %bitreverse4 = call i32 @llvm.bitreverse.i32(i32 %ld4)
+ %bitreverse5 = call i32 @llvm.bitreverse.i32(i32 %ld5)
+ %bitreverse6 = call i32 @llvm.bitreverse.i32(i32 %ld6)
+ %bitreverse7 = call i32 @llvm.bitreverse.i32(i32 %ld7)
+ store i32 %bitreverse0, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 0), align 2
+ store i32 %bitreverse1, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 1), align 2
+ store i32 %bitreverse2, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 2), align 2
+ store i32 %bitreverse3, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 3), align 2
+ store i32 %bitreverse4, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 4), align 2
+ store i32 %bitreverse5, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 5), align 2
+ store i32 %bitreverse6, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 6), align 2
+ store i32 %bitreverse7, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 7), align 2
+ ret void
+}
+
+define void @bitreverse_8i16() #0 {
+; CHECK-LABEL: @bitreverse_8i16(
+; CHECK-NEXT: [[LD0:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 0), align 2
+; CHECK-NEXT: [[LD1:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 1), align 2
+; CHECK-NEXT: [[LD2:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 2), align 2
+; CHECK-NEXT: [[LD3:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 3), align 2
+; CHECK-NEXT: [[LD4:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 4), align 2
+; CHECK-NEXT: [[LD5:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 5), align 2
+; CHECK-NEXT: [[LD6:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 6), align 2
+; CHECK-NEXT: [[LD7:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 7), align 2
+; CHECK-NEXT: [[BITREVERSE0:%.*]] = call i16 @llvm.bitreverse.i16(i16 [[LD0]])
+; CHECK-NEXT: [[BITREVERSE1:%.*]] = call i16 @llvm.bitreverse.i16(i16 [[LD1]])
+; CHECK-NEXT: [[BITREVERSE2:%.*]] = call i16 @llvm.bitreverse.i16(i16 [[LD2]])
+; CHECK-NEXT: [[BITREVERSE3:%.*]] = call i16 @llvm.bitreverse.i16(i16 [[LD3]])
+; CHECK-NEXT: [[BITREVERSE4:%.*]] = call i16 @llvm.bitreverse.i16(i16 [[LD4]])
+; CHECK-NEXT: [[BITREVERSE5:%.*]] = call i16 @llvm.bitreverse.i16(i16 [[LD5]])
+; CHECK-NEXT: [[BITREVERSE6:%.*]] = call i16 @llvm.bitreverse.i16(i16 [[LD6]])
+; CHECK-NEXT: [[BITREVERSE7:%.*]] = call i16 @llvm.bitreverse.i16(i16 [[LD7]])
+; CHECK-NEXT: store i16 [[BITREVERSE0]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 0), align 2
+; CHECK-NEXT: store i16 [[BITREVERSE1]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 1), align 2
+; CHECK-NEXT: store i16 [[BITREVERSE2]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 2), align 2
+; CHECK-NEXT: store i16 [[BITREVERSE3]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 3), align 2
+; CHECK-NEXT: store i16 [[BITREVERSE4]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 4), align 2
+; CHECK-NEXT: store i16 [[BITREVERSE5]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 5), align 2
+; CHECK-NEXT: store i16 [[BITREVERSE6]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 6), align 2
+; CHECK-NEXT: store i16 [[BITREVERSE7]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 7), align 2
+; CHECK-NEXT: ret void
+;
+ %ld0 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 0), align 2
+ %ld1 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 1), align 2
+ %ld2 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 2), align 2
+ %ld3 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 3), align 2
+ %ld4 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 4), align 2
+ %ld5 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 5), align 2
+ %ld6 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 6), align 2
+ %ld7 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 7), align 2
+ %bitreverse0 = call i16 @llvm.bitreverse.i16(i16 %ld0)
+ %bitreverse1 = call i16 @llvm.bitreverse.i16(i16 %ld1)
+ %bitreverse2 = call i16 @llvm.bitreverse.i16(i16 %ld2)
+ %bitreverse3 = call i16 @llvm.bitreverse.i16(i16 %ld3)
+ %bitreverse4 = call i16 @llvm.bitreverse.i16(i16 %ld4)
+ %bitreverse5 = call i16 @llvm.bitreverse.i16(i16 %ld5)
+ %bitreverse6 = call i16 @llvm.bitreverse.i16(i16 %ld6)
+ %bitreverse7 = call i16 @llvm.bitreverse.i16(i16 %ld7)
+ store i16 %bitreverse0, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 0), align 2
+ store i16 %bitreverse1, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 1), align 2
+ store i16 %bitreverse2, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 2), align 2
+ store i16 %bitreverse3, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 3), align 2
+ store i16 %bitreverse4, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 4), align 2
+ store i16 %bitreverse5, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 5), align 2
+ store i16 %bitreverse6, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 6), align 2
+ store i16 %bitreverse7, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 7), align 2
+ ret void
+}
+
+define void @bitreverse_16i16() #0 {
+; CHECK-LABEL: @bitreverse_16i16(
+; CHECK-NEXT: [[LD0:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 0), align 2
+; CHECK-NEXT: [[LD1:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 1), align 2
+; CHECK-NEXT: [[LD2:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 2), align 2
+; CHECK-NEXT: [[LD3:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 3), align 2
+; CHECK-NEXT: [[LD4:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 4), align 2
+; CHECK-NEXT: [[LD5:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 5), align 2
+; CHECK-NEXT: [[LD6:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 6), align 2
+; CHECK-NEXT: [[LD7:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 7), align 2
+; CHECK-NEXT: [[LD8:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 8), align 2
+; CHECK-NEXT: [[LD9:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 9), align 2
+; CHECK-NEXT: [[LD10:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 10), align 2
+; CHECK-NEXT: [[LD11:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 11), align 2
+; CHECK-NEXT: [[LD12:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 12), align 2
+; CHECK-NEXT: [[LD13:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 13), align 2
+; CHECK-NEXT: [[LD14:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 14), align 2
+; CHECK-NEXT: [[LD15:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 15), align 2
+; CHECK-NEXT: [[BITREVERSE0:%.*]] = call i16 @llvm.bitreverse.i16(i16 [[LD0]])
+; CHECK-NEXT: [[BITREVERSE1:%.*]] = call i16 @llvm.bitreverse.i16(i16 [[LD1]])
+; CHECK-NEXT: [[BITREVERSE2:%.*]] = call i16 @llvm.bitreverse.i16(i16 [[LD2]])
+; CHECK-NEXT: [[BITREVERSE3:%.*]] = call i16 @llvm.bitreverse.i16(i16 [[LD3]])
+; CHECK-NEXT: [[BITREVERSE4:%.*]] = call i16 @llvm.bitreverse.i16(i16 [[LD4]])
+; CHECK-NEXT: [[BITREVERSE5:%.*]] = call i16 @llvm.bitreverse.i16(i16 [[LD5]])
+; CHECK-NEXT: [[BITREVERSE6:%.*]] = call i16 @llvm.bitreverse.i16(i16 [[LD6]])
+; CHECK-NEXT: [[BITREVERSE7:%.*]] = call i16 @llvm.bitreverse.i16(i16 [[LD7]])
+; CHECK-NEXT: [[BITREVERSE8:%.*]] = call i16 @llvm.bitreverse.i16(i16 [[LD8]])
+; CHECK-NEXT: [[BITREVERSE9:%.*]] = call i16 @llvm.bitreverse.i16(i16 [[LD9]])
+; CHECK-NEXT: [[BITREVERSE10:%.*]] = call i16 @llvm.bitreverse.i16(i16 [[LD10]])
+; CHECK-NEXT: [[BITREVERSE11:%.*]] = call i16 @llvm.bitreverse.i16(i16 [[LD11]])
+; CHECK-NEXT: [[BITREVERSE12:%.*]] = call i16 @llvm.bitreverse.i16(i16 [[LD12]])
+; CHECK-NEXT: [[BITREVERSE13:%.*]] = call i16 @llvm.bitreverse.i16(i16 [[LD13]])
+; CHECK-NEXT: [[BITREVERSE14:%.*]] = call i16 @llvm.bitreverse.i16(i16 [[LD14]])
+; CHECK-NEXT: [[BITREVERSE15:%.*]] = call i16 @llvm.bitreverse.i16(i16 [[LD15]])
+; CHECK-NEXT: store i16 [[BITREVERSE0]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 0), align 2
+; CHECK-NEXT: store i16 [[BITREVERSE1]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 1), align 2
+; CHECK-NEXT: store i16 [[BITREVERSE2]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 2), align 2
+; CHECK-NEXT: store i16 [[BITREVERSE3]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 3), align 2
+; CHECK-NEXT: store i16 [[BITREVERSE4]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 4), align 2
+; CHECK-NEXT: store i16 [[BITREVERSE5]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 5), align 2
+; CHECK-NEXT: store i16 [[BITREVERSE6]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 6), align 2
+; CHECK-NEXT: store i16 [[BITREVERSE7]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 7), align 2
+; CHECK-NEXT: store i16 [[BITREVERSE8]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 8), align 2
+; CHECK-NEXT: store i16 [[BITREVERSE9]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 9), align 2
+; CHECK-NEXT: store i16 [[BITREVERSE10]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 10), align 2
+; CHECK-NEXT: store i16 [[BITREVERSE11]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 11), align 2
+; CHECK-NEXT: store i16 [[BITREVERSE12]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 12), align 2
+; CHECK-NEXT: store i16 [[BITREVERSE13]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 13), align 2
+; CHECK-NEXT: store i16 [[BITREVERSE14]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 14), align 2
+; CHECK-NEXT: store i16 [[BITREVERSE15]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 15), align 2
+; CHECK-NEXT: ret void
+;
+ %ld0 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 0), align 2
+ %ld1 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 1), align 2
+ %ld2 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 2), align 2
+ %ld3 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 3), align 2
+ %ld4 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 4), align 2
+ %ld5 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 5), align 2
+ %ld6 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 6), align 2
+ %ld7 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 7), align 2
+ %ld8 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 8), align 2
+ %ld9 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 9), align 2
+ %ld10 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 10), align 2
+ %ld11 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 11), align 2
+ %ld12 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 12), align 2
+ %ld13 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 13), align 2
+ %ld14 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 14), align 2
+ %ld15 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 15), align 2
+ %bitreverse0 = call i16 @llvm.bitreverse.i16(i16 %ld0)
+ %bitreverse1 = call i16 @llvm.bitreverse.i16(i16 %ld1)
+ %bitreverse2 = call i16 @llvm.bitreverse.i16(i16 %ld2)
+ %bitreverse3 = call i16 @llvm.bitreverse.i16(i16 %ld3)
+ %bitreverse4 = call i16 @llvm.bitreverse.i16(i16 %ld4)
+ %bitreverse5 = call i16 @llvm.bitreverse.i16(i16 %ld5)
+ %bitreverse6 = call i16 @llvm.bitreverse.i16(i16 %ld6)
+ %bitreverse7 = call i16 @llvm.bitreverse.i16(i16 %ld7)
+ %bitreverse8 = call i16 @llvm.bitreverse.i16(i16 %ld8)
+ %bitreverse9 = call i16 @llvm.bitreverse.i16(i16 %ld9)
+ %bitreverse10 = call i16 @llvm.bitreverse.i16(i16 %ld10)
+ %bitreverse11 = call i16 @llvm.bitreverse.i16(i16 %ld11)
+ %bitreverse12 = call i16 @llvm.bitreverse.i16(i16 %ld12)
+ %bitreverse13 = call i16 @llvm.bitreverse.i16(i16 %ld13)
+ %bitreverse14 = call i16 @llvm.bitreverse.i16(i16 %ld14)
+ %bitreverse15 = call i16 @llvm.bitreverse.i16(i16 %ld15)
+ store i16 %bitreverse0 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 0), align 2
+ store i16 %bitreverse1 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 1), align 2
+ store i16 %bitreverse2 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 2), align 2
+ store i16 %bitreverse3 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 3), align 2
+ store i16 %bitreverse4 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 4), align 2
+ store i16 %bitreverse5 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 5), align 2
+ store i16 %bitreverse6 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 6), align 2
+ store i16 %bitreverse7 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 7), align 2
+ store i16 %bitreverse8 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 8), align 2
+ store i16 %bitreverse9 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 9), align 2
+ store i16 %bitreverse10, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 10), align 2
+ store i16 %bitreverse11, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 11), align 2
+ store i16 %bitreverse12, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 12), align 2
+ store i16 %bitreverse13, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 13), align 2
+ store i16 %bitreverse14, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 14), align 2
+ store i16 %bitreverse15, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 15), align 2
+ ret void
+}
+
+define void @bitreverse_16i8() #0 {
+; CHECK-LABEL: @bitreverse_16i8(
+; CHECK-NEXT: [[LD0:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 0), align 1
+; CHECK-NEXT: [[LD1:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 1), align 1
+; CHECK-NEXT: [[LD2:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 2), align 1
+; CHECK-NEXT: [[LD3:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 3), align 1
+; CHECK-NEXT: [[LD4:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 4), align 1
+; CHECK-NEXT: [[LD5:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 5), align 1
+; CHECK-NEXT: [[LD6:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 6), align 1
+; CHECK-NEXT: [[LD7:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 7), align 1
+; CHECK-NEXT: [[LD8:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 8), align 1
+; CHECK-NEXT: [[LD9:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 9), align 1
+; CHECK-NEXT: [[LD10:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 10), align 1
+; CHECK-NEXT: [[LD11:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 11), align 1
+; CHECK-NEXT: [[LD12:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 12), align 1
+; CHECK-NEXT: [[LD13:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 13), align 1
+; CHECK-NEXT: [[LD14:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 14), align 1
+; CHECK-NEXT: [[LD15:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 15), align 1
+; CHECK-NEXT: [[BITREVERSE0:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD0]])
+; CHECK-NEXT: [[BITREVERSE1:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD1]])
+; CHECK-NEXT: [[BITREVERSE2:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD2]])
+; CHECK-NEXT: [[BITREVERSE3:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD3]])
+; CHECK-NEXT: [[BITREVERSE4:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD4]])
+; CHECK-NEXT: [[BITREVERSE5:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD5]])
+; CHECK-NEXT: [[BITREVERSE6:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD6]])
+; CHECK-NEXT: [[BITREVERSE7:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD7]])
+; CHECK-NEXT: [[BITREVERSE8:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD8]])
+; CHECK-NEXT: [[BITREVERSE9:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD9]])
+; CHECK-NEXT: [[BITREVERSE10:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD10]])
+; CHECK-NEXT: [[BITREVERSE11:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD11]])
+; CHECK-NEXT: [[BITREVERSE12:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD12]])
+; CHECK-NEXT: [[BITREVERSE13:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD13]])
+; CHECK-NEXT: [[BITREVERSE14:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD14]])
+; CHECK-NEXT: [[BITREVERSE15:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD15]])
+; CHECK-NEXT: store i8 [[BITREVERSE0]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 0), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE1]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 1), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE2]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 2), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE3]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 3), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE4]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 4), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE5]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 5), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE6]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 6), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE7]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 7), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE8]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 8), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE9]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 9), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE10]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 10), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE11]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 11), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE12]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 12), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE13]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 13), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE14]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 14), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE15]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 15), align 1
+; CHECK-NEXT: ret void
+;
+ %ld0 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 0), align 1
+ %ld1 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 1), align 1
+ %ld2 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 2), align 1
+ %ld3 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 3), align 1
+ %ld4 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 4), align 1
+ %ld5 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 5), align 1
+ %ld6 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 6), align 1
+ %ld7 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 7), align 1
+ %ld8 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 8), align 1
+ %ld9 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 9), align 1
+ %ld10 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 10), align 1
+ %ld11 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 11), align 1
+ %ld12 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 12), align 1
+ %ld13 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 13), align 1
+ %ld14 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 14), align 1
+ %ld15 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 15), align 1
+ %bitreverse0 = call i8 @llvm.bitreverse.i8(i8 %ld0)
+ %bitreverse1 = call i8 @llvm.bitreverse.i8(i8 %ld1)
+ %bitreverse2 = call i8 @llvm.bitreverse.i8(i8 %ld2)
+ %bitreverse3 = call i8 @llvm.bitreverse.i8(i8 %ld3)
+ %bitreverse4 = call i8 @llvm.bitreverse.i8(i8 %ld4)
+ %bitreverse5 = call i8 @llvm.bitreverse.i8(i8 %ld5)
+ %bitreverse6 = call i8 @llvm.bitreverse.i8(i8 %ld6)
+ %bitreverse7 = call i8 @llvm.bitreverse.i8(i8 %ld7)
+ %bitreverse8 = call i8 @llvm.bitreverse.i8(i8 %ld8)
+ %bitreverse9 = call i8 @llvm.bitreverse.i8(i8 %ld9)
+ %bitreverse10 = call i8 @llvm.bitreverse.i8(i8 %ld10)
+ %bitreverse11 = call i8 @llvm.bitreverse.i8(i8 %ld11)
+ %bitreverse12 = call i8 @llvm.bitreverse.i8(i8 %ld12)
+ %bitreverse13 = call i8 @llvm.bitreverse.i8(i8 %ld13)
+ %bitreverse14 = call i8 @llvm.bitreverse.i8(i8 %ld14)
+ %bitreverse15 = call i8 @llvm.bitreverse.i8(i8 %ld15)
+ store i8 %bitreverse0 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 0), align 1
+ store i8 %bitreverse1 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 1), align 1
+ store i8 %bitreverse2 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 2), align 1
+ store i8 %bitreverse3 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 3), align 1
+ store i8 %bitreverse4 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 4), align 1
+ store i8 %bitreverse5 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 5), align 1
+ store i8 %bitreverse6 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 6), align 1
+ store i8 %bitreverse7 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 7), align 1
+ store i8 %bitreverse8 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 8), align 1
+ store i8 %bitreverse9 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 9), align 1
+ store i8 %bitreverse10, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 10), align 1
+ store i8 %bitreverse11, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 11), align 1
+ store i8 %bitreverse12, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 12), align 1
+ store i8 %bitreverse13, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 13), align 1
+ store i8 %bitreverse14, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 14), align 1
+ store i8 %bitreverse15, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 15), align 1
+ ret void
+}
+
+define void @bitreverse_32i8() #0 {
+; CHECK-LABEL: @bitreverse_32i8(
+; CHECK-NEXT: [[LD0:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 0), align 1
+; CHECK-NEXT: [[LD1:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 1), align 1
+; CHECK-NEXT: [[LD2:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 2), align 1
+; CHECK-NEXT: [[LD3:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 3), align 1
+; CHECK-NEXT: [[LD4:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 4), align 1
+; CHECK-NEXT: [[LD5:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 5), align 1
+; CHECK-NEXT: [[LD6:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 6), align 1
+; CHECK-NEXT: [[LD7:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 7), align 1
+; CHECK-NEXT: [[LD8:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 8), align 1
+; CHECK-NEXT: [[LD9:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 9), align 1
+; CHECK-NEXT: [[LD10:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 10), align 1
+; CHECK-NEXT: [[LD11:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 11), align 1
+; CHECK-NEXT: [[LD12:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 12), align 1
+; CHECK-NEXT: [[LD13:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 13), align 1
+; CHECK-NEXT: [[LD14:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 14), align 1
+; CHECK-NEXT: [[LD15:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 15), align 1
+; CHECK-NEXT: [[LD16:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 16), align 1
+; CHECK-NEXT: [[LD17:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 17), align 1
+; CHECK-NEXT: [[LD18:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 18), align 1
+; CHECK-NEXT: [[LD19:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 19), align 1
+; CHECK-NEXT: [[LD20:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 20), align 1
+; CHECK-NEXT: [[LD21:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 21), align 1
+; CHECK-NEXT: [[LD22:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 22), align 1
+; CHECK-NEXT: [[LD23:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 23), align 1
+; CHECK-NEXT: [[LD24:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 24), align 1
+; CHECK-NEXT: [[LD25:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 25), align 1
+; CHECK-NEXT: [[LD26:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 26), align 1
+; CHECK-NEXT: [[LD27:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 27), align 1
+; CHECK-NEXT: [[LD28:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 28), align 1
+; CHECK-NEXT: [[LD29:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 29), align 1
+; CHECK-NEXT: [[LD30:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 30), align 1
+; CHECK-NEXT: [[LD31:%.*]] = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 31), align 1
+; CHECK-NEXT: [[BITREVERSE0:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD0]])
+; CHECK-NEXT: [[BITREVERSE1:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD1]])
+; CHECK-NEXT: [[BITREVERSE2:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD2]])
+; CHECK-NEXT: [[BITREVERSE3:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD3]])
+; CHECK-NEXT: [[BITREVERSE4:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD4]])
+; CHECK-NEXT: [[BITREVERSE5:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD5]])
+; CHECK-NEXT: [[BITREVERSE6:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD6]])
+; CHECK-NEXT: [[BITREVERSE7:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD7]])
+; CHECK-NEXT: [[BITREVERSE8:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD8]])
+; CHECK-NEXT: [[BITREVERSE9:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD9]])
+; CHECK-NEXT: [[BITREVERSE10:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD10]])
+; CHECK-NEXT: [[BITREVERSE11:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD11]])
+; CHECK-NEXT: [[BITREVERSE12:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD12]])
+; CHECK-NEXT: [[BITREVERSE13:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD13]])
+; CHECK-NEXT: [[BITREVERSE14:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD14]])
+; CHECK-NEXT: [[BITREVERSE15:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD15]])
+; CHECK-NEXT: [[BITREVERSE16:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD16]])
+; CHECK-NEXT: [[BITREVERSE17:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD17]])
+; CHECK-NEXT: [[BITREVERSE18:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD18]])
+; CHECK-NEXT: [[BITREVERSE19:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD19]])
+; CHECK-NEXT: [[BITREVERSE20:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD20]])
+; CHECK-NEXT: [[BITREVERSE21:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD21]])
+; CHECK-NEXT: [[BITREVERSE22:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD22]])
+; CHECK-NEXT: [[BITREVERSE23:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD23]])
+; CHECK-NEXT: [[BITREVERSE24:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD24]])
+; CHECK-NEXT: [[BITREVERSE25:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD25]])
+; CHECK-NEXT: [[BITREVERSE26:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD26]])
+; CHECK-NEXT: [[BITREVERSE27:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD27]])
+; CHECK-NEXT: [[BITREVERSE28:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD28]])
+; CHECK-NEXT: [[BITREVERSE29:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD29]])
+; CHECK-NEXT: [[BITREVERSE30:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD30]])
+; CHECK-NEXT: [[BITREVERSE31:%.*]] = call i8 @llvm.bitreverse.i8(i8 [[LD31]])
+; CHECK-NEXT: store i8 [[BITREVERSE0]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 0), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE1]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 1), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE2]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 2), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE3]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 3), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE4]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 4), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE5]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 5), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE6]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 6), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE7]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 7), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE8]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 8), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE9]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 9), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE10]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 10), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE11]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 11), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE12]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 12), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE13]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 13), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE14]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 14), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE15]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 15), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE16]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 16), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE17]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 17), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE18]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 18), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE19]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 19), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE20]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 20), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE21]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 21), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE22]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 22), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE23]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 23), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE24]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 24), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE25]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 25), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE26]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 26), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE27]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 27), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE28]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 28), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE29]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 29), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE30]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 30), align 1
+; CHECK-NEXT: store i8 [[BITREVERSE31]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 31), align 1
+; CHECK-NEXT: ret void
+;
+ %ld0 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 0), align 1
+ %ld1 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 1), align 1
+ %ld2 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 2), align 1
+ %ld3 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 3), align 1
+ %ld4 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 4), align 1
+ %ld5 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 5), align 1
+ %ld6 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 6), align 1
+ %ld7 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 7), align 1
+ %ld8 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 8), align 1
+ %ld9 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 9), align 1
+ %ld10 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 10), align 1
+ %ld11 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 11), align 1
+ %ld12 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 12), align 1
+ %ld13 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 13), align 1
+ %ld14 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 14), align 1
+ %ld15 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 15), align 1
+ %ld16 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 16), align 1
+ %ld17 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 17), align 1
+ %ld18 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 18), align 1
+ %ld19 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 19), align 1
+ %ld20 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 20), align 1
+ %ld21 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 21), align 1
+ %ld22 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 22), align 1
+ %ld23 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 23), align 1
+ %ld24 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 24), align 1
+ %ld25 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 25), align 1
+ %ld26 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 26), align 1
+ %ld27 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 27), align 1
+ %ld28 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 28), align 1
+ %ld29 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 29), align 1
+ %ld30 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 30), align 1
+ %ld31 = load i8, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 31), align 1
+ %bitreverse0 = call i8 @llvm.bitreverse.i8(i8 %ld0)
+ %bitreverse1 = call i8 @llvm.bitreverse.i8(i8 %ld1)
+ %bitreverse2 = call i8 @llvm.bitreverse.i8(i8 %ld2)
+ %bitreverse3 = call i8 @llvm.bitreverse.i8(i8 %ld3)
+ %bitreverse4 = call i8 @llvm.bitreverse.i8(i8 %ld4)
+ %bitreverse5 = call i8 @llvm.bitreverse.i8(i8 %ld5)
+ %bitreverse6 = call i8 @llvm.bitreverse.i8(i8 %ld6)
+ %bitreverse7 = call i8 @llvm.bitreverse.i8(i8 %ld7)
+ %bitreverse8 = call i8 @llvm.bitreverse.i8(i8 %ld8)
+ %bitreverse9 = call i8 @llvm.bitreverse.i8(i8 %ld9)
+ %bitreverse10 = call i8 @llvm.bitreverse.i8(i8 %ld10)
+ %bitreverse11 = call i8 @llvm.bitreverse.i8(i8 %ld11)
+ %bitreverse12 = call i8 @llvm.bitreverse.i8(i8 %ld12)
+ %bitreverse13 = call i8 @llvm.bitreverse.i8(i8 %ld13)
+ %bitreverse14 = call i8 @llvm.bitreverse.i8(i8 %ld14)
+ %bitreverse15 = call i8 @llvm.bitreverse.i8(i8 %ld15)
+ %bitreverse16 = call i8 @llvm.bitreverse.i8(i8 %ld16)
+ %bitreverse17 = call i8 @llvm.bitreverse.i8(i8 %ld17)
+ %bitreverse18 = call i8 @llvm.bitreverse.i8(i8 %ld18)
+ %bitreverse19 = call i8 @llvm.bitreverse.i8(i8 %ld19)
+ %bitreverse20 = call i8 @llvm.bitreverse.i8(i8 %ld20)
+ %bitreverse21 = call i8 @llvm.bitreverse.i8(i8 %ld21)
+ %bitreverse22 = call i8 @llvm.bitreverse.i8(i8 %ld22)
+ %bitreverse23 = call i8 @llvm.bitreverse.i8(i8 %ld23)
+ %bitreverse24 = call i8 @llvm.bitreverse.i8(i8 %ld24)
+ %bitreverse25 = call i8 @llvm.bitreverse.i8(i8 %ld25)
+ %bitreverse26 = call i8 @llvm.bitreverse.i8(i8 %ld26)
+ %bitreverse27 = call i8 @llvm.bitreverse.i8(i8 %ld27)
+ %bitreverse28 = call i8 @llvm.bitreverse.i8(i8 %ld28)
+ %bitreverse29 = call i8 @llvm.bitreverse.i8(i8 %ld29)
+ %bitreverse30 = call i8 @llvm.bitreverse.i8(i8 %ld30)
+ %bitreverse31 = call i8 @llvm.bitreverse.i8(i8 %ld31)
+ store i8 %bitreverse0 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 0), align 1
+ store i8 %bitreverse1 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 1), align 1
+ store i8 %bitreverse2 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 2), align 1
+ store i8 %bitreverse3 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 3), align 1
+ store i8 %bitreverse4 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 4), align 1
+ store i8 %bitreverse5 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 5), align 1
+ store i8 %bitreverse6 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 6), align 1
+ store i8 %bitreverse7 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 7), align 1
+ store i8 %bitreverse8 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 8), align 1
+ store i8 %bitreverse9 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 9), align 1
+ store i8 %bitreverse10, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 10), align 1
+ store i8 %bitreverse11, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 11), align 1
+ store i8 %bitreverse12, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 12), align 1
+ store i8 %bitreverse13, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 13), align 1
+ store i8 %bitreverse14, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 14), align 1
+ store i8 %bitreverse15, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 15), align 1
+ store i8 %bitreverse16, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 16), align 1
+ store i8 %bitreverse17, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 17), align 1
+ store i8 %bitreverse18, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 18), align 1
+ store i8 %bitreverse19, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 19), align 1
+ store i8 %bitreverse20, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 20), align 1
+ store i8 %bitreverse21, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 21), align 1
+ store i8 %bitreverse22, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 22), align 1
+ store i8 %bitreverse23, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 23), align 1
+ store i8 %bitreverse24, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 24), align 1
+ store i8 %bitreverse25, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 25), align 1
+ store i8 %bitreverse26, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 26), align 1
+ store i8 %bitreverse27, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 27), align 1
+ store i8 %bitreverse28, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 28), align 1
+ store i8 %bitreverse29, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 29), align 1
+ store i8 %bitreverse30, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 30), align 1
+ store i8 %bitreverse31, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 31), align 1
+ ret void
+}
+
+attributes #0 = { nounwind }
+
--- /dev/null
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt < %s -mtriple=x86_64-unknown -basicaa -slp-vectorizer -S | FileCheck %s --check-prefix=CHECK --check-prefix=SSE
+; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=corei7-avx -basicaa -slp-vectorizer -S | FileCheck %s --check-prefix=CHECK --check-prefix=AVX
+; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=core-avx2 -basicaa -slp-vectorizer -S | FileCheck %s --check-prefix=CHECK --check-prefix=AVX
+
+target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
+
+@src64 = common global [4 x i64] zeroinitializer, align 32
+@dst64 = common global [4 x i64] zeroinitializer, align 32
+@src32 = common global [8 x i32] zeroinitializer, align 32
+@dst32 = common global [8 x i32] zeroinitializer, align 32
+@src16 = common global [16 x i16] zeroinitializer, align 32
+@dst16 = common global [16 x i16] zeroinitializer, align 32
+
+declare i64 @llvm.bswap.i64(i64)
+declare i32 @llvm.bswap.i32(i32)
+declare i16 @llvm.bswap.i16(i16)
+
+define void @bswap_2i64() #0 {
+; CHECK-LABEL: @bswap_2i64(
+; CHECK-NEXT: [[LD0:%.*]] = load i64, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @src64, i32 0, i64 0), align 8
+; CHECK-NEXT: [[LD1:%.*]] = load i64, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @src64, i32 0, i64 1), align 8
+; CHECK-NEXT: [[BSWAP0:%.*]] = call i64 @llvm.bswap.i64(i64 [[LD0]])
+; CHECK-NEXT: [[BSWAP1:%.*]] = call i64 @llvm.bswap.i64(i64 [[LD1]])
+; CHECK-NEXT: store i64 [[BSWAP0]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i32 0, i64 0), align 8
+; CHECK-NEXT: store i64 [[BSWAP1]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i32 0, i64 1), align 8
+; CHECK-NEXT: ret void
+;
+ %ld0 = load i64, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @src64, i32 0, i64 0), align 8
+ %ld1 = load i64, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @src64, i32 0, i64 1), align 8
+ %bswap0 = call i64 @llvm.bswap.i64(i64 %ld0)
+ %bswap1 = call i64 @llvm.bswap.i64(i64 %ld1)
+ store i64 %bswap0, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i32 0, i64 0), align 8
+ store i64 %bswap1, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i32 0, i64 1), align 8
+ ret void
+}
+
+define void @bswap_4i64() #0 {
+; CHECK-LABEL: @bswap_4i64(
+; CHECK-NEXT: [[LD0:%.*]] = load i64, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @src64, i64 0, i64 0), align 4
+; CHECK-NEXT: [[LD1:%.*]] = load i64, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @src64, i64 0, i64 1), align 4
+; CHECK-NEXT: [[LD2:%.*]] = load i64, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @src64, i64 0, i64 2), align 4
+; CHECK-NEXT: [[LD3:%.*]] = load i64, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @src64, i64 0, i64 3), align 4
+; CHECK-NEXT: [[BSWAP0:%.*]] = call i64 @llvm.bswap.i64(i64 [[LD0]])
+; CHECK-NEXT: [[BSWAP1:%.*]] = call i64 @llvm.bswap.i64(i64 [[LD1]])
+; CHECK-NEXT: [[BSWAP2:%.*]] = call i64 @llvm.bswap.i64(i64 [[LD2]])
+; CHECK-NEXT: [[BSWAP3:%.*]] = call i64 @llvm.bswap.i64(i64 [[LD3]])
+; CHECK-NEXT: store i64 [[BSWAP0]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 0), align 4
+; CHECK-NEXT: store i64 [[BSWAP1]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 1), align 4
+; CHECK-NEXT: store i64 [[BSWAP2]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 2), align 4
+; CHECK-NEXT: store i64 [[BSWAP3]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 3), align 4
+; CHECK-NEXT: ret void
+;
+ %ld0 = load i64, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @src64, i64 0, i64 0), align 4
+ %ld1 = load i64, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @src64, i64 0, i64 1), align 4
+ %ld2 = load i64, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @src64, i64 0, i64 2), align 4
+ %ld3 = load i64, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @src64, i64 0, i64 3), align 4
+ %bswap0 = call i64 @llvm.bswap.i64(i64 %ld0)
+ %bswap1 = call i64 @llvm.bswap.i64(i64 %ld1)
+ %bswap2 = call i64 @llvm.bswap.i64(i64 %ld2)
+ %bswap3 = call i64 @llvm.bswap.i64(i64 %ld3)
+ store i64 %bswap0, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 0), align 4
+ store i64 %bswap1, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 1), align 4
+ store i64 %bswap2, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 2), align 4
+ store i64 %bswap3, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 3), align 4
+ ret void
+}
+
+define void @bswap_4i32() #0 {
+; CHECK-LABEL: @bswap_4i32(
+; CHECK-NEXT: [[LD0:%.*]] = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 0), align 4
+; CHECK-NEXT: [[LD1:%.*]] = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 1), align 4
+; CHECK-NEXT: [[LD2:%.*]] = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 2), align 4
+; CHECK-NEXT: [[LD3:%.*]] = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 3), align 4
+; CHECK-NEXT: [[BSWAP0:%.*]] = call i32 @llvm.bswap.i32(i32 [[LD0]])
+; CHECK-NEXT: [[BSWAP1:%.*]] = call i32 @llvm.bswap.i32(i32 [[LD1]])
+; CHECK-NEXT: [[BSWAP2:%.*]] = call i32 @llvm.bswap.i32(i32 [[LD2]])
+; CHECK-NEXT: [[BSWAP3:%.*]] = call i32 @llvm.bswap.i32(i32 [[LD3]])
+; CHECK-NEXT: store i32 [[BSWAP0]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 0), align 4
+; CHECK-NEXT: store i32 [[BSWAP1]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 1), align 4
+; CHECK-NEXT: store i32 [[BSWAP2]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 2), align 4
+; CHECK-NEXT: store i32 [[BSWAP3]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 3), align 4
+; CHECK-NEXT: ret void
+;
+ %ld0 = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 0), align 4
+ %ld1 = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 1), align 4
+ %ld2 = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 2), align 4
+ %ld3 = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 3), align 4
+ %bswap0 = call i32 @llvm.bswap.i32(i32 %ld0)
+ %bswap1 = call i32 @llvm.bswap.i32(i32 %ld1)
+ %bswap2 = call i32 @llvm.bswap.i32(i32 %ld2)
+ %bswap3 = call i32 @llvm.bswap.i32(i32 %ld3)
+ store i32 %bswap0, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 0), align 4
+ store i32 %bswap1, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 1), align 4
+ store i32 %bswap2, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 2), align 4
+ store i32 %bswap3, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 3), align 4
+ ret void
+}
+
+define void @bswap_8i32() #0 {
+; CHECK-LABEL: @bswap_8i32(
+; CHECK-NEXT: [[LD0:%.*]] = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 0), align 2
+; CHECK-NEXT: [[LD1:%.*]] = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 1), align 2
+; CHECK-NEXT: [[LD2:%.*]] = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 2), align 2
+; CHECK-NEXT: [[LD3:%.*]] = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 3), align 2
+; CHECK-NEXT: [[LD4:%.*]] = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 4), align 2
+; CHECK-NEXT: [[LD5:%.*]] = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 5), align 2
+; CHECK-NEXT: [[LD6:%.*]] = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 6), align 2
+; CHECK-NEXT: [[LD7:%.*]] = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 7), align 2
+; CHECK-NEXT: [[BSWAP0:%.*]] = call i32 @llvm.bswap.i32(i32 [[LD0]])
+; CHECK-NEXT: [[BSWAP1:%.*]] = call i32 @llvm.bswap.i32(i32 [[LD1]])
+; CHECK-NEXT: [[BSWAP2:%.*]] = call i32 @llvm.bswap.i32(i32 [[LD2]])
+; CHECK-NEXT: [[BSWAP3:%.*]] = call i32 @llvm.bswap.i32(i32 [[LD3]])
+; CHECK-NEXT: [[BSWAP4:%.*]] = call i32 @llvm.bswap.i32(i32 [[LD4]])
+; CHECK-NEXT: [[BSWAP5:%.*]] = call i32 @llvm.bswap.i32(i32 [[LD5]])
+; CHECK-NEXT: [[BSWAP6:%.*]] = call i32 @llvm.bswap.i32(i32 [[LD6]])
+; CHECK-NEXT: [[BSWAP7:%.*]] = call i32 @llvm.bswap.i32(i32 [[LD7]])
+; CHECK-NEXT: store i32 [[BSWAP0]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 0), align 2
+; CHECK-NEXT: store i32 [[BSWAP1]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 1), align 2
+; CHECK-NEXT: store i32 [[BSWAP2]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 2), align 2
+; CHECK-NEXT: store i32 [[BSWAP3]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 3), align 2
+; CHECK-NEXT: store i32 [[BSWAP4]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 4), align 2
+; CHECK-NEXT: store i32 [[BSWAP5]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 5), align 2
+; CHECK-NEXT: store i32 [[BSWAP6]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 6), align 2
+; CHECK-NEXT: store i32 [[BSWAP7]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 7), align 2
+; CHECK-NEXT: ret void
+;
+ %ld0 = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 0), align 2
+ %ld1 = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 1), align 2
+ %ld2 = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 2), align 2
+ %ld3 = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 3), align 2
+ %ld4 = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 4), align 2
+ %ld5 = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 5), align 2
+ %ld6 = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 6), align 2
+ %ld7 = load i32, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 7), align 2
+ %bswap0 = call i32 @llvm.bswap.i32(i32 %ld0)
+ %bswap1 = call i32 @llvm.bswap.i32(i32 %ld1)
+ %bswap2 = call i32 @llvm.bswap.i32(i32 %ld2)
+ %bswap3 = call i32 @llvm.bswap.i32(i32 %ld3)
+ %bswap4 = call i32 @llvm.bswap.i32(i32 %ld4)
+ %bswap5 = call i32 @llvm.bswap.i32(i32 %ld5)
+ %bswap6 = call i32 @llvm.bswap.i32(i32 %ld6)
+ %bswap7 = call i32 @llvm.bswap.i32(i32 %ld7)
+ store i32 %bswap0, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 0), align 2
+ store i32 %bswap1, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 1), align 2
+ store i32 %bswap2, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 2), align 2
+ store i32 %bswap3, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 3), align 2
+ store i32 %bswap4, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 4), align 2
+ store i32 %bswap5, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 5), align 2
+ store i32 %bswap6, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 6), align 2
+ store i32 %bswap7, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 7), align 2
+ ret void
+}
+
+define void @bswap_8i16() #0 {
+; CHECK-LABEL: @bswap_8i16(
+; CHECK-NEXT: [[LD0:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 0), align 2
+; CHECK-NEXT: [[LD1:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 1), align 2
+; CHECK-NEXT: [[LD2:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 2), align 2
+; CHECK-NEXT: [[LD3:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 3), align 2
+; CHECK-NEXT: [[LD4:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 4), align 2
+; CHECK-NEXT: [[LD5:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 5), align 2
+; CHECK-NEXT: [[LD6:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 6), align 2
+; CHECK-NEXT: [[LD7:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 7), align 2
+; CHECK-NEXT: [[BSWAP0:%.*]] = call i16 @llvm.bswap.i16(i16 [[LD0]])
+; CHECK-NEXT: [[BSWAP1:%.*]] = call i16 @llvm.bswap.i16(i16 [[LD1]])
+; CHECK-NEXT: [[BSWAP2:%.*]] = call i16 @llvm.bswap.i16(i16 [[LD2]])
+; CHECK-NEXT: [[BSWAP3:%.*]] = call i16 @llvm.bswap.i16(i16 [[LD3]])
+; CHECK-NEXT: [[BSWAP4:%.*]] = call i16 @llvm.bswap.i16(i16 [[LD4]])
+; CHECK-NEXT: [[BSWAP5:%.*]] = call i16 @llvm.bswap.i16(i16 [[LD5]])
+; CHECK-NEXT: [[BSWAP6:%.*]] = call i16 @llvm.bswap.i16(i16 [[LD6]])
+; CHECK-NEXT: [[BSWAP7:%.*]] = call i16 @llvm.bswap.i16(i16 [[LD7]])
+; CHECK-NEXT: store i16 [[BSWAP0]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 0), align 2
+; CHECK-NEXT: store i16 [[BSWAP1]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 1), align 2
+; CHECK-NEXT: store i16 [[BSWAP2]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 2), align 2
+; CHECK-NEXT: store i16 [[BSWAP3]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 3), align 2
+; CHECK-NEXT: store i16 [[BSWAP4]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 4), align 2
+; CHECK-NEXT: store i16 [[BSWAP5]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 5), align 2
+; CHECK-NEXT: store i16 [[BSWAP6]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 6), align 2
+; CHECK-NEXT: store i16 [[BSWAP7]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 7), align 2
+; CHECK-NEXT: ret void
+;
+ %ld0 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 0), align 2
+ %ld1 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 1), align 2
+ %ld2 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 2), align 2
+ %ld3 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 3), align 2
+ %ld4 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 4), align 2
+ %ld5 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 5), align 2
+ %ld6 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 6), align 2
+ %ld7 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 7), align 2
+ %bswap0 = call i16 @llvm.bswap.i16(i16 %ld0)
+ %bswap1 = call i16 @llvm.bswap.i16(i16 %ld1)
+ %bswap2 = call i16 @llvm.bswap.i16(i16 %ld2)
+ %bswap3 = call i16 @llvm.bswap.i16(i16 %ld3)
+ %bswap4 = call i16 @llvm.bswap.i16(i16 %ld4)
+ %bswap5 = call i16 @llvm.bswap.i16(i16 %ld5)
+ %bswap6 = call i16 @llvm.bswap.i16(i16 %ld6)
+ %bswap7 = call i16 @llvm.bswap.i16(i16 %ld7)
+ store i16 %bswap0, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 0), align 2
+ store i16 %bswap1, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 1), align 2
+ store i16 %bswap2, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 2), align 2
+ store i16 %bswap3, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 3), align 2
+ store i16 %bswap4, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 4), align 2
+ store i16 %bswap5, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 5), align 2
+ store i16 %bswap6, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 6), align 2
+ store i16 %bswap7, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 7), align 2
+ ret void
+}
+
+define void @bswap_16i16() #0 {
+; CHECK-LABEL: @bswap_16i16(
+; CHECK-NEXT: [[LD0:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 0), align 2
+; CHECK-NEXT: [[LD1:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 1), align 2
+; CHECK-NEXT: [[LD2:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 2), align 2
+; CHECK-NEXT: [[LD3:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 3), align 2
+; CHECK-NEXT: [[LD4:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 4), align 2
+; CHECK-NEXT: [[LD5:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 5), align 2
+; CHECK-NEXT: [[LD6:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 6), align 2
+; CHECK-NEXT: [[LD7:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 7), align 2
+; CHECK-NEXT: [[LD8:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 8), align 2
+; CHECK-NEXT: [[LD9:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 9), align 2
+; CHECK-NEXT: [[LD10:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 10), align 2
+; CHECK-NEXT: [[LD11:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 11), align 2
+; CHECK-NEXT: [[LD12:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 12), align 2
+; CHECK-NEXT: [[LD13:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 13), align 2
+; CHECK-NEXT: [[LD14:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 14), align 2
+; CHECK-NEXT: [[LD15:%.*]] = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 15), align 2
+; CHECK-NEXT: [[BSWAP0:%.*]] = call i16 @llvm.bswap.i16(i16 [[LD0]])
+; CHECK-NEXT: [[BSWAP1:%.*]] = call i16 @llvm.bswap.i16(i16 [[LD1]])
+; CHECK-NEXT: [[BSWAP2:%.*]] = call i16 @llvm.bswap.i16(i16 [[LD2]])
+; CHECK-NEXT: [[BSWAP3:%.*]] = call i16 @llvm.bswap.i16(i16 [[LD3]])
+; CHECK-NEXT: [[BSWAP4:%.*]] = call i16 @llvm.bswap.i16(i16 [[LD4]])
+; CHECK-NEXT: [[BSWAP5:%.*]] = call i16 @llvm.bswap.i16(i16 [[LD5]])
+; CHECK-NEXT: [[BSWAP6:%.*]] = call i16 @llvm.bswap.i16(i16 [[LD6]])
+; CHECK-NEXT: [[BSWAP7:%.*]] = call i16 @llvm.bswap.i16(i16 [[LD7]])
+; CHECK-NEXT: [[BSWAP8:%.*]] = call i16 @llvm.bswap.i16(i16 [[LD8]])
+; CHECK-NEXT: [[BSWAP9:%.*]] = call i16 @llvm.bswap.i16(i16 [[LD9]])
+; CHECK-NEXT: [[BSWAP10:%.*]] = call i16 @llvm.bswap.i16(i16 [[LD10]])
+; CHECK-NEXT: [[BSWAP11:%.*]] = call i16 @llvm.bswap.i16(i16 [[LD11]])
+; CHECK-NEXT: [[BSWAP12:%.*]] = call i16 @llvm.bswap.i16(i16 [[LD12]])
+; CHECK-NEXT: [[BSWAP13:%.*]] = call i16 @llvm.bswap.i16(i16 [[LD13]])
+; CHECK-NEXT: [[BSWAP14:%.*]] = call i16 @llvm.bswap.i16(i16 [[LD14]])
+; CHECK-NEXT: [[BSWAP15:%.*]] = call i16 @llvm.bswap.i16(i16 [[LD15]])
+; CHECK-NEXT: store i16 [[BSWAP0]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 0), align 2
+; CHECK-NEXT: store i16 [[BSWAP1]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 1), align 2
+; CHECK-NEXT: store i16 [[BSWAP2]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 2), align 2
+; CHECK-NEXT: store i16 [[BSWAP3]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 3), align 2
+; CHECK-NEXT: store i16 [[BSWAP4]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 4), align 2
+; CHECK-NEXT: store i16 [[BSWAP5]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 5), align 2
+; CHECK-NEXT: store i16 [[BSWAP6]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 6), align 2
+; CHECK-NEXT: store i16 [[BSWAP7]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 7), align 2
+; CHECK-NEXT: store i16 [[BSWAP8]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 8), align 2
+; CHECK-NEXT: store i16 [[BSWAP9]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 9), align 2
+; CHECK-NEXT: store i16 [[BSWAP10]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 10), align 2
+; CHECK-NEXT: store i16 [[BSWAP11]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 11), align 2
+; CHECK-NEXT: store i16 [[BSWAP12]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 12), align 2
+; CHECK-NEXT: store i16 [[BSWAP13]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 13), align 2
+; CHECK-NEXT: store i16 [[BSWAP14]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 14), align 2
+; CHECK-NEXT: store i16 [[BSWAP15]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 15), align 2
+; CHECK-NEXT: ret void
+;
+ %ld0 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 0), align 2
+ %ld1 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 1), align 2
+ %ld2 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 2), align 2
+ %ld3 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 3), align 2
+ %ld4 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 4), align 2
+ %ld5 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 5), align 2
+ %ld6 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 6), align 2
+ %ld7 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 7), align 2
+ %ld8 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 8), align 2
+ %ld9 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 9), align 2
+ %ld10 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 10), align 2
+ %ld11 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 11), align 2
+ %ld12 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 12), align 2
+ %ld13 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 13), align 2
+ %ld14 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 14), align 2
+ %ld15 = load i16, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 15), align 2
+ %bswap0 = call i16 @llvm.bswap.i16(i16 %ld0)
+ %bswap1 = call i16 @llvm.bswap.i16(i16 %ld1)
+ %bswap2 = call i16 @llvm.bswap.i16(i16 %ld2)
+ %bswap3 = call i16 @llvm.bswap.i16(i16 %ld3)
+ %bswap4 = call i16 @llvm.bswap.i16(i16 %ld4)
+ %bswap5 = call i16 @llvm.bswap.i16(i16 %ld5)
+ %bswap6 = call i16 @llvm.bswap.i16(i16 %ld6)
+ %bswap7 = call i16 @llvm.bswap.i16(i16 %ld7)
+ %bswap8 = call i16 @llvm.bswap.i16(i16 %ld8)
+ %bswap9 = call i16 @llvm.bswap.i16(i16 %ld9)
+ %bswap10 = call i16 @llvm.bswap.i16(i16 %ld10)
+ %bswap11 = call i16 @llvm.bswap.i16(i16 %ld11)
+ %bswap12 = call i16 @llvm.bswap.i16(i16 %ld12)
+ %bswap13 = call i16 @llvm.bswap.i16(i16 %ld13)
+ %bswap14 = call i16 @llvm.bswap.i16(i16 %ld14)
+ %bswap15 = call i16 @llvm.bswap.i16(i16 %ld15)
+ store i16 %bswap0 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 0), align 2
+ store i16 %bswap1 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 1), align 2
+ store i16 %bswap2 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 2), align 2
+ store i16 %bswap3 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 3), align 2
+ store i16 %bswap4 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 4), align 2
+ store i16 %bswap5 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 5), align 2
+ store i16 %bswap6 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 6), align 2
+ store i16 %bswap7 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 7), align 2
+ store i16 %bswap8 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 8), align 2
+ store i16 %bswap9 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 9), align 2
+ store i16 %bswap10, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 10), align 2
+ store i16 %bswap11, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 11), align 2
+ store i16 %bswap12, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 12), align 2
+ store i16 %bswap13, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 13), align 2
+ store i16 %bswap14, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 14), align 2
+ store i16 %bswap15, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 15), align 2
+ ret void
+}
+
+attributes #0 = { nounwind }