[llvm] r268800 - [SLPVectorizer][X86] Added CTPOP/CTLZ/CTTZ vectorization tests
Simon Pilgrim via llvm-commits
llvm-commits at lists.llvm.org
Fri May 6 14:33:02 PDT 2016
Author: rksimon
Date: Fri May 6 16:33:01 2016
New Revision: 268800
URL: http://llvm.org/viewvc/llvm-project?rev=268800&view=rev
Log:
[SLPVectorizer][X86] Added CTPOP/CTLZ/CTTZ vectorization tests
Added:
llvm/trunk/test/Transforms/SLPVectorizer/X86/ctlz.ll
llvm/trunk/test/Transforms/SLPVectorizer/X86/ctpop.ll
llvm/trunk/test/Transforms/SLPVectorizer/X86/cttz.ll
Added: llvm/trunk/test/Transforms/SLPVectorizer/X86/ctlz.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Transforms/SLPVectorizer/X86/ctlz.ll?rev=268800&view=auto
==============================================================================
--- llvm/trunk/test/Transforms/SLPVectorizer/X86/ctlz.ll (added)
+++ llvm/trunk/test/Transforms/SLPVectorizer/X86/ctlz.ll Fri May 6 16:33:01 2016
@@ -0,0 +1,1222 @@
+; 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"
+
+ at src64 = common global [4 x i64] zeroinitializer, align 32
+ at dst64 = common global [4 x i64] zeroinitializer, align 32
+ at src32 = common global [8 x i32] zeroinitializer, align 32
+ at dst32 = common global [8 x i32] zeroinitializer, align 32
+ at src16 = common global [16 x i16] zeroinitializer, align 32
+ at dst16 = common global [16 x i16] zeroinitializer, align 32
+ at src8 = common global [32 x i8] zeroinitializer, align 32
+ at dst8 = common global [32 x i8] zeroinitializer, align 32
+
+declare i64 @llvm.ctlz.i64(i64, i1)
+declare i32 @llvm.ctlz.i32(i32, i1)
+declare i16 @llvm.ctlz.i16(i16, i1)
+declare i8 @llvm.ctlz.i8(i8, i1)
+
+;
+; CTLZ
+;
+
+define void @ctlz_2i64() #0 {
+; CHECK-LABEL: @ctlz_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: [[CTLZ0:%.*]] = call i64 @llvm.ctlz.i64(i64 [[LD0]], i1 false)
+; CHECK-NEXT: [[CTLZ1:%.*]] = call i64 @llvm.ctlz.i64(i64 [[LD1]], i1 false)
+; CHECK-NEXT: store i64 [[CTLZ0]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i32 0, i64 0), align 8
+; CHECK-NEXT: store i64 [[CTLZ1]], 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
+ %ctlz0 = call i64 @llvm.ctlz.i64(i64 %ld0, i1 0)
+ %ctlz1 = call i64 @llvm.ctlz.i64(i64 %ld1, i1 0)
+ store i64 %ctlz0, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i32 0, i64 0), align 8
+ store i64 %ctlz1, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i32 0, i64 1), align 8
+ ret void
+}
+
+define void @ctlz_4i64() #0 {
+; CHECK-LABEL: @ctlz_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: [[CTLZ0:%.*]] = call i64 @llvm.ctlz.i64(i64 [[LD0]], i1 false)
+; CHECK-NEXT: [[CTLZ1:%.*]] = call i64 @llvm.ctlz.i64(i64 [[LD1]], i1 false)
+; CHECK-NEXT: [[CTLZ2:%.*]] = call i64 @llvm.ctlz.i64(i64 [[LD2]], i1 false)
+; CHECK-NEXT: [[CTLZ3:%.*]] = call i64 @llvm.ctlz.i64(i64 [[LD3]], i1 false)
+; CHECK-NEXT: store i64 [[CTLZ0]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 0), align 4
+; CHECK-NEXT: store i64 [[CTLZ1]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 1), align 4
+; CHECK-NEXT: store i64 [[CTLZ2]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 2), align 4
+; CHECK-NEXT: store i64 [[CTLZ3]], 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
+ %ctlz0 = call i64 @llvm.ctlz.i64(i64 %ld0, i1 0)
+ %ctlz1 = call i64 @llvm.ctlz.i64(i64 %ld1, i1 0)
+ %ctlz2 = call i64 @llvm.ctlz.i64(i64 %ld2, i1 0)
+ %ctlz3 = call i64 @llvm.ctlz.i64(i64 %ld3, i1 0)
+ store i64 %ctlz0, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 0), align 4
+ store i64 %ctlz1, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 1), align 4
+ store i64 %ctlz2, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 2), align 4
+ store i64 %ctlz3, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 3), align 4
+ ret void
+}
+
+define void @ctlz_4i32() #0 {
+; CHECK-LABEL: @ctlz_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: [[CTLZ0:%.*]] = call i32 @llvm.ctlz.i32(i32 [[LD0]], i1 false)
+; CHECK-NEXT: [[CTLZ1:%.*]] = call i32 @llvm.ctlz.i32(i32 [[LD1]], i1 false)
+; CHECK-NEXT: [[CTLZ2:%.*]] = call i32 @llvm.ctlz.i32(i32 [[LD2]], i1 false)
+; CHECK-NEXT: [[CTLZ3:%.*]] = call i32 @llvm.ctlz.i32(i32 [[LD3]], i1 false)
+; CHECK-NEXT: store i32 [[CTLZ0]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 0), align 4
+; CHECK-NEXT: store i32 [[CTLZ1]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 1), align 4
+; CHECK-NEXT: store i32 [[CTLZ2]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 2), align 4
+; CHECK-NEXT: store i32 [[CTLZ3]], 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
+ %ctlz0 = call i32 @llvm.ctlz.i32(i32 %ld0, i1 0)
+ %ctlz1 = call i32 @llvm.ctlz.i32(i32 %ld1, i1 0)
+ %ctlz2 = call i32 @llvm.ctlz.i32(i32 %ld2, i1 0)
+ %ctlz3 = call i32 @llvm.ctlz.i32(i32 %ld3, i1 0)
+ store i32 %ctlz0, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 0), align 4
+ store i32 %ctlz1, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 1), align 4
+ store i32 %ctlz2, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 2), align 4
+ store i32 %ctlz3, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 3), align 4
+ ret void
+}
+
+define void @ctlz_8i32() #0 {
+; CHECK-LABEL: @ctlz_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: [[CTLZ0:%.*]] = call i32 @llvm.ctlz.i32(i32 [[LD0]], i1 false)
+; CHECK-NEXT: [[CTLZ1:%.*]] = call i32 @llvm.ctlz.i32(i32 [[LD1]], i1 false)
+; CHECK-NEXT: [[CTLZ2:%.*]] = call i32 @llvm.ctlz.i32(i32 [[LD2]], i1 false)
+; CHECK-NEXT: [[CTLZ3:%.*]] = call i32 @llvm.ctlz.i32(i32 [[LD3]], i1 false)
+; CHECK-NEXT: [[CTLZ4:%.*]] = call i32 @llvm.ctlz.i32(i32 [[LD4]], i1 false)
+; CHECK-NEXT: [[CTLZ5:%.*]] = call i32 @llvm.ctlz.i32(i32 [[LD5]], i1 false)
+; CHECK-NEXT: [[CTLZ6:%.*]] = call i32 @llvm.ctlz.i32(i32 [[LD6]], i1 false)
+; CHECK-NEXT: [[CTLZ7:%.*]] = call i32 @llvm.ctlz.i32(i32 [[LD7]], i1 false)
+; CHECK-NEXT: store i32 [[CTLZ0]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 0), align 2
+; CHECK-NEXT: store i32 [[CTLZ1]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 1), align 2
+; CHECK-NEXT: store i32 [[CTLZ2]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 2), align 2
+; CHECK-NEXT: store i32 [[CTLZ3]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 3), align 2
+; CHECK-NEXT: store i32 [[CTLZ4]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 4), align 2
+; CHECK-NEXT: store i32 [[CTLZ5]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 5), align 2
+; CHECK-NEXT: store i32 [[CTLZ6]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 6), align 2
+; CHECK-NEXT: store i32 [[CTLZ7]], 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
+ %ctlz0 = call i32 @llvm.ctlz.i32(i32 %ld0, i1 0)
+ %ctlz1 = call i32 @llvm.ctlz.i32(i32 %ld1, i1 0)
+ %ctlz2 = call i32 @llvm.ctlz.i32(i32 %ld2, i1 0)
+ %ctlz3 = call i32 @llvm.ctlz.i32(i32 %ld3, i1 0)
+ %ctlz4 = call i32 @llvm.ctlz.i32(i32 %ld4, i1 0)
+ %ctlz5 = call i32 @llvm.ctlz.i32(i32 %ld5, i1 0)
+ %ctlz6 = call i32 @llvm.ctlz.i32(i32 %ld6, i1 0)
+ %ctlz7 = call i32 @llvm.ctlz.i32(i32 %ld7, i1 0)
+ store i32 %ctlz0, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 0), align 2
+ store i32 %ctlz1, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 1), align 2
+ store i32 %ctlz2, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 2), align 2
+ store i32 %ctlz3, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 3), align 2
+ store i32 %ctlz4, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 4), align 2
+ store i32 %ctlz5, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 5), align 2
+ store i32 %ctlz6, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 6), align 2
+ store i32 %ctlz7, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 7), align 2
+ ret void
+}
+
+define void @ctlz_8i16() #0 {
+; CHECK-LABEL: @ctlz_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: [[CTLZ0:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD0]], i1 false)
+; CHECK-NEXT: [[CTLZ1:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD1]], i1 false)
+; CHECK-NEXT: [[CTLZ2:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD2]], i1 false)
+; CHECK-NEXT: [[CTLZ3:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD3]], i1 false)
+; CHECK-NEXT: [[CTLZ4:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD4]], i1 false)
+; CHECK-NEXT: [[CTLZ5:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD5]], i1 false)
+; CHECK-NEXT: [[CTLZ6:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD6]], i1 false)
+; CHECK-NEXT: [[CTLZ7:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD7]], i1 false)
+; CHECK-NEXT: store i16 [[CTLZ0]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 0), align 2
+; CHECK-NEXT: store i16 [[CTLZ1]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 1), align 2
+; CHECK-NEXT: store i16 [[CTLZ2]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 2), align 2
+; CHECK-NEXT: store i16 [[CTLZ3]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 3), align 2
+; CHECK-NEXT: store i16 [[CTLZ4]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 4), align 2
+; CHECK-NEXT: store i16 [[CTLZ5]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 5), align 2
+; CHECK-NEXT: store i16 [[CTLZ6]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 6), align 2
+; CHECK-NEXT: store i16 [[CTLZ7]], 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
+ %ctlz0 = call i16 @llvm.ctlz.i16(i16 %ld0, i1 0)
+ %ctlz1 = call i16 @llvm.ctlz.i16(i16 %ld1, i1 0)
+ %ctlz2 = call i16 @llvm.ctlz.i16(i16 %ld2, i1 0)
+ %ctlz3 = call i16 @llvm.ctlz.i16(i16 %ld3, i1 0)
+ %ctlz4 = call i16 @llvm.ctlz.i16(i16 %ld4, i1 0)
+ %ctlz5 = call i16 @llvm.ctlz.i16(i16 %ld5, i1 0)
+ %ctlz6 = call i16 @llvm.ctlz.i16(i16 %ld6, i1 0)
+ %ctlz7 = call i16 @llvm.ctlz.i16(i16 %ld7, i1 0)
+ store i16 %ctlz0, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 0), align 2
+ store i16 %ctlz1, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 1), align 2
+ store i16 %ctlz2, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 2), align 2
+ store i16 %ctlz3, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 3), align 2
+ store i16 %ctlz4, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 4), align 2
+ store i16 %ctlz5, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 5), align 2
+ store i16 %ctlz6, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 6), align 2
+ store i16 %ctlz7, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 7), align 2
+ ret void
+}
+
+define void @ctlz_16i16() #0 {
+; CHECK-LABEL: @ctlz_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: [[CTLZ0:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD0]], i1 false)
+; CHECK-NEXT: [[CTLZ1:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD1]], i1 false)
+; CHECK-NEXT: [[CTLZ2:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD2]], i1 false)
+; CHECK-NEXT: [[CTLZ3:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD3]], i1 false)
+; CHECK-NEXT: [[CTLZ4:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD4]], i1 false)
+; CHECK-NEXT: [[CTLZ5:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD5]], i1 false)
+; CHECK-NEXT: [[CTLZ6:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD6]], i1 false)
+; CHECK-NEXT: [[CTLZ7:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD7]], i1 false)
+; CHECK-NEXT: [[CTLZ8:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD8]], i1 false)
+; CHECK-NEXT: [[CTLZ9:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD9]], i1 false)
+; CHECK-NEXT: [[CTLZ10:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD10]], i1 false)
+; CHECK-NEXT: [[CTLZ11:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD11]], i1 false)
+; CHECK-NEXT: [[CTLZ12:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD12]], i1 false)
+; CHECK-NEXT: [[CTLZ13:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD13]], i1 false)
+; CHECK-NEXT: [[CTLZ14:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD14]], i1 false)
+; CHECK-NEXT: [[CTLZ15:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD15]], i1 false)
+; CHECK-NEXT: store i16 [[CTLZ0]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 0), align 2
+; CHECK-NEXT: store i16 [[CTLZ1]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 1), align 2
+; CHECK-NEXT: store i16 [[CTLZ2]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 2), align 2
+; CHECK-NEXT: store i16 [[CTLZ3]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 3), align 2
+; CHECK-NEXT: store i16 [[CTLZ4]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 4), align 2
+; CHECK-NEXT: store i16 [[CTLZ5]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 5), align 2
+; CHECK-NEXT: store i16 [[CTLZ6]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 6), align 2
+; CHECK-NEXT: store i16 [[CTLZ7]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 7), align 2
+; CHECK-NEXT: store i16 [[CTLZ8]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 8), align 2
+; CHECK-NEXT: store i16 [[CTLZ9]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 9), align 2
+; CHECK-NEXT: store i16 [[CTLZ10]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 10), align 2
+; CHECK-NEXT: store i16 [[CTLZ11]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 11), align 2
+; CHECK-NEXT: store i16 [[CTLZ12]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 12), align 2
+; CHECK-NEXT: store i16 [[CTLZ13]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 13), align 2
+; CHECK-NEXT: store i16 [[CTLZ14]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 14), align 2
+; CHECK-NEXT: store i16 [[CTLZ15]], 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
+ %ctlz0 = call i16 @llvm.ctlz.i16(i16 %ld0, i1 0)
+ %ctlz1 = call i16 @llvm.ctlz.i16(i16 %ld1, i1 0)
+ %ctlz2 = call i16 @llvm.ctlz.i16(i16 %ld2, i1 0)
+ %ctlz3 = call i16 @llvm.ctlz.i16(i16 %ld3, i1 0)
+ %ctlz4 = call i16 @llvm.ctlz.i16(i16 %ld4, i1 0)
+ %ctlz5 = call i16 @llvm.ctlz.i16(i16 %ld5, i1 0)
+ %ctlz6 = call i16 @llvm.ctlz.i16(i16 %ld6, i1 0)
+ %ctlz7 = call i16 @llvm.ctlz.i16(i16 %ld7, i1 0)
+ %ctlz8 = call i16 @llvm.ctlz.i16(i16 %ld8, i1 0)
+ %ctlz9 = call i16 @llvm.ctlz.i16(i16 %ld9, i1 0)
+ %ctlz10 = call i16 @llvm.ctlz.i16(i16 %ld10, i1 0)
+ %ctlz11 = call i16 @llvm.ctlz.i16(i16 %ld11, i1 0)
+ %ctlz12 = call i16 @llvm.ctlz.i16(i16 %ld12, i1 0)
+ %ctlz13 = call i16 @llvm.ctlz.i16(i16 %ld13, i1 0)
+ %ctlz14 = call i16 @llvm.ctlz.i16(i16 %ld14, i1 0)
+ %ctlz15 = call i16 @llvm.ctlz.i16(i16 %ld15, i1 0)
+ store i16 %ctlz0 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 0), align 2
+ store i16 %ctlz1 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 1), align 2
+ store i16 %ctlz2 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 2), align 2
+ store i16 %ctlz3 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 3), align 2
+ store i16 %ctlz4 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 4), align 2
+ store i16 %ctlz5 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 5), align 2
+ store i16 %ctlz6 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 6), align 2
+ store i16 %ctlz7 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 7), align 2
+ store i16 %ctlz8 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 8), align 2
+ store i16 %ctlz9 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 9), align 2
+ store i16 %ctlz10, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 10), align 2
+ store i16 %ctlz11, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 11), align 2
+ store i16 %ctlz12, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 12), align 2
+ store i16 %ctlz13, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 13), align 2
+ store i16 %ctlz14, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 14), align 2
+ store i16 %ctlz15, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 15), align 2
+ ret void
+}
+
+define void @ctlz_16i8() #0 {
+; CHECK-LABEL: @ctlz_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: [[CTLZ0:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD0]], i1 false)
+; CHECK-NEXT: [[CTLZ1:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD1]], i1 false)
+; CHECK-NEXT: [[CTLZ2:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD2]], i1 false)
+; CHECK-NEXT: [[CTLZ3:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD3]], i1 false)
+; CHECK-NEXT: [[CTLZ4:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD4]], i1 false)
+; CHECK-NEXT: [[CTLZ5:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD5]], i1 false)
+; CHECK-NEXT: [[CTLZ6:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD6]], i1 false)
+; CHECK-NEXT: [[CTLZ7:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD7]], i1 false)
+; CHECK-NEXT: [[CTLZ8:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD8]], i1 false)
+; CHECK-NEXT: [[CTLZ9:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD9]], i1 false)
+; CHECK-NEXT: [[CTLZ10:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD10]], i1 false)
+; CHECK-NEXT: [[CTLZ11:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD11]], i1 false)
+; CHECK-NEXT: [[CTLZ12:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD12]], i1 false)
+; CHECK-NEXT: [[CTLZ13:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD13]], i1 false)
+; CHECK-NEXT: [[CTLZ14:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD14]], i1 false)
+; CHECK-NEXT: [[CTLZ15:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD15]], i1 false)
+; CHECK-NEXT: store i8 [[CTLZ0]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 0), align 1
+; CHECK-NEXT: store i8 [[CTLZ1]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 1), align 1
+; CHECK-NEXT: store i8 [[CTLZ2]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 2), align 1
+; CHECK-NEXT: store i8 [[CTLZ3]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 3), align 1
+; CHECK-NEXT: store i8 [[CTLZ4]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 4), align 1
+; CHECK-NEXT: store i8 [[CTLZ5]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 5), align 1
+; CHECK-NEXT: store i8 [[CTLZ6]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 6), align 1
+; CHECK-NEXT: store i8 [[CTLZ7]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 7), align 1
+; CHECK-NEXT: store i8 [[CTLZ8]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 8), align 1
+; CHECK-NEXT: store i8 [[CTLZ9]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 9), align 1
+; CHECK-NEXT: store i8 [[CTLZ10]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 10), align 1
+; CHECK-NEXT: store i8 [[CTLZ11]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 11), align 1
+; CHECK-NEXT: store i8 [[CTLZ12]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 12), align 1
+; CHECK-NEXT: store i8 [[CTLZ13]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 13), align 1
+; CHECK-NEXT: store i8 [[CTLZ14]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 14), align 1
+; CHECK-NEXT: store i8 [[CTLZ15]], 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
+ %ctlz0 = call i8 @llvm.ctlz.i8(i8 %ld0, i1 0)
+ %ctlz1 = call i8 @llvm.ctlz.i8(i8 %ld1, i1 0)
+ %ctlz2 = call i8 @llvm.ctlz.i8(i8 %ld2, i1 0)
+ %ctlz3 = call i8 @llvm.ctlz.i8(i8 %ld3, i1 0)
+ %ctlz4 = call i8 @llvm.ctlz.i8(i8 %ld4, i1 0)
+ %ctlz5 = call i8 @llvm.ctlz.i8(i8 %ld5, i1 0)
+ %ctlz6 = call i8 @llvm.ctlz.i8(i8 %ld6, i1 0)
+ %ctlz7 = call i8 @llvm.ctlz.i8(i8 %ld7, i1 0)
+ %ctlz8 = call i8 @llvm.ctlz.i8(i8 %ld8, i1 0)
+ %ctlz9 = call i8 @llvm.ctlz.i8(i8 %ld9, i1 0)
+ %ctlz10 = call i8 @llvm.ctlz.i8(i8 %ld10, i1 0)
+ %ctlz11 = call i8 @llvm.ctlz.i8(i8 %ld11, i1 0)
+ %ctlz12 = call i8 @llvm.ctlz.i8(i8 %ld12, i1 0)
+ %ctlz13 = call i8 @llvm.ctlz.i8(i8 %ld13, i1 0)
+ %ctlz14 = call i8 @llvm.ctlz.i8(i8 %ld14, i1 0)
+ %ctlz15 = call i8 @llvm.ctlz.i8(i8 %ld15, i1 0)
+ store i8 %ctlz0 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 0), align 1
+ store i8 %ctlz1 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 1), align 1
+ store i8 %ctlz2 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 2), align 1
+ store i8 %ctlz3 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 3), align 1
+ store i8 %ctlz4 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 4), align 1
+ store i8 %ctlz5 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 5), align 1
+ store i8 %ctlz6 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 6), align 1
+ store i8 %ctlz7 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 7), align 1
+ store i8 %ctlz8 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 8), align 1
+ store i8 %ctlz9 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 9), align 1
+ store i8 %ctlz10, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 10), align 1
+ store i8 %ctlz11, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 11), align 1
+ store i8 %ctlz12, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 12), align 1
+ store i8 %ctlz13, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 13), align 1
+ store i8 %ctlz14, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 14), align 1
+ store i8 %ctlz15, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 15), align 1
+ ret void
+}
+
+define void @ctlz_32i8() #0 {
+; CHECK-LABEL: @ctlz_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: [[CTLZ0:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD0]], i1 false)
+; CHECK-NEXT: [[CTLZ1:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD1]], i1 false)
+; CHECK-NEXT: [[CTLZ2:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD2]], i1 false)
+; CHECK-NEXT: [[CTLZ3:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD3]], i1 false)
+; CHECK-NEXT: [[CTLZ4:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD4]], i1 false)
+; CHECK-NEXT: [[CTLZ5:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD5]], i1 false)
+; CHECK-NEXT: [[CTLZ6:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD6]], i1 false)
+; CHECK-NEXT: [[CTLZ7:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD7]], i1 false)
+; CHECK-NEXT: [[CTLZ8:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD8]], i1 false)
+; CHECK-NEXT: [[CTLZ9:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD9]], i1 false)
+; CHECK-NEXT: [[CTLZ10:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD10]], i1 false)
+; CHECK-NEXT: [[CTLZ11:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD11]], i1 false)
+; CHECK-NEXT: [[CTLZ12:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD12]], i1 false)
+; CHECK-NEXT: [[CTLZ13:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD13]], i1 false)
+; CHECK-NEXT: [[CTLZ14:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD14]], i1 false)
+; CHECK-NEXT: [[CTLZ15:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD15]], i1 false)
+; CHECK-NEXT: [[CTLZ16:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD16]], i1 false)
+; CHECK-NEXT: [[CTLZ17:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD17]], i1 false)
+; CHECK-NEXT: [[CTLZ18:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD18]], i1 false)
+; CHECK-NEXT: [[CTLZ19:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD19]], i1 false)
+; CHECK-NEXT: [[CTLZ20:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD20]], i1 false)
+; CHECK-NEXT: [[CTLZ21:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD21]], i1 false)
+; CHECK-NEXT: [[CTLZ22:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD22]], i1 false)
+; CHECK-NEXT: [[CTLZ23:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD23]], i1 false)
+; CHECK-NEXT: [[CTLZ24:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD24]], i1 false)
+; CHECK-NEXT: [[CTLZ25:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD25]], i1 false)
+; CHECK-NEXT: [[CTLZ26:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD26]], i1 false)
+; CHECK-NEXT: [[CTLZ27:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD27]], i1 false)
+; CHECK-NEXT: [[CTLZ28:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD28]], i1 false)
+; CHECK-NEXT: [[CTLZ29:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD29]], i1 false)
+; CHECK-NEXT: [[CTLZ30:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD30]], i1 false)
+; CHECK-NEXT: [[CTLZ31:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD31]], i1 false)
+; CHECK-NEXT: store i8 [[CTLZ0]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 0), align 1
+; CHECK-NEXT: store i8 [[CTLZ1]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 1), align 1
+; CHECK-NEXT: store i8 [[CTLZ2]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 2), align 1
+; CHECK-NEXT: store i8 [[CTLZ3]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 3), align 1
+; CHECK-NEXT: store i8 [[CTLZ4]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 4), align 1
+; CHECK-NEXT: store i8 [[CTLZ5]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 5), align 1
+; CHECK-NEXT: store i8 [[CTLZ6]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 6), align 1
+; CHECK-NEXT: store i8 [[CTLZ7]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 7), align 1
+; CHECK-NEXT: store i8 [[CTLZ8]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 8), align 1
+; CHECK-NEXT: store i8 [[CTLZ9]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 9), align 1
+; CHECK-NEXT: store i8 [[CTLZ10]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 10), align 1
+; CHECK-NEXT: store i8 [[CTLZ11]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 11), align 1
+; CHECK-NEXT: store i8 [[CTLZ12]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 12), align 1
+; CHECK-NEXT: store i8 [[CTLZ13]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 13), align 1
+; CHECK-NEXT: store i8 [[CTLZ14]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 14), align 1
+; CHECK-NEXT: store i8 [[CTLZ15]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 15), align 1
+; CHECK-NEXT: store i8 [[CTLZ16]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 16), align 1
+; CHECK-NEXT: store i8 [[CTLZ17]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 17), align 1
+; CHECK-NEXT: store i8 [[CTLZ18]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 18), align 1
+; CHECK-NEXT: store i8 [[CTLZ19]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 19), align 1
+; CHECK-NEXT: store i8 [[CTLZ20]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 20), align 1
+; CHECK-NEXT: store i8 [[CTLZ21]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 21), align 1
+; CHECK-NEXT: store i8 [[CTLZ22]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 22), align 1
+; CHECK-NEXT: store i8 [[CTLZ23]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 23), align 1
+; CHECK-NEXT: store i8 [[CTLZ24]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 24), align 1
+; CHECK-NEXT: store i8 [[CTLZ25]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 25), align 1
+; CHECK-NEXT: store i8 [[CTLZ26]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 26), align 1
+; CHECK-NEXT: store i8 [[CTLZ27]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 27), align 1
+; CHECK-NEXT: store i8 [[CTLZ28]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 28), align 1
+; CHECK-NEXT: store i8 [[CTLZ29]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 29), align 1
+; CHECK-NEXT: store i8 [[CTLZ30]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 30), align 1
+; CHECK-NEXT: store i8 [[CTLZ31]], 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
+ %ctlz0 = call i8 @llvm.ctlz.i8(i8 %ld0, i1 0)
+ %ctlz1 = call i8 @llvm.ctlz.i8(i8 %ld1, i1 0)
+ %ctlz2 = call i8 @llvm.ctlz.i8(i8 %ld2, i1 0)
+ %ctlz3 = call i8 @llvm.ctlz.i8(i8 %ld3, i1 0)
+ %ctlz4 = call i8 @llvm.ctlz.i8(i8 %ld4, i1 0)
+ %ctlz5 = call i8 @llvm.ctlz.i8(i8 %ld5, i1 0)
+ %ctlz6 = call i8 @llvm.ctlz.i8(i8 %ld6, i1 0)
+ %ctlz7 = call i8 @llvm.ctlz.i8(i8 %ld7, i1 0)
+ %ctlz8 = call i8 @llvm.ctlz.i8(i8 %ld8, i1 0)
+ %ctlz9 = call i8 @llvm.ctlz.i8(i8 %ld9, i1 0)
+ %ctlz10 = call i8 @llvm.ctlz.i8(i8 %ld10, i1 0)
+ %ctlz11 = call i8 @llvm.ctlz.i8(i8 %ld11, i1 0)
+ %ctlz12 = call i8 @llvm.ctlz.i8(i8 %ld12, i1 0)
+ %ctlz13 = call i8 @llvm.ctlz.i8(i8 %ld13, i1 0)
+ %ctlz14 = call i8 @llvm.ctlz.i8(i8 %ld14, i1 0)
+ %ctlz15 = call i8 @llvm.ctlz.i8(i8 %ld15, i1 0)
+ %ctlz16 = call i8 @llvm.ctlz.i8(i8 %ld16, i1 0)
+ %ctlz17 = call i8 @llvm.ctlz.i8(i8 %ld17, i1 0)
+ %ctlz18 = call i8 @llvm.ctlz.i8(i8 %ld18, i1 0)
+ %ctlz19 = call i8 @llvm.ctlz.i8(i8 %ld19, i1 0)
+ %ctlz20 = call i8 @llvm.ctlz.i8(i8 %ld20, i1 0)
+ %ctlz21 = call i8 @llvm.ctlz.i8(i8 %ld21, i1 0)
+ %ctlz22 = call i8 @llvm.ctlz.i8(i8 %ld22, i1 0)
+ %ctlz23 = call i8 @llvm.ctlz.i8(i8 %ld23, i1 0)
+ %ctlz24 = call i8 @llvm.ctlz.i8(i8 %ld24, i1 0)
+ %ctlz25 = call i8 @llvm.ctlz.i8(i8 %ld25, i1 0)
+ %ctlz26 = call i8 @llvm.ctlz.i8(i8 %ld26, i1 0)
+ %ctlz27 = call i8 @llvm.ctlz.i8(i8 %ld27, i1 0)
+ %ctlz28 = call i8 @llvm.ctlz.i8(i8 %ld28, i1 0)
+ %ctlz29 = call i8 @llvm.ctlz.i8(i8 %ld29, i1 0)
+ %ctlz30 = call i8 @llvm.ctlz.i8(i8 %ld30, i1 0)
+ %ctlz31 = call i8 @llvm.ctlz.i8(i8 %ld31, i1 0)
+ store i8 %ctlz0 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 0), align 1
+ store i8 %ctlz1 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 1), align 1
+ store i8 %ctlz2 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 2), align 1
+ store i8 %ctlz3 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 3), align 1
+ store i8 %ctlz4 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 4), align 1
+ store i8 %ctlz5 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 5), align 1
+ store i8 %ctlz6 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 6), align 1
+ store i8 %ctlz7 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 7), align 1
+ store i8 %ctlz8 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 8), align 1
+ store i8 %ctlz9 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 9), align 1
+ store i8 %ctlz10, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 10), align 1
+ store i8 %ctlz11, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 11), align 1
+ store i8 %ctlz12, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 12), align 1
+ store i8 %ctlz13, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 13), align 1
+ store i8 %ctlz14, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 14), align 1
+ store i8 %ctlz15, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 15), align 1
+ store i8 %ctlz16, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 16), align 1
+ store i8 %ctlz17, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 17), align 1
+ store i8 %ctlz18, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 18), align 1
+ store i8 %ctlz19, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 19), align 1
+ store i8 %ctlz20, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 20), align 1
+ store i8 %ctlz21, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 21), align 1
+ store i8 %ctlz22, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 22), align 1
+ store i8 %ctlz23, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 23), align 1
+ store i8 %ctlz24, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 24), align 1
+ store i8 %ctlz25, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 25), align 1
+ store i8 %ctlz26, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 26), align 1
+ store i8 %ctlz27, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 27), align 1
+ store i8 %ctlz28, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 28), align 1
+ store i8 %ctlz29, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 29), align 1
+ store i8 %ctlz30, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 30), align 1
+ store i8 %ctlz31, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 31), align 1
+ ret void
+}
+
+;
+; CTLZ_ZERO_UNDEF
+;
+
+define void @ctlz_undef_2i64() #0 {
+; CHECK-LABEL: @ctlz_undef_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: [[CTLZ0:%.*]] = call i64 @llvm.ctlz.i64(i64 [[LD0]], i1 true)
+; CHECK-NEXT: [[CTLZ1:%.*]] = call i64 @llvm.ctlz.i64(i64 [[LD1]], i1 true)
+; CHECK-NEXT: store i64 [[CTLZ0]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i32 0, i64 0), align 8
+; CHECK-NEXT: store i64 [[CTLZ1]], 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
+ %ctlz0 = call i64 @llvm.ctlz.i64(i64 %ld0, i1 -1)
+ %ctlz1 = call i64 @llvm.ctlz.i64(i64 %ld1, i1 -1)
+ store i64 %ctlz0, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i32 0, i64 0), align 8
+ store i64 %ctlz1, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i32 0, i64 1), align 8
+ ret void
+}
+
+define void @ctlz_undef_4i64() #0 {
+; CHECK-LABEL: @ctlz_undef_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: [[CTLZ0:%.*]] = call i64 @llvm.ctlz.i64(i64 [[LD0]], i1 true)
+; CHECK-NEXT: [[CTLZ1:%.*]] = call i64 @llvm.ctlz.i64(i64 [[LD1]], i1 true)
+; CHECK-NEXT: [[CTLZ2:%.*]] = call i64 @llvm.ctlz.i64(i64 [[LD2]], i1 true)
+; CHECK-NEXT: [[CTLZ3:%.*]] = call i64 @llvm.ctlz.i64(i64 [[LD3]], i1 true)
+; CHECK-NEXT: store i64 [[CTLZ0]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 0), align 4
+; CHECK-NEXT: store i64 [[CTLZ1]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 1), align 4
+; CHECK-NEXT: store i64 [[CTLZ2]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 2), align 4
+; CHECK-NEXT: store i64 [[CTLZ3]], 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
+ %ctlz0 = call i64 @llvm.ctlz.i64(i64 %ld0, i1 -1)
+ %ctlz1 = call i64 @llvm.ctlz.i64(i64 %ld1, i1 -1)
+ %ctlz2 = call i64 @llvm.ctlz.i64(i64 %ld2, i1 -1)
+ %ctlz3 = call i64 @llvm.ctlz.i64(i64 %ld3, i1 -1)
+ store i64 %ctlz0, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 0), align 4
+ store i64 %ctlz1, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 1), align 4
+ store i64 %ctlz2, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 2), align 4
+ store i64 %ctlz3, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 3), align 4
+ ret void
+}
+
+define void @ctlz_undef_4i32() #0 {
+; CHECK-LABEL: @ctlz_undef_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: [[CTLZ0:%.*]] = call i32 @llvm.ctlz.i32(i32 [[LD0]], i1 true)
+; CHECK-NEXT: [[CTLZ1:%.*]] = call i32 @llvm.ctlz.i32(i32 [[LD1]], i1 true)
+; CHECK-NEXT: [[CTLZ2:%.*]] = call i32 @llvm.ctlz.i32(i32 [[LD2]], i1 true)
+; CHECK-NEXT: [[CTLZ3:%.*]] = call i32 @llvm.ctlz.i32(i32 [[LD3]], i1 true)
+; CHECK-NEXT: store i32 [[CTLZ0]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 0), align 4
+; CHECK-NEXT: store i32 [[CTLZ1]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 1), align 4
+; CHECK-NEXT: store i32 [[CTLZ2]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 2), align 4
+; CHECK-NEXT: store i32 [[CTLZ3]], 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
+ %ctlz0 = call i32 @llvm.ctlz.i32(i32 %ld0, i1 -1)
+ %ctlz1 = call i32 @llvm.ctlz.i32(i32 %ld1, i1 -1)
+ %ctlz2 = call i32 @llvm.ctlz.i32(i32 %ld2, i1 -1)
+ %ctlz3 = call i32 @llvm.ctlz.i32(i32 %ld3, i1 -1)
+ store i32 %ctlz0, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 0), align 4
+ store i32 %ctlz1, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 1), align 4
+ store i32 %ctlz2, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 2), align 4
+ store i32 %ctlz3, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 3), align 4
+ ret void
+}
+
+define void @ctlz_undef_8i32() #0 {
+; CHECK-LABEL: @ctlz_undef_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: [[CTLZ0:%.*]] = call i32 @llvm.ctlz.i32(i32 [[LD0]], i1 true)
+; CHECK-NEXT: [[CTLZ1:%.*]] = call i32 @llvm.ctlz.i32(i32 [[LD1]], i1 true)
+; CHECK-NEXT: [[CTLZ2:%.*]] = call i32 @llvm.ctlz.i32(i32 [[LD2]], i1 true)
+; CHECK-NEXT: [[CTLZ3:%.*]] = call i32 @llvm.ctlz.i32(i32 [[LD3]], i1 true)
+; CHECK-NEXT: [[CTLZ4:%.*]] = call i32 @llvm.ctlz.i32(i32 [[LD4]], i1 true)
+; CHECK-NEXT: [[CTLZ5:%.*]] = call i32 @llvm.ctlz.i32(i32 [[LD5]], i1 true)
+; CHECK-NEXT: [[CTLZ6:%.*]] = call i32 @llvm.ctlz.i32(i32 [[LD6]], i1 true)
+; CHECK-NEXT: [[CTLZ7:%.*]] = call i32 @llvm.ctlz.i32(i32 [[LD7]], i1 true)
+; CHECK-NEXT: store i32 [[CTLZ0]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 0), align 2
+; CHECK-NEXT: store i32 [[CTLZ1]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 1), align 2
+; CHECK-NEXT: store i32 [[CTLZ2]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 2), align 2
+; CHECK-NEXT: store i32 [[CTLZ3]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 3), align 2
+; CHECK-NEXT: store i32 [[CTLZ4]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 4), align 2
+; CHECK-NEXT: store i32 [[CTLZ5]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 5), align 2
+; CHECK-NEXT: store i32 [[CTLZ6]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 6), align 2
+; CHECK-NEXT: store i32 [[CTLZ7]], 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
+ %ctlz0 = call i32 @llvm.ctlz.i32(i32 %ld0, i1 -1)
+ %ctlz1 = call i32 @llvm.ctlz.i32(i32 %ld1, i1 -1)
+ %ctlz2 = call i32 @llvm.ctlz.i32(i32 %ld2, i1 -1)
+ %ctlz3 = call i32 @llvm.ctlz.i32(i32 %ld3, i1 -1)
+ %ctlz4 = call i32 @llvm.ctlz.i32(i32 %ld4, i1 -1)
+ %ctlz5 = call i32 @llvm.ctlz.i32(i32 %ld5, i1 -1)
+ %ctlz6 = call i32 @llvm.ctlz.i32(i32 %ld6, i1 -1)
+ %ctlz7 = call i32 @llvm.ctlz.i32(i32 %ld7, i1 -1)
+ store i32 %ctlz0, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 0), align 2
+ store i32 %ctlz1, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 1), align 2
+ store i32 %ctlz2, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 2), align 2
+ store i32 %ctlz3, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 3), align 2
+ store i32 %ctlz4, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 4), align 2
+ store i32 %ctlz5, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 5), align 2
+ store i32 %ctlz6, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 6), align 2
+ store i32 %ctlz7, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 7), align 2
+ ret void
+}
+
+define void @ctlz_undef_8i16() #0 {
+; CHECK-LABEL: @ctlz_undef_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: [[CTLZ0:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD0]], i1 true)
+; CHECK-NEXT: [[CTLZ1:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD1]], i1 true)
+; CHECK-NEXT: [[CTLZ2:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD2]], i1 true)
+; CHECK-NEXT: [[CTLZ3:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD3]], i1 true)
+; CHECK-NEXT: [[CTLZ4:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD4]], i1 true)
+; CHECK-NEXT: [[CTLZ5:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD5]], i1 true)
+; CHECK-NEXT: [[CTLZ6:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD6]], i1 true)
+; CHECK-NEXT: [[CTLZ7:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD7]], i1 true)
+; CHECK-NEXT: store i16 [[CTLZ0]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 0), align 2
+; CHECK-NEXT: store i16 [[CTLZ1]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 1), align 2
+; CHECK-NEXT: store i16 [[CTLZ2]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 2), align 2
+; CHECK-NEXT: store i16 [[CTLZ3]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 3), align 2
+; CHECK-NEXT: store i16 [[CTLZ4]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 4), align 2
+; CHECK-NEXT: store i16 [[CTLZ5]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 5), align 2
+; CHECK-NEXT: store i16 [[CTLZ6]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 6), align 2
+; CHECK-NEXT: store i16 [[CTLZ7]], 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
+ %ctlz0 = call i16 @llvm.ctlz.i16(i16 %ld0, i1 -1)
+ %ctlz1 = call i16 @llvm.ctlz.i16(i16 %ld1, i1 -1)
+ %ctlz2 = call i16 @llvm.ctlz.i16(i16 %ld2, i1 -1)
+ %ctlz3 = call i16 @llvm.ctlz.i16(i16 %ld3, i1 -1)
+ %ctlz4 = call i16 @llvm.ctlz.i16(i16 %ld4, i1 -1)
+ %ctlz5 = call i16 @llvm.ctlz.i16(i16 %ld5, i1 -1)
+ %ctlz6 = call i16 @llvm.ctlz.i16(i16 %ld6, i1 -1)
+ %ctlz7 = call i16 @llvm.ctlz.i16(i16 %ld7, i1 -1)
+ store i16 %ctlz0, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 0), align 2
+ store i16 %ctlz1, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 1), align 2
+ store i16 %ctlz2, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 2), align 2
+ store i16 %ctlz3, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 3), align 2
+ store i16 %ctlz4, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 4), align 2
+ store i16 %ctlz5, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 5), align 2
+ store i16 %ctlz6, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 6), align 2
+ store i16 %ctlz7, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 7), align 2
+ ret void
+}
+
+define void @ctlz_undef_16i16() #0 {
+; CHECK-LABEL: @ctlz_undef_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: [[CTLZ0:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD0]], i1 true)
+; CHECK-NEXT: [[CTLZ1:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD1]], i1 true)
+; CHECK-NEXT: [[CTLZ2:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD2]], i1 true)
+; CHECK-NEXT: [[CTLZ3:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD3]], i1 true)
+; CHECK-NEXT: [[CTLZ4:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD4]], i1 true)
+; CHECK-NEXT: [[CTLZ5:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD5]], i1 true)
+; CHECK-NEXT: [[CTLZ6:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD6]], i1 true)
+; CHECK-NEXT: [[CTLZ7:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD7]], i1 true)
+; CHECK-NEXT: [[CTLZ8:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD8]], i1 true)
+; CHECK-NEXT: [[CTLZ9:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD9]], i1 true)
+; CHECK-NEXT: [[CTLZ10:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD10]], i1 true)
+; CHECK-NEXT: [[CTLZ11:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD11]], i1 true)
+; CHECK-NEXT: [[CTLZ12:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD12]], i1 true)
+; CHECK-NEXT: [[CTLZ13:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD13]], i1 true)
+; CHECK-NEXT: [[CTLZ14:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD14]], i1 true)
+; CHECK-NEXT: [[CTLZ15:%.*]] = call i16 @llvm.ctlz.i16(i16 [[LD15]], i1 true)
+; CHECK-NEXT: store i16 [[CTLZ0]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 0), align 2
+; CHECK-NEXT: store i16 [[CTLZ1]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 1), align 2
+; CHECK-NEXT: store i16 [[CTLZ2]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 2), align 2
+; CHECK-NEXT: store i16 [[CTLZ3]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 3), align 2
+; CHECK-NEXT: store i16 [[CTLZ4]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 4), align 2
+; CHECK-NEXT: store i16 [[CTLZ5]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 5), align 2
+; CHECK-NEXT: store i16 [[CTLZ6]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 6), align 2
+; CHECK-NEXT: store i16 [[CTLZ7]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 7), align 2
+; CHECK-NEXT: store i16 [[CTLZ8]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 8), align 2
+; CHECK-NEXT: store i16 [[CTLZ9]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 9), align 2
+; CHECK-NEXT: store i16 [[CTLZ10]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 10), align 2
+; CHECK-NEXT: store i16 [[CTLZ11]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 11), align 2
+; CHECK-NEXT: store i16 [[CTLZ12]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 12), align 2
+; CHECK-NEXT: store i16 [[CTLZ13]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 13), align 2
+; CHECK-NEXT: store i16 [[CTLZ14]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 14), align 2
+; CHECK-NEXT: store i16 [[CTLZ15]], 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
+ %ctlz0 = call i16 @llvm.ctlz.i16(i16 %ld0, i1 -1)
+ %ctlz1 = call i16 @llvm.ctlz.i16(i16 %ld1, i1 -1)
+ %ctlz2 = call i16 @llvm.ctlz.i16(i16 %ld2, i1 -1)
+ %ctlz3 = call i16 @llvm.ctlz.i16(i16 %ld3, i1 -1)
+ %ctlz4 = call i16 @llvm.ctlz.i16(i16 %ld4, i1 -1)
+ %ctlz5 = call i16 @llvm.ctlz.i16(i16 %ld5, i1 -1)
+ %ctlz6 = call i16 @llvm.ctlz.i16(i16 %ld6, i1 -1)
+ %ctlz7 = call i16 @llvm.ctlz.i16(i16 %ld7, i1 -1)
+ %ctlz8 = call i16 @llvm.ctlz.i16(i16 %ld8, i1 -1)
+ %ctlz9 = call i16 @llvm.ctlz.i16(i16 %ld9, i1 -1)
+ %ctlz10 = call i16 @llvm.ctlz.i16(i16 %ld10, i1 -1)
+ %ctlz11 = call i16 @llvm.ctlz.i16(i16 %ld11, i1 -1)
+ %ctlz12 = call i16 @llvm.ctlz.i16(i16 %ld12, i1 -1)
+ %ctlz13 = call i16 @llvm.ctlz.i16(i16 %ld13, i1 -1)
+ %ctlz14 = call i16 @llvm.ctlz.i16(i16 %ld14, i1 -1)
+ %ctlz15 = call i16 @llvm.ctlz.i16(i16 %ld15, i1 -1)
+ store i16 %ctlz0 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 0), align 2
+ store i16 %ctlz1 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 1), align 2
+ store i16 %ctlz2 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 2), align 2
+ store i16 %ctlz3 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 3), align 2
+ store i16 %ctlz4 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 4), align 2
+ store i16 %ctlz5 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 5), align 2
+ store i16 %ctlz6 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 6), align 2
+ store i16 %ctlz7 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 7), align 2
+ store i16 %ctlz8 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 8), align 2
+ store i16 %ctlz9 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 9), align 2
+ store i16 %ctlz10, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 10), align 2
+ store i16 %ctlz11, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 11), align 2
+ store i16 %ctlz12, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 12), align 2
+ store i16 %ctlz13, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 13), align 2
+ store i16 %ctlz14, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 14), align 2
+ store i16 %ctlz15, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 15), align 2
+ ret void
+}
+
+define void @ctlz_undef_16i8() #0 {
+; CHECK-LABEL: @ctlz_undef_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: [[CTLZ0:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD0]], i1 true)
+; CHECK-NEXT: [[CTLZ1:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD1]], i1 true)
+; CHECK-NEXT: [[CTLZ2:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD2]], i1 true)
+; CHECK-NEXT: [[CTLZ3:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD3]], i1 true)
+; CHECK-NEXT: [[CTLZ4:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD4]], i1 true)
+; CHECK-NEXT: [[CTLZ5:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD5]], i1 true)
+; CHECK-NEXT: [[CTLZ6:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD6]], i1 true)
+; CHECK-NEXT: [[CTLZ7:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD7]], i1 true)
+; CHECK-NEXT: [[CTLZ8:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD8]], i1 true)
+; CHECK-NEXT: [[CTLZ9:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD9]], i1 true)
+; CHECK-NEXT: [[CTLZ10:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD10]], i1 true)
+; CHECK-NEXT: [[CTLZ11:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD11]], i1 true)
+; CHECK-NEXT: [[CTLZ12:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD12]], i1 true)
+; CHECK-NEXT: [[CTLZ13:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD13]], i1 true)
+; CHECK-NEXT: [[CTLZ14:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD14]], i1 true)
+; CHECK-NEXT: [[CTLZ15:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD15]], i1 true)
+; CHECK-NEXT: store i8 [[CTLZ0]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 0), align 1
+; CHECK-NEXT: store i8 [[CTLZ1]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 1), align 1
+; CHECK-NEXT: store i8 [[CTLZ2]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 2), align 1
+; CHECK-NEXT: store i8 [[CTLZ3]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 3), align 1
+; CHECK-NEXT: store i8 [[CTLZ4]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 4), align 1
+; CHECK-NEXT: store i8 [[CTLZ5]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 5), align 1
+; CHECK-NEXT: store i8 [[CTLZ6]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 6), align 1
+; CHECK-NEXT: store i8 [[CTLZ7]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 7), align 1
+; CHECK-NEXT: store i8 [[CTLZ8]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 8), align 1
+; CHECK-NEXT: store i8 [[CTLZ9]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 9), align 1
+; CHECK-NEXT: store i8 [[CTLZ10]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 10), align 1
+; CHECK-NEXT: store i8 [[CTLZ11]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 11), align 1
+; CHECK-NEXT: store i8 [[CTLZ12]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 12), align 1
+; CHECK-NEXT: store i8 [[CTLZ13]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 13), align 1
+; CHECK-NEXT: store i8 [[CTLZ14]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 14), align 1
+; CHECK-NEXT: store i8 [[CTLZ15]], 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
+ %ctlz0 = call i8 @llvm.ctlz.i8(i8 %ld0, i1 -1)
+ %ctlz1 = call i8 @llvm.ctlz.i8(i8 %ld1, i1 -1)
+ %ctlz2 = call i8 @llvm.ctlz.i8(i8 %ld2, i1 -1)
+ %ctlz3 = call i8 @llvm.ctlz.i8(i8 %ld3, i1 -1)
+ %ctlz4 = call i8 @llvm.ctlz.i8(i8 %ld4, i1 -1)
+ %ctlz5 = call i8 @llvm.ctlz.i8(i8 %ld5, i1 -1)
+ %ctlz6 = call i8 @llvm.ctlz.i8(i8 %ld6, i1 -1)
+ %ctlz7 = call i8 @llvm.ctlz.i8(i8 %ld7, i1 -1)
+ %ctlz8 = call i8 @llvm.ctlz.i8(i8 %ld8, i1 -1)
+ %ctlz9 = call i8 @llvm.ctlz.i8(i8 %ld9, i1 -1)
+ %ctlz10 = call i8 @llvm.ctlz.i8(i8 %ld10, i1 -1)
+ %ctlz11 = call i8 @llvm.ctlz.i8(i8 %ld11, i1 -1)
+ %ctlz12 = call i8 @llvm.ctlz.i8(i8 %ld12, i1 -1)
+ %ctlz13 = call i8 @llvm.ctlz.i8(i8 %ld13, i1 -1)
+ %ctlz14 = call i8 @llvm.ctlz.i8(i8 %ld14, i1 -1)
+ %ctlz15 = call i8 @llvm.ctlz.i8(i8 %ld15, i1 -1)
+ store i8 %ctlz0 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 0), align 1
+ store i8 %ctlz1 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 1), align 1
+ store i8 %ctlz2 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 2), align 1
+ store i8 %ctlz3 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 3), align 1
+ store i8 %ctlz4 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 4), align 1
+ store i8 %ctlz5 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 5), align 1
+ store i8 %ctlz6 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 6), align 1
+ store i8 %ctlz7 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 7), align 1
+ store i8 %ctlz8 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 8), align 1
+ store i8 %ctlz9 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 9), align 1
+ store i8 %ctlz10, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 10), align 1
+ store i8 %ctlz11, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 11), align 1
+ store i8 %ctlz12, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 12), align 1
+ store i8 %ctlz13, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 13), align 1
+ store i8 %ctlz14, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 14), align 1
+ store i8 %ctlz15, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 15), align 1
+ ret void
+}
+
+define void @ctlz_undef_32i8() #0 {
+; CHECK-LABEL: @ctlz_undef_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: [[CTLZ0:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD0]], i1 true)
+; CHECK-NEXT: [[CTLZ1:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD1]], i1 true)
+; CHECK-NEXT: [[CTLZ2:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD2]], i1 true)
+; CHECK-NEXT: [[CTLZ3:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD3]], i1 true)
+; CHECK-NEXT: [[CTLZ4:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD4]], i1 true)
+; CHECK-NEXT: [[CTLZ5:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD5]], i1 true)
+; CHECK-NEXT: [[CTLZ6:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD6]], i1 true)
+; CHECK-NEXT: [[CTLZ7:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD7]], i1 true)
+; CHECK-NEXT: [[CTLZ8:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD8]], i1 true)
+; CHECK-NEXT: [[CTLZ9:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD9]], i1 true)
+; CHECK-NEXT: [[CTLZ10:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD10]], i1 true)
+; CHECK-NEXT: [[CTLZ11:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD11]], i1 true)
+; CHECK-NEXT: [[CTLZ12:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD12]], i1 true)
+; CHECK-NEXT: [[CTLZ13:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD13]], i1 true)
+; CHECK-NEXT: [[CTLZ14:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD14]], i1 true)
+; CHECK-NEXT: [[CTLZ15:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD15]], i1 true)
+; CHECK-NEXT: [[CTLZ16:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD16]], i1 true)
+; CHECK-NEXT: [[CTLZ17:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD17]], i1 true)
+; CHECK-NEXT: [[CTLZ18:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD18]], i1 true)
+; CHECK-NEXT: [[CTLZ19:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD19]], i1 true)
+; CHECK-NEXT: [[CTLZ20:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD20]], i1 true)
+; CHECK-NEXT: [[CTLZ21:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD21]], i1 true)
+; CHECK-NEXT: [[CTLZ22:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD22]], i1 true)
+; CHECK-NEXT: [[CTLZ23:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD23]], i1 true)
+; CHECK-NEXT: [[CTLZ24:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD24]], i1 true)
+; CHECK-NEXT: [[CTLZ25:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD25]], i1 true)
+; CHECK-NEXT: [[CTLZ26:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD26]], i1 true)
+; CHECK-NEXT: [[CTLZ27:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD27]], i1 true)
+; CHECK-NEXT: [[CTLZ28:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD28]], i1 true)
+; CHECK-NEXT: [[CTLZ29:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD29]], i1 true)
+; CHECK-NEXT: [[CTLZ30:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD30]], i1 true)
+; CHECK-NEXT: [[CTLZ31:%.*]] = call i8 @llvm.ctlz.i8(i8 [[LD31]], i1 true)
+; CHECK-NEXT: store i8 [[CTLZ0]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 0), align 1
+; CHECK-NEXT: store i8 [[CTLZ1]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 1), align 1
+; CHECK-NEXT: store i8 [[CTLZ2]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 2), align 1
+; CHECK-NEXT: store i8 [[CTLZ3]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 3), align 1
+; CHECK-NEXT: store i8 [[CTLZ4]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 4), align 1
+; CHECK-NEXT: store i8 [[CTLZ5]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 5), align 1
+; CHECK-NEXT: store i8 [[CTLZ6]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 6), align 1
+; CHECK-NEXT: store i8 [[CTLZ7]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 7), align 1
+; CHECK-NEXT: store i8 [[CTLZ8]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 8), align 1
+; CHECK-NEXT: store i8 [[CTLZ9]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 9), align 1
+; CHECK-NEXT: store i8 [[CTLZ10]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 10), align 1
+; CHECK-NEXT: store i8 [[CTLZ11]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 11), align 1
+; CHECK-NEXT: store i8 [[CTLZ12]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 12), align 1
+; CHECK-NEXT: store i8 [[CTLZ13]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 13), align 1
+; CHECK-NEXT: store i8 [[CTLZ14]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 14), align 1
+; CHECK-NEXT: store i8 [[CTLZ15]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 15), align 1
+; CHECK-NEXT: store i8 [[CTLZ16]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 16), align 1
+; CHECK-NEXT: store i8 [[CTLZ17]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 17), align 1
+; CHECK-NEXT: store i8 [[CTLZ18]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 18), align 1
+; CHECK-NEXT: store i8 [[CTLZ19]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 19), align 1
+; CHECK-NEXT: store i8 [[CTLZ20]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 20), align 1
+; CHECK-NEXT: store i8 [[CTLZ21]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 21), align 1
+; CHECK-NEXT: store i8 [[CTLZ22]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 22), align 1
+; CHECK-NEXT: store i8 [[CTLZ23]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 23), align 1
+; CHECK-NEXT: store i8 [[CTLZ24]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 24), align 1
+; CHECK-NEXT: store i8 [[CTLZ25]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 25), align 1
+; CHECK-NEXT: store i8 [[CTLZ26]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 26), align 1
+; CHECK-NEXT: store i8 [[CTLZ27]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 27), align 1
+; CHECK-NEXT: store i8 [[CTLZ28]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 28), align 1
+; CHECK-NEXT: store i8 [[CTLZ29]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 29), align 1
+; CHECK-NEXT: store i8 [[CTLZ30]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 30), align 1
+; CHECK-NEXT: store i8 [[CTLZ31]], 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
+ %ctlz0 = call i8 @llvm.ctlz.i8(i8 %ld0, i1 -1)
+ %ctlz1 = call i8 @llvm.ctlz.i8(i8 %ld1, i1 -1)
+ %ctlz2 = call i8 @llvm.ctlz.i8(i8 %ld2, i1 -1)
+ %ctlz3 = call i8 @llvm.ctlz.i8(i8 %ld3, i1 -1)
+ %ctlz4 = call i8 @llvm.ctlz.i8(i8 %ld4, i1 -1)
+ %ctlz5 = call i8 @llvm.ctlz.i8(i8 %ld5, i1 -1)
+ %ctlz6 = call i8 @llvm.ctlz.i8(i8 %ld6, i1 -1)
+ %ctlz7 = call i8 @llvm.ctlz.i8(i8 %ld7, i1 -1)
+ %ctlz8 = call i8 @llvm.ctlz.i8(i8 %ld8, i1 -1)
+ %ctlz9 = call i8 @llvm.ctlz.i8(i8 %ld9, i1 -1)
+ %ctlz10 = call i8 @llvm.ctlz.i8(i8 %ld10, i1 -1)
+ %ctlz11 = call i8 @llvm.ctlz.i8(i8 %ld11, i1 -1)
+ %ctlz12 = call i8 @llvm.ctlz.i8(i8 %ld12, i1 -1)
+ %ctlz13 = call i8 @llvm.ctlz.i8(i8 %ld13, i1 -1)
+ %ctlz14 = call i8 @llvm.ctlz.i8(i8 %ld14, i1 -1)
+ %ctlz15 = call i8 @llvm.ctlz.i8(i8 %ld15, i1 -1)
+ %ctlz16 = call i8 @llvm.ctlz.i8(i8 %ld16, i1 -1)
+ %ctlz17 = call i8 @llvm.ctlz.i8(i8 %ld17, i1 -1)
+ %ctlz18 = call i8 @llvm.ctlz.i8(i8 %ld18, i1 -1)
+ %ctlz19 = call i8 @llvm.ctlz.i8(i8 %ld19, i1 -1)
+ %ctlz20 = call i8 @llvm.ctlz.i8(i8 %ld20, i1 -1)
+ %ctlz21 = call i8 @llvm.ctlz.i8(i8 %ld21, i1 -1)
+ %ctlz22 = call i8 @llvm.ctlz.i8(i8 %ld22, i1 -1)
+ %ctlz23 = call i8 @llvm.ctlz.i8(i8 %ld23, i1 -1)
+ %ctlz24 = call i8 @llvm.ctlz.i8(i8 %ld24, i1 -1)
+ %ctlz25 = call i8 @llvm.ctlz.i8(i8 %ld25, i1 -1)
+ %ctlz26 = call i8 @llvm.ctlz.i8(i8 %ld26, i1 -1)
+ %ctlz27 = call i8 @llvm.ctlz.i8(i8 %ld27, i1 -1)
+ %ctlz28 = call i8 @llvm.ctlz.i8(i8 %ld28, i1 -1)
+ %ctlz29 = call i8 @llvm.ctlz.i8(i8 %ld29, i1 -1)
+ %ctlz30 = call i8 @llvm.ctlz.i8(i8 %ld30, i1 -1)
+ %ctlz31 = call i8 @llvm.ctlz.i8(i8 %ld31, i1 -1)
+ store i8 %ctlz0 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 0), align 1
+ store i8 %ctlz1 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 1), align 1
+ store i8 %ctlz2 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 2), align 1
+ store i8 %ctlz3 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 3), align 1
+ store i8 %ctlz4 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 4), align 1
+ store i8 %ctlz5 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 5), align 1
+ store i8 %ctlz6 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 6), align 1
+ store i8 %ctlz7 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 7), align 1
+ store i8 %ctlz8 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 8), align 1
+ store i8 %ctlz9 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 9), align 1
+ store i8 %ctlz10, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 10), align 1
+ store i8 %ctlz11, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 11), align 1
+ store i8 %ctlz12, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 12), align 1
+ store i8 %ctlz13, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 13), align 1
+ store i8 %ctlz14, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 14), align 1
+ store i8 %ctlz15, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 15), align 1
+ store i8 %ctlz16, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 16), align 1
+ store i8 %ctlz17, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 17), align 1
+ store i8 %ctlz18, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 18), align 1
+ store i8 %ctlz19, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 19), align 1
+ store i8 %ctlz20, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 20), align 1
+ store i8 %ctlz21, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 21), align 1
+ store i8 %ctlz22, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 22), align 1
+ store i8 %ctlz23, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 23), align 1
+ store i8 %ctlz24, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 24), align 1
+ store i8 %ctlz25, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 25), align 1
+ store i8 %ctlz26, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 26), align 1
+ store i8 %ctlz27, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 27), align 1
+ store i8 %ctlz28, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 28), align 1
+ store i8 %ctlz29, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 29), align 1
+ store i8 %ctlz30, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 30), align 1
+ store i8 %ctlz31, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 31), align 1
+ ret void
+}
+
+attributes #0 = { nounwind }
Added: llvm/trunk/test/Transforms/SLPVectorizer/X86/ctpop.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Transforms/SLPVectorizer/X86/ctpop.ll?rev=268800&view=auto
==============================================================================
--- llvm/trunk/test/Transforms/SLPVectorizer/X86/ctpop.ll (added)
+++ llvm/trunk/test/Transforms/SLPVectorizer/X86/ctpop.ll Fri May 6 16:33:01 2016
@@ -0,0 +1,403 @@
+; 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"
+
+ at src64 = common global [4 x i64] zeroinitializer, align 32
+ at dst64 = common global [4 x i64] zeroinitializer, align 32
+ at src32 = common global [8 x i32] zeroinitializer, align 32
+ at dst32 = common global [8 x i32] zeroinitializer, align 32
+ at src16 = common global [16 x i16] zeroinitializer, align 32
+ at dst16 = common global [16 x i16] zeroinitializer, align 32
+ at src8 = common global [32 x i8] zeroinitializer, align 32
+ at dst8 = common global [32 x i8] zeroinitializer, align 32
+
+declare i64 @llvm.ctpop.i64(i64)
+declare i32 @llvm.ctpop.i32(i32)
+declare i16 @llvm.ctpop.i16(i16)
+declare i8 @llvm.ctpop.i8(i8)
+
+define void @ctpop_2i64() #0 {
+; CHECK-LABEL: @ctpop_2i64(
+; CHECK-NEXT: [[TMP1:%.*]] = load <2 x i64>, <2 x i64>* bitcast ([4 x i64]* @src64 to <2 x i64>*), align 8
+; CHECK-NEXT: [[TMP2:%.*]] = call <2 x i64> @llvm.ctpop.v2i64(<2 x i64> [[TMP1]])
+; CHECK-NEXT: store <2 x i64> [[TMP2]], <2 x i64>* bitcast ([4 x i64]* @dst64 to <2 x i64>*), 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
+ %ctpop0 = call i64 @llvm.ctpop.i64(i64 %ld0)
+ %ctpop1 = call i64 @llvm.ctpop.i64(i64 %ld1)
+ store i64 %ctpop0, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i32 0, i64 0), align 8
+ store i64 %ctpop1, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i32 0, i64 1), align 8
+ ret void
+}
+
+define void @ctpop_4i64() #0 {
+; SSE-LABEL: @ctpop_4i64(
+; SSE-NEXT: [[TMP1:%.*]] = load <2 x i64>, <2 x i64>* bitcast ([4 x i64]* @src64 to <2 x i64>*), align 4
+; SSE-NEXT: [[TMP2:%.*]] = load <2 x i64>, <2 x i64>* bitcast (i64* getelementptr inbounds ([4 x i64], [4 x i64]* @src64, i64 0, i64 2) to <2 x i64>*), align 4
+; SSE-NEXT: [[TMP3:%.*]] = call <2 x i64> @llvm.ctpop.v2i64(<2 x i64> [[TMP1]])
+; SSE-NEXT: [[TMP4:%.*]] = call <2 x i64> @llvm.ctpop.v2i64(<2 x i64> [[TMP2]])
+; SSE-NEXT: store <2 x i64> [[TMP3]], <2 x i64>* bitcast ([4 x i64]* @dst64 to <2 x i64>*), align 4
+; SSE-NEXT: store <2 x i64> [[TMP4]], <2 x i64>* bitcast (i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 2) to <2 x i64>*), align 4
+; SSE-NEXT: ret void
+;
+; AVX-LABEL: @ctpop_4i64(
+; AVX-NEXT: [[TMP1:%.*]] = load <4 x i64>, <4 x i64>* bitcast ([4 x i64]* @src64 to <4 x i64>*), align 4
+; AVX-NEXT: [[TMP2:%.*]] = call <4 x i64> @llvm.ctpop.v4i64(<4 x i64> [[TMP1]])
+; AVX-NEXT: store <4 x i64> [[TMP2]], <4 x i64>* bitcast ([4 x i64]* @dst64 to <4 x i64>*), align 4
+; AVX-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
+ %ctpop0 = call i64 @llvm.ctpop.i64(i64 %ld0)
+ %ctpop1 = call i64 @llvm.ctpop.i64(i64 %ld1)
+ %ctpop2 = call i64 @llvm.ctpop.i64(i64 %ld2)
+ %ctpop3 = call i64 @llvm.ctpop.i64(i64 %ld3)
+ store i64 %ctpop0, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 0), align 4
+ store i64 %ctpop1, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 1), align 4
+ store i64 %ctpop2, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 2), align 4
+ store i64 %ctpop3, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 3), align 4
+ ret void
+}
+
+define void @ctpop_4i32() #0 {
+; CHECK-LABEL: @ctpop_4i32(
+; CHECK-NEXT: [[TMP1:%.*]] = load <4 x i32>, <4 x i32>* bitcast ([8 x i32]* @src32 to <4 x i32>*), align 4
+; CHECK-NEXT: [[TMP2:%.*]] = call <4 x i32> @llvm.ctpop.v4i32(<4 x i32> [[TMP1]])
+; CHECK-NEXT: store <4 x i32> [[TMP2]], <4 x i32>* bitcast ([8 x i32]* @dst32 to <4 x i32>*), 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
+ %ctpop0 = call i32 @llvm.ctpop.i32(i32 %ld0)
+ %ctpop1 = call i32 @llvm.ctpop.i32(i32 %ld1)
+ %ctpop2 = call i32 @llvm.ctpop.i32(i32 %ld2)
+ %ctpop3 = call i32 @llvm.ctpop.i32(i32 %ld3)
+ store i32 %ctpop0, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 0), align 4
+ store i32 %ctpop1, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 1), align 4
+ store i32 %ctpop2, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 2), align 4
+ store i32 %ctpop3, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 3), align 4
+ ret void
+}
+
+define void @ctpop_8i32() #0 {
+; SSE-LABEL: @ctpop_8i32(
+; SSE-NEXT: [[TMP1:%.*]] = load <4 x i32>, <4 x i32>* bitcast ([8 x i32]* @src32 to <4 x i32>*), align 2
+; SSE-NEXT: [[TMP2:%.*]] = load <4 x i32>, <4 x i32>* bitcast (i32* getelementptr inbounds ([8 x i32], [8 x i32]* @src32, i32 0, i64 4) to <4 x i32>*), align 2
+; SSE-NEXT: [[TMP3:%.*]] = call <4 x i32> @llvm.ctpop.v4i32(<4 x i32> [[TMP1]])
+; SSE-NEXT: [[TMP4:%.*]] = call <4 x i32> @llvm.ctpop.v4i32(<4 x i32> [[TMP2]])
+; SSE-NEXT: store <4 x i32> [[TMP3]], <4 x i32>* bitcast ([8 x i32]* @dst32 to <4 x i32>*), align 2
+; SSE-NEXT: store <4 x i32> [[TMP4]], <4 x i32>* bitcast (i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 4) to <4 x i32>*), align 2
+; SSE-NEXT: ret void
+;
+; AVX-LABEL: @ctpop_8i32(
+; AVX-NEXT: [[TMP1:%.*]] = load <8 x i32>, <8 x i32>* bitcast ([8 x i32]* @src32 to <8 x i32>*), align 2
+; AVX-NEXT: [[TMP2:%.*]] = call <8 x i32> @llvm.ctpop.v8i32(<8 x i32> [[TMP1]])
+; AVX-NEXT: store <8 x i32> [[TMP2]], <8 x i32>* bitcast ([8 x i32]* @dst32 to <8 x i32>*), align 2
+; AVX-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
+ %ctpop0 = call i32 @llvm.ctpop.i32(i32 %ld0)
+ %ctpop1 = call i32 @llvm.ctpop.i32(i32 %ld1)
+ %ctpop2 = call i32 @llvm.ctpop.i32(i32 %ld2)
+ %ctpop3 = call i32 @llvm.ctpop.i32(i32 %ld3)
+ %ctpop4 = call i32 @llvm.ctpop.i32(i32 %ld4)
+ %ctpop5 = call i32 @llvm.ctpop.i32(i32 %ld5)
+ %ctpop6 = call i32 @llvm.ctpop.i32(i32 %ld6)
+ %ctpop7 = call i32 @llvm.ctpop.i32(i32 %ld7)
+ store i32 %ctpop0, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 0), align 2
+ store i32 %ctpop1, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 1), align 2
+ store i32 %ctpop2, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 2), align 2
+ store i32 %ctpop3, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 3), align 2
+ store i32 %ctpop4, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 4), align 2
+ store i32 %ctpop5, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 5), align 2
+ store i32 %ctpop6, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 6), align 2
+ store i32 %ctpop7, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 7), align 2
+ ret void
+}
+
+define void @ctpop_8i16() #0 {
+; CHECK-LABEL: @ctpop_8i16(
+; CHECK-NEXT: [[TMP1:%.*]] = load <8 x i16>, <8 x i16>* bitcast ([16 x i16]* @src16 to <8 x i16>*), align 2
+; CHECK-NEXT: [[TMP2:%.*]] = call <8 x i16> @llvm.ctpop.v8i16(<8 x i16> [[TMP1]])
+; CHECK-NEXT: store <8 x i16> [[TMP2]], <8 x i16>* bitcast ([16 x i16]* @dst16 to <8 x i16>*), 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
+ %ctpop0 = call i16 @llvm.ctpop.i16(i16 %ld0)
+ %ctpop1 = call i16 @llvm.ctpop.i16(i16 %ld1)
+ %ctpop2 = call i16 @llvm.ctpop.i16(i16 %ld2)
+ %ctpop3 = call i16 @llvm.ctpop.i16(i16 %ld3)
+ %ctpop4 = call i16 @llvm.ctpop.i16(i16 %ld4)
+ %ctpop5 = call i16 @llvm.ctpop.i16(i16 %ld5)
+ %ctpop6 = call i16 @llvm.ctpop.i16(i16 %ld6)
+ %ctpop7 = call i16 @llvm.ctpop.i16(i16 %ld7)
+ store i16 %ctpop0, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 0), align 2
+ store i16 %ctpop1, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 1), align 2
+ store i16 %ctpop2, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 2), align 2
+ store i16 %ctpop3, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 3), align 2
+ store i16 %ctpop4, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 4), align 2
+ store i16 %ctpop5, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 5), align 2
+ store i16 %ctpop6, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 6), align 2
+ store i16 %ctpop7, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 7), align 2
+ ret void
+}
+
+define void @ctpop_16i16() #0 {
+; SSE-LABEL: @ctpop_16i16(
+; SSE-NEXT: [[TMP1:%.*]] = load <8 x i16>, <8 x i16>* bitcast ([16 x i16]* @src16 to <8 x i16>*), align 2
+; SSE-NEXT: [[TMP2:%.*]] = load <8 x i16>, <8 x i16>* bitcast (i16* getelementptr inbounds ([16 x i16], [16 x i16]* @src16, i16 0, i64 8) to <8 x i16>*), align 2
+; SSE-NEXT: [[TMP3:%.*]] = call <8 x i16> @llvm.ctpop.v8i16(<8 x i16> [[TMP1]])
+; SSE-NEXT: [[TMP4:%.*]] = call <8 x i16> @llvm.ctpop.v8i16(<8 x i16> [[TMP2]])
+; SSE-NEXT: store <8 x i16> [[TMP3]], <8 x i16>* bitcast ([16 x i16]* @dst16 to <8 x i16>*), align 2
+; SSE-NEXT: store <8 x i16> [[TMP4]], <8 x i16>* bitcast (i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 8) to <8 x i16>*), align 2
+; SSE-NEXT: ret void
+;
+; AVX-LABEL: @ctpop_16i16(
+; AVX-NEXT: [[TMP1:%.*]] = load <16 x i16>, <16 x i16>* bitcast ([16 x i16]* @src16 to <16 x i16>*), align 2
+; AVX-NEXT: [[TMP2:%.*]] = call <16 x i16> @llvm.ctpop.v16i16(<16 x i16> [[TMP1]])
+; AVX-NEXT: store <16 x i16> [[TMP2]], <16 x i16>* bitcast ([16 x i16]* @dst16 to <16 x i16>*), align 2
+; AVX-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
+ %ctpop0 = call i16 @llvm.ctpop.i16(i16 %ld0)
+ %ctpop1 = call i16 @llvm.ctpop.i16(i16 %ld1)
+ %ctpop2 = call i16 @llvm.ctpop.i16(i16 %ld2)
+ %ctpop3 = call i16 @llvm.ctpop.i16(i16 %ld3)
+ %ctpop4 = call i16 @llvm.ctpop.i16(i16 %ld4)
+ %ctpop5 = call i16 @llvm.ctpop.i16(i16 %ld5)
+ %ctpop6 = call i16 @llvm.ctpop.i16(i16 %ld6)
+ %ctpop7 = call i16 @llvm.ctpop.i16(i16 %ld7)
+ %ctpop8 = call i16 @llvm.ctpop.i16(i16 %ld8)
+ %ctpop9 = call i16 @llvm.ctpop.i16(i16 %ld9)
+ %ctpop10 = call i16 @llvm.ctpop.i16(i16 %ld10)
+ %ctpop11 = call i16 @llvm.ctpop.i16(i16 %ld11)
+ %ctpop12 = call i16 @llvm.ctpop.i16(i16 %ld12)
+ %ctpop13 = call i16 @llvm.ctpop.i16(i16 %ld13)
+ %ctpop14 = call i16 @llvm.ctpop.i16(i16 %ld14)
+ %ctpop15 = call i16 @llvm.ctpop.i16(i16 %ld15)
+ store i16 %ctpop0 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 0), align 2
+ store i16 %ctpop1 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 1), align 2
+ store i16 %ctpop2 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 2), align 2
+ store i16 %ctpop3 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 3), align 2
+ store i16 %ctpop4 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 4), align 2
+ store i16 %ctpop5 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 5), align 2
+ store i16 %ctpop6 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 6), align 2
+ store i16 %ctpop7 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 7), align 2
+ store i16 %ctpop8 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 8), align 2
+ store i16 %ctpop9 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 9), align 2
+ store i16 %ctpop10, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 10), align 2
+ store i16 %ctpop11, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 11), align 2
+ store i16 %ctpop12, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 12), align 2
+ store i16 %ctpop13, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 13), align 2
+ store i16 %ctpop14, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 14), align 2
+ store i16 %ctpop15, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 15), align 2
+ ret void
+}
+
+define void @ctpop_16i8() #0 {
+; CHECK-LABEL: @ctpop_16i8(
+; CHECK-NEXT: [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* bitcast ([32 x i8]* @src8 to <16 x i8>*), align 1
+; CHECK-NEXT: [[TMP2:%.*]] = call <16 x i8> @llvm.ctpop.v16i8(<16 x i8> [[TMP1]])
+; CHECK-NEXT: store <16 x i8> [[TMP2]], <16 x i8>* bitcast ([32 x i8]* @dst8 to <16 x i8>*), 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
+ %ctpop0 = call i8 @llvm.ctpop.i8(i8 %ld0)
+ %ctpop1 = call i8 @llvm.ctpop.i8(i8 %ld1)
+ %ctpop2 = call i8 @llvm.ctpop.i8(i8 %ld2)
+ %ctpop3 = call i8 @llvm.ctpop.i8(i8 %ld3)
+ %ctpop4 = call i8 @llvm.ctpop.i8(i8 %ld4)
+ %ctpop5 = call i8 @llvm.ctpop.i8(i8 %ld5)
+ %ctpop6 = call i8 @llvm.ctpop.i8(i8 %ld6)
+ %ctpop7 = call i8 @llvm.ctpop.i8(i8 %ld7)
+ %ctpop8 = call i8 @llvm.ctpop.i8(i8 %ld8)
+ %ctpop9 = call i8 @llvm.ctpop.i8(i8 %ld9)
+ %ctpop10 = call i8 @llvm.ctpop.i8(i8 %ld10)
+ %ctpop11 = call i8 @llvm.ctpop.i8(i8 %ld11)
+ %ctpop12 = call i8 @llvm.ctpop.i8(i8 %ld12)
+ %ctpop13 = call i8 @llvm.ctpop.i8(i8 %ld13)
+ %ctpop14 = call i8 @llvm.ctpop.i8(i8 %ld14)
+ %ctpop15 = call i8 @llvm.ctpop.i8(i8 %ld15)
+ store i8 %ctpop0 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 0), align 1
+ store i8 %ctpop1 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 1), align 1
+ store i8 %ctpop2 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 2), align 1
+ store i8 %ctpop3 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 3), align 1
+ store i8 %ctpop4 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 4), align 1
+ store i8 %ctpop5 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 5), align 1
+ store i8 %ctpop6 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 6), align 1
+ store i8 %ctpop7 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 7), align 1
+ store i8 %ctpop8 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 8), align 1
+ store i8 %ctpop9 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 9), align 1
+ store i8 %ctpop10, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 10), align 1
+ store i8 %ctpop11, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 11), align 1
+ store i8 %ctpop12, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 12), align 1
+ store i8 %ctpop13, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 13), align 1
+ store i8 %ctpop14, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 14), align 1
+ store i8 %ctpop15, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 15), align 1
+ ret void
+}
+
+define void @ctpop_32i8() #0 {
+; CHECK-LABEL: @ctpop_32i8(
+; CHECK-NEXT: [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* bitcast ([32 x i8]* @src8 to <16 x i8>*), align 1
+; CHECK-NEXT: [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* bitcast (i8* getelementptr inbounds ([32 x i8], [32 x i8]* @src8, i8 0, i64 16) to <16 x i8>*), align 1
+; CHECK-NEXT: [[TMP3:%.*]] = call <16 x i8> @llvm.ctpop.v16i8(<16 x i8> [[TMP1]])
+; CHECK-NEXT: [[TMP4:%.*]] = call <16 x i8> @llvm.ctpop.v16i8(<16 x i8> [[TMP2]])
+; CHECK-NEXT: store <16 x i8> [[TMP3]], <16 x i8>* bitcast ([32 x i8]* @dst8 to <16 x i8>*), align 1
+; CHECK-NEXT: store <16 x i8> [[TMP4]], <16 x i8>* bitcast (i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 16) to <16 x i8>*), 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
+ %ctpop0 = call i8 @llvm.ctpop.i8(i8 %ld0)
+ %ctpop1 = call i8 @llvm.ctpop.i8(i8 %ld1)
+ %ctpop2 = call i8 @llvm.ctpop.i8(i8 %ld2)
+ %ctpop3 = call i8 @llvm.ctpop.i8(i8 %ld3)
+ %ctpop4 = call i8 @llvm.ctpop.i8(i8 %ld4)
+ %ctpop5 = call i8 @llvm.ctpop.i8(i8 %ld5)
+ %ctpop6 = call i8 @llvm.ctpop.i8(i8 %ld6)
+ %ctpop7 = call i8 @llvm.ctpop.i8(i8 %ld7)
+ %ctpop8 = call i8 @llvm.ctpop.i8(i8 %ld8)
+ %ctpop9 = call i8 @llvm.ctpop.i8(i8 %ld9)
+ %ctpop10 = call i8 @llvm.ctpop.i8(i8 %ld10)
+ %ctpop11 = call i8 @llvm.ctpop.i8(i8 %ld11)
+ %ctpop12 = call i8 @llvm.ctpop.i8(i8 %ld12)
+ %ctpop13 = call i8 @llvm.ctpop.i8(i8 %ld13)
+ %ctpop14 = call i8 @llvm.ctpop.i8(i8 %ld14)
+ %ctpop15 = call i8 @llvm.ctpop.i8(i8 %ld15)
+ %ctpop16 = call i8 @llvm.ctpop.i8(i8 %ld16)
+ %ctpop17 = call i8 @llvm.ctpop.i8(i8 %ld17)
+ %ctpop18 = call i8 @llvm.ctpop.i8(i8 %ld18)
+ %ctpop19 = call i8 @llvm.ctpop.i8(i8 %ld19)
+ %ctpop20 = call i8 @llvm.ctpop.i8(i8 %ld20)
+ %ctpop21 = call i8 @llvm.ctpop.i8(i8 %ld21)
+ %ctpop22 = call i8 @llvm.ctpop.i8(i8 %ld22)
+ %ctpop23 = call i8 @llvm.ctpop.i8(i8 %ld23)
+ %ctpop24 = call i8 @llvm.ctpop.i8(i8 %ld24)
+ %ctpop25 = call i8 @llvm.ctpop.i8(i8 %ld25)
+ %ctpop26 = call i8 @llvm.ctpop.i8(i8 %ld26)
+ %ctpop27 = call i8 @llvm.ctpop.i8(i8 %ld27)
+ %ctpop28 = call i8 @llvm.ctpop.i8(i8 %ld28)
+ %ctpop29 = call i8 @llvm.ctpop.i8(i8 %ld29)
+ %ctpop30 = call i8 @llvm.ctpop.i8(i8 %ld30)
+ %ctpop31 = call i8 @llvm.ctpop.i8(i8 %ld31)
+ store i8 %ctpop0 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 0), align 1
+ store i8 %ctpop1 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 1), align 1
+ store i8 %ctpop2 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 2), align 1
+ store i8 %ctpop3 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 3), align 1
+ store i8 %ctpop4 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 4), align 1
+ store i8 %ctpop5 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 5), align 1
+ store i8 %ctpop6 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 6), align 1
+ store i8 %ctpop7 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 7), align 1
+ store i8 %ctpop8 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 8), align 1
+ store i8 %ctpop9 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 9), align 1
+ store i8 %ctpop10, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 10), align 1
+ store i8 %ctpop11, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 11), align 1
+ store i8 %ctpop12, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 12), align 1
+ store i8 %ctpop13, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 13), align 1
+ store i8 %ctpop14, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 14), align 1
+ store i8 %ctpop15, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 15), align 1
+ store i8 %ctpop16, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 16), align 1
+ store i8 %ctpop17, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 17), align 1
+ store i8 %ctpop18, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 18), align 1
+ store i8 %ctpop19, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 19), align 1
+ store i8 %ctpop20, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 20), align 1
+ store i8 %ctpop21, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 21), align 1
+ store i8 %ctpop22, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 22), align 1
+ store i8 %ctpop23, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 23), align 1
+ store i8 %ctpop24, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 24), align 1
+ store i8 %ctpop25, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 25), align 1
+ store i8 %ctpop26, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 26), align 1
+ store i8 %ctpop27, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 27), align 1
+ store i8 %ctpop28, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 28), align 1
+ store i8 %ctpop29, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 29), align 1
+ store i8 %ctpop30, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 30), align 1
+ store i8 %ctpop31, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 31), align 1
+ ret void
+}
+
+attributes #0 = { nounwind }
+
Added: llvm/trunk/test/Transforms/SLPVectorizer/X86/cttz.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Transforms/SLPVectorizer/X86/cttz.ll?rev=268800&view=auto
==============================================================================
--- llvm/trunk/test/Transforms/SLPVectorizer/X86/cttz.ll (added)
+++ llvm/trunk/test/Transforms/SLPVectorizer/X86/cttz.ll Fri May 6 16:33:01 2016
@@ -0,0 +1,1222 @@
+; 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"
+
+ at src64 = common global [4 x i64] zeroinitializer, align 32
+ at dst64 = common global [4 x i64] zeroinitializer, align 32
+ at src32 = common global [8 x i32] zeroinitializer, align 32
+ at dst32 = common global [8 x i32] zeroinitializer, align 32
+ at src16 = common global [16 x i16] zeroinitializer, align 32
+ at dst16 = common global [16 x i16] zeroinitializer, align 32
+ at src8 = common global [32 x i8] zeroinitializer, align 32
+ at dst8 = common global [32 x i8] zeroinitializer, align 32
+
+declare i64 @llvm.cttz.i64(i64, i1)
+declare i32 @llvm.cttz.i32(i32, i1)
+declare i16 @llvm.cttz.i16(i16, i1)
+declare i8 @llvm.cttz.i8(i8, i1)
+
+;
+; CTTZ
+;
+
+define void @cttz_2i64() #0 {
+; CHECK-LABEL: @cttz_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: [[CTTZ0:%.*]] = call i64 @llvm.cttz.i64(i64 [[LD0]], i1 false)
+; CHECK-NEXT: [[CTTZ1:%.*]] = call i64 @llvm.cttz.i64(i64 [[LD1]], i1 false)
+; CHECK-NEXT: store i64 [[CTTZ0]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i32 0, i64 0), align 8
+; CHECK-NEXT: store i64 [[CTTZ1]], 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
+ %cttz0 = call i64 @llvm.cttz.i64(i64 %ld0, i1 0)
+ %cttz1 = call i64 @llvm.cttz.i64(i64 %ld1, i1 0)
+ store i64 %cttz0, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i32 0, i64 0), align 8
+ store i64 %cttz1, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i32 0, i64 1), align 8
+ ret void
+}
+
+define void @cttz_4i64() #0 {
+; CHECK-LABEL: @cttz_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: [[CTTZ0:%.*]] = call i64 @llvm.cttz.i64(i64 [[LD0]], i1 false)
+; CHECK-NEXT: [[CTTZ1:%.*]] = call i64 @llvm.cttz.i64(i64 [[LD1]], i1 false)
+; CHECK-NEXT: [[CTTZ2:%.*]] = call i64 @llvm.cttz.i64(i64 [[LD2]], i1 false)
+; CHECK-NEXT: [[CTTZ3:%.*]] = call i64 @llvm.cttz.i64(i64 [[LD3]], i1 false)
+; CHECK-NEXT: store i64 [[CTTZ0]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 0), align 4
+; CHECK-NEXT: store i64 [[CTTZ1]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 1), align 4
+; CHECK-NEXT: store i64 [[CTTZ2]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 2), align 4
+; CHECK-NEXT: store i64 [[CTTZ3]], 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
+ %cttz0 = call i64 @llvm.cttz.i64(i64 %ld0, i1 0)
+ %cttz1 = call i64 @llvm.cttz.i64(i64 %ld1, i1 0)
+ %cttz2 = call i64 @llvm.cttz.i64(i64 %ld2, i1 0)
+ %cttz3 = call i64 @llvm.cttz.i64(i64 %ld3, i1 0)
+ store i64 %cttz0, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 0), align 4
+ store i64 %cttz1, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 1), align 4
+ store i64 %cttz2, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 2), align 4
+ store i64 %cttz3, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 3), align 4
+ ret void
+}
+
+define void @cttz_4i32() #0 {
+; CHECK-LABEL: @cttz_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: [[CTTZ0:%.*]] = call i32 @llvm.cttz.i32(i32 [[LD0]], i1 false)
+; CHECK-NEXT: [[CTTZ1:%.*]] = call i32 @llvm.cttz.i32(i32 [[LD1]], i1 false)
+; CHECK-NEXT: [[CTTZ2:%.*]] = call i32 @llvm.cttz.i32(i32 [[LD2]], i1 false)
+; CHECK-NEXT: [[CTTZ3:%.*]] = call i32 @llvm.cttz.i32(i32 [[LD3]], i1 false)
+; CHECK-NEXT: store i32 [[CTTZ0]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 0), align 4
+; CHECK-NEXT: store i32 [[CTTZ1]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 1), align 4
+; CHECK-NEXT: store i32 [[CTTZ2]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 2), align 4
+; CHECK-NEXT: store i32 [[CTTZ3]], 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
+ %cttz0 = call i32 @llvm.cttz.i32(i32 %ld0, i1 0)
+ %cttz1 = call i32 @llvm.cttz.i32(i32 %ld1, i1 0)
+ %cttz2 = call i32 @llvm.cttz.i32(i32 %ld2, i1 0)
+ %cttz3 = call i32 @llvm.cttz.i32(i32 %ld3, i1 0)
+ store i32 %cttz0, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 0), align 4
+ store i32 %cttz1, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 1), align 4
+ store i32 %cttz2, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 2), align 4
+ store i32 %cttz3, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 3), align 4
+ ret void
+}
+
+define void @cttz_8i32() #0 {
+; CHECK-LABEL: @cttz_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: [[CTTZ0:%.*]] = call i32 @llvm.cttz.i32(i32 [[LD0]], i1 false)
+; CHECK-NEXT: [[CTTZ1:%.*]] = call i32 @llvm.cttz.i32(i32 [[LD1]], i1 false)
+; CHECK-NEXT: [[CTTZ2:%.*]] = call i32 @llvm.cttz.i32(i32 [[LD2]], i1 false)
+; CHECK-NEXT: [[CTTZ3:%.*]] = call i32 @llvm.cttz.i32(i32 [[LD3]], i1 false)
+; CHECK-NEXT: [[CTTZ4:%.*]] = call i32 @llvm.cttz.i32(i32 [[LD4]], i1 false)
+; CHECK-NEXT: [[CTTZ5:%.*]] = call i32 @llvm.cttz.i32(i32 [[LD5]], i1 false)
+; CHECK-NEXT: [[CTTZ6:%.*]] = call i32 @llvm.cttz.i32(i32 [[LD6]], i1 false)
+; CHECK-NEXT: [[CTTZ7:%.*]] = call i32 @llvm.cttz.i32(i32 [[LD7]], i1 false)
+; CHECK-NEXT: store i32 [[CTTZ0]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 0), align 2
+; CHECK-NEXT: store i32 [[CTTZ1]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 1), align 2
+; CHECK-NEXT: store i32 [[CTTZ2]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 2), align 2
+; CHECK-NEXT: store i32 [[CTTZ3]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 3), align 2
+; CHECK-NEXT: store i32 [[CTTZ4]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 4), align 2
+; CHECK-NEXT: store i32 [[CTTZ5]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 5), align 2
+; CHECK-NEXT: store i32 [[CTTZ6]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 6), align 2
+; CHECK-NEXT: store i32 [[CTTZ7]], 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
+ %cttz0 = call i32 @llvm.cttz.i32(i32 %ld0, i1 0)
+ %cttz1 = call i32 @llvm.cttz.i32(i32 %ld1, i1 0)
+ %cttz2 = call i32 @llvm.cttz.i32(i32 %ld2, i1 0)
+ %cttz3 = call i32 @llvm.cttz.i32(i32 %ld3, i1 0)
+ %cttz4 = call i32 @llvm.cttz.i32(i32 %ld4, i1 0)
+ %cttz5 = call i32 @llvm.cttz.i32(i32 %ld5, i1 0)
+ %cttz6 = call i32 @llvm.cttz.i32(i32 %ld6, i1 0)
+ %cttz7 = call i32 @llvm.cttz.i32(i32 %ld7, i1 0)
+ store i32 %cttz0, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 0), align 2
+ store i32 %cttz1, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 1), align 2
+ store i32 %cttz2, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 2), align 2
+ store i32 %cttz3, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 3), align 2
+ store i32 %cttz4, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 4), align 2
+ store i32 %cttz5, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 5), align 2
+ store i32 %cttz6, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 6), align 2
+ store i32 %cttz7, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 7), align 2
+ ret void
+}
+
+define void @cttz_8i16() #0 {
+; CHECK-LABEL: @cttz_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: [[CTTZ0:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD0]], i1 false)
+; CHECK-NEXT: [[CTTZ1:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD1]], i1 false)
+; CHECK-NEXT: [[CTTZ2:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD2]], i1 false)
+; CHECK-NEXT: [[CTTZ3:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD3]], i1 false)
+; CHECK-NEXT: [[CTTZ4:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD4]], i1 false)
+; CHECK-NEXT: [[CTTZ5:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD5]], i1 false)
+; CHECK-NEXT: [[CTTZ6:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD6]], i1 false)
+; CHECK-NEXT: [[CTTZ7:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD7]], i1 false)
+; CHECK-NEXT: store i16 [[CTTZ0]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 0), align 2
+; CHECK-NEXT: store i16 [[CTTZ1]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 1), align 2
+; CHECK-NEXT: store i16 [[CTTZ2]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 2), align 2
+; CHECK-NEXT: store i16 [[CTTZ3]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 3), align 2
+; CHECK-NEXT: store i16 [[CTTZ4]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 4), align 2
+; CHECK-NEXT: store i16 [[CTTZ5]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 5), align 2
+; CHECK-NEXT: store i16 [[CTTZ6]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 6), align 2
+; CHECK-NEXT: store i16 [[CTTZ7]], 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
+ %cttz0 = call i16 @llvm.cttz.i16(i16 %ld0, i1 0)
+ %cttz1 = call i16 @llvm.cttz.i16(i16 %ld1, i1 0)
+ %cttz2 = call i16 @llvm.cttz.i16(i16 %ld2, i1 0)
+ %cttz3 = call i16 @llvm.cttz.i16(i16 %ld3, i1 0)
+ %cttz4 = call i16 @llvm.cttz.i16(i16 %ld4, i1 0)
+ %cttz5 = call i16 @llvm.cttz.i16(i16 %ld5, i1 0)
+ %cttz6 = call i16 @llvm.cttz.i16(i16 %ld6, i1 0)
+ %cttz7 = call i16 @llvm.cttz.i16(i16 %ld7, i1 0)
+ store i16 %cttz0, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 0), align 2
+ store i16 %cttz1, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 1), align 2
+ store i16 %cttz2, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 2), align 2
+ store i16 %cttz3, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 3), align 2
+ store i16 %cttz4, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 4), align 2
+ store i16 %cttz5, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 5), align 2
+ store i16 %cttz6, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 6), align 2
+ store i16 %cttz7, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 7), align 2
+ ret void
+}
+
+define void @cttz_16i16() #0 {
+; CHECK-LABEL: @cttz_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: [[CTTZ0:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD0]], i1 false)
+; CHECK-NEXT: [[CTTZ1:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD1]], i1 false)
+; CHECK-NEXT: [[CTTZ2:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD2]], i1 false)
+; CHECK-NEXT: [[CTTZ3:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD3]], i1 false)
+; CHECK-NEXT: [[CTTZ4:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD4]], i1 false)
+; CHECK-NEXT: [[CTTZ5:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD5]], i1 false)
+; CHECK-NEXT: [[CTTZ6:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD6]], i1 false)
+; CHECK-NEXT: [[CTTZ7:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD7]], i1 false)
+; CHECK-NEXT: [[CTTZ8:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD8]], i1 false)
+; CHECK-NEXT: [[CTTZ9:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD9]], i1 false)
+; CHECK-NEXT: [[CTTZ10:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD10]], i1 false)
+; CHECK-NEXT: [[CTTZ11:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD11]], i1 false)
+; CHECK-NEXT: [[CTTZ12:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD12]], i1 false)
+; CHECK-NEXT: [[CTTZ13:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD13]], i1 false)
+; CHECK-NEXT: [[CTTZ14:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD14]], i1 false)
+; CHECK-NEXT: [[CTTZ15:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD15]], i1 false)
+; CHECK-NEXT: store i16 [[CTTZ0]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 0), align 2
+; CHECK-NEXT: store i16 [[CTTZ1]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 1), align 2
+; CHECK-NEXT: store i16 [[CTTZ2]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 2), align 2
+; CHECK-NEXT: store i16 [[CTTZ3]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 3), align 2
+; CHECK-NEXT: store i16 [[CTTZ4]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 4), align 2
+; CHECK-NEXT: store i16 [[CTTZ5]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 5), align 2
+; CHECK-NEXT: store i16 [[CTTZ6]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 6), align 2
+; CHECK-NEXT: store i16 [[CTTZ7]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 7), align 2
+; CHECK-NEXT: store i16 [[CTTZ8]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 8), align 2
+; CHECK-NEXT: store i16 [[CTTZ9]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 9), align 2
+; CHECK-NEXT: store i16 [[CTTZ10]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 10), align 2
+; CHECK-NEXT: store i16 [[CTTZ11]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 11), align 2
+; CHECK-NEXT: store i16 [[CTTZ12]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 12), align 2
+; CHECK-NEXT: store i16 [[CTTZ13]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 13), align 2
+; CHECK-NEXT: store i16 [[CTTZ14]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 14), align 2
+; CHECK-NEXT: store i16 [[CTTZ15]], 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
+ %cttz0 = call i16 @llvm.cttz.i16(i16 %ld0, i1 0)
+ %cttz1 = call i16 @llvm.cttz.i16(i16 %ld1, i1 0)
+ %cttz2 = call i16 @llvm.cttz.i16(i16 %ld2, i1 0)
+ %cttz3 = call i16 @llvm.cttz.i16(i16 %ld3, i1 0)
+ %cttz4 = call i16 @llvm.cttz.i16(i16 %ld4, i1 0)
+ %cttz5 = call i16 @llvm.cttz.i16(i16 %ld5, i1 0)
+ %cttz6 = call i16 @llvm.cttz.i16(i16 %ld6, i1 0)
+ %cttz7 = call i16 @llvm.cttz.i16(i16 %ld7, i1 0)
+ %cttz8 = call i16 @llvm.cttz.i16(i16 %ld8, i1 0)
+ %cttz9 = call i16 @llvm.cttz.i16(i16 %ld9, i1 0)
+ %cttz10 = call i16 @llvm.cttz.i16(i16 %ld10, i1 0)
+ %cttz11 = call i16 @llvm.cttz.i16(i16 %ld11, i1 0)
+ %cttz12 = call i16 @llvm.cttz.i16(i16 %ld12, i1 0)
+ %cttz13 = call i16 @llvm.cttz.i16(i16 %ld13, i1 0)
+ %cttz14 = call i16 @llvm.cttz.i16(i16 %ld14, i1 0)
+ %cttz15 = call i16 @llvm.cttz.i16(i16 %ld15, i1 0)
+ store i16 %cttz0 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 0), align 2
+ store i16 %cttz1 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 1), align 2
+ store i16 %cttz2 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 2), align 2
+ store i16 %cttz3 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 3), align 2
+ store i16 %cttz4 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 4), align 2
+ store i16 %cttz5 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 5), align 2
+ store i16 %cttz6 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 6), align 2
+ store i16 %cttz7 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 7), align 2
+ store i16 %cttz8 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 8), align 2
+ store i16 %cttz9 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 9), align 2
+ store i16 %cttz10, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 10), align 2
+ store i16 %cttz11, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 11), align 2
+ store i16 %cttz12, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 12), align 2
+ store i16 %cttz13, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 13), align 2
+ store i16 %cttz14, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 14), align 2
+ store i16 %cttz15, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 15), align 2
+ ret void
+}
+
+define void @cttz_16i8() #0 {
+; CHECK-LABEL: @cttz_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: [[CTTZ0:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD0]], i1 false)
+; CHECK-NEXT: [[CTTZ1:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD1]], i1 false)
+; CHECK-NEXT: [[CTTZ2:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD2]], i1 false)
+; CHECK-NEXT: [[CTTZ3:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD3]], i1 false)
+; CHECK-NEXT: [[CTTZ4:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD4]], i1 false)
+; CHECK-NEXT: [[CTTZ5:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD5]], i1 false)
+; CHECK-NEXT: [[CTTZ6:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD6]], i1 false)
+; CHECK-NEXT: [[CTTZ7:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD7]], i1 false)
+; CHECK-NEXT: [[CTTZ8:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD8]], i1 false)
+; CHECK-NEXT: [[CTTZ9:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD9]], i1 false)
+; CHECK-NEXT: [[CTTZ10:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD10]], i1 false)
+; CHECK-NEXT: [[CTTZ11:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD11]], i1 false)
+; CHECK-NEXT: [[CTTZ12:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD12]], i1 false)
+; CHECK-NEXT: [[CTTZ13:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD13]], i1 false)
+; CHECK-NEXT: [[CTTZ14:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD14]], i1 false)
+; CHECK-NEXT: [[CTTZ15:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD15]], i1 false)
+; CHECK-NEXT: store i8 [[CTTZ0]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 0), align 1
+; CHECK-NEXT: store i8 [[CTTZ1]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 1), align 1
+; CHECK-NEXT: store i8 [[CTTZ2]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 2), align 1
+; CHECK-NEXT: store i8 [[CTTZ3]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 3), align 1
+; CHECK-NEXT: store i8 [[CTTZ4]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 4), align 1
+; CHECK-NEXT: store i8 [[CTTZ5]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 5), align 1
+; CHECK-NEXT: store i8 [[CTTZ6]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 6), align 1
+; CHECK-NEXT: store i8 [[CTTZ7]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 7), align 1
+; CHECK-NEXT: store i8 [[CTTZ8]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 8), align 1
+; CHECK-NEXT: store i8 [[CTTZ9]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 9), align 1
+; CHECK-NEXT: store i8 [[CTTZ10]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 10), align 1
+; CHECK-NEXT: store i8 [[CTTZ11]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 11), align 1
+; CHECK-NEXT: store i8 [[CTTZ12]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 12), align 1
+; CHECK-NEXT: store i8 [[CTTZ13]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 13), align 1
+; CHECK-NEXT: store i8 [[CTTZ14]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 14), align 1
+; CHECK-NEXT: store i8 [[CTTZ15]], 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
+ %cttz0 = call i8 @llvm.cttz.i8(i8 %ld0, i1 0)
+ %cttz1 = call i8 @llvm.cttz.i8(i8 %ld1, i1 0)
+ %cttz2 = call i8 @llvm.cttz.i8(i8 %ld2, i1 0)
+ %cttz3 = call i8 @llvm.cttz.i8(i8 %ld3, i1 0)
+ %cttz4 = call i8 @llvm.cttz.i8(i8 %ld4, i1 0)
+ %cttz5 = call i8 @llvm.cttz.i8(i8 %ld5, i1 0)
+ %cttz6 = call i8 @llvm.cttz.i8(i8 %ld6, i1 0)
+ %cttz7 = call i8 @llvm.cttz.i8(i8 %ld7, i1 0)
+ %cttz8 = call i8 @llvm.cttz.i8(i8 %ld8, i1 0)
+ %cttz9 = call i8 @llvm.cttz.i8(i8 %ld9, i1 0)
+ %cttz10 = call i8 @llvm.cttz.i8(i8 %ld10, i1 0)
+ %cttz11 = call i8 @llvm.cttz.i8(i8 %ld11, i1 0)
+ %cttz12 = call i8 @llvm.cttz.i8(i8 %ld12, i1 0)
+ %cttz13 = call i8 @llvm.cttz.i8(i8 %ld13, i1 0)
+ %cttz14 = call i8 @llvm.cttz.i8(i8 %ld14, i1 0)
+ %cttz15 = call i8 @llvm.cttz.i8(i8 %ld15, i1 0)
+ store i8 %cttz0 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 0), align 1
+ store i8 %cttz1 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 1), align 1
+ store i8 %cttz2 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 2), align 1
+ store i8 %cttz3 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 3), align 1
+ store i8 %cttz4 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 4), align 1
+ store i8 %cttz5 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 5), align 1
+ store i8 %cttz6 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 6), align 1
+ store i8 %cttz7 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 7), align 1
+ store i8 %cttz8 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 8), align 1
+ store i8 %cttz9 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 9), align 1
+ store i8 %cttz10, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 10), align 1
+ store i8 %cttz11, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 11), align 1
+ store i8 %cttz12, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 12), align 1
+ store i8 %cttz13, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 13), align 1
+ store i8 %cttz14, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 14), align 1
+ store i8 %cttz15, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 15), align 1
+ ret void
+}
+
+define void @cttz_32i8() #0 {
+; CHECK-LABEL: @cttz_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: [[CTTZ0:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD0]], i1 false)
+; CHECK-NEXT: [[CTTZ1:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD1]], i1 false)
+; CHECK-NEXT: [[CTTZ2:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD2]], i1 false)
+; CHECK-NEXT: [[CTTZ3:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD3]], i1 false)
+; CHECK-NEXT: [[CTTZ4:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD4]], i1 false)
+; CHECK-NEXT: [[CTTZ5:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD5]], i1 false)
+; CHECK-NEXT: [[CTTZ6:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD6]], i1 false)
+; CHECK-NEXT: [[CTTZ7:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD7]], i1 false)
+; CHECK-NEXT: [[CTTZ8:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD8]], i1 false)
+; CHECK-NEXT: [[CTTZ9:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD9]], i1 false)
+; CHECK-NEXT: [[CTTZ10:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD10]], i1 false)
+; CHECK-NEXT: [[CTTZ11:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD11]], i1 false)
+; CHECK-NEXT: [[CTTZ12:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD12]], i1 false)
+; CHECK-NEXT: [[CTTZ13:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD13]], i1 false)
+; CHECK-NEXT: [[CTTZ14:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD14]], i1 false)
+; CHECK-NEXT: [[CTTZ15:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD15]], i1 false)
+; CHECK-NEXT: [[CTTZ16:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD16]], i1 false)
+; CHECK-NEXT: [[CTTZ17:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD17]], i1 false)
+; CHECK-NEXT: [[CTTZ18:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD18]], i1 false)
+; CHECK-NEXT: [[CTTZ19:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD19]], i1 false)
+; CHECK-NEXT: [[CTTZ20:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD20]], i1 false)
+; CHECK-NEXT: [[CTTZ21:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD21]], i1 false)
+; CHECK-NEXT: [[CTTZ22:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD22]], i1 false)
+; CHECK-NEXT: [[CTTZ23:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD23]], i1 false)
+; CHECK-NEXT: [[CTTZ24:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD24]], i1 false)
+; CHECK-NEXT: [[CTTZ25:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD25]], i1 false)
+; CHECK-NEXT: [[CTTZ26:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD26]], i1 false)
+; CHECK-NEXT: [[CTTZ27:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD27]], i1 false)
+; CHECK-NEXT: [[CTTZ28:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD28]], i1 false)
+; CHECK-NEXT: [[CTTZ29:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD29]], i1 false)
+; CHECK-NEXT: [[CTTZ30:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD30]], i1 false)
+; CHECK-NEXT: [[CTTZ31:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD31]], i1 false)
+; CHECK-NEXT: store i8 [[CTTZ0]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 0), align 1
+; CHECK-NEXT: store i8 [[CTTZ1]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 1), align 1
+; CHECK-NEXT: store i8 [[CTTZ2]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 2), align 1
+; CHECK-NEXT: store i8 [[CTTZ3]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 3), align 1
+; CHECK-NEXT: store i8 [[CTTZ4]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 4), align 1
+; CHECK-NEXT: store i8 [[CTTZ5]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 5), align 1
+; CHECK-NEXT: store i8 [[CTTZ6]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 6), align 1
+; CHECK-NEXT: store i8 [[CTTZ7]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 7), align 1
+; CHECK-NEXT: store i8 [[CTTZ8]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 8), align 1
+; CHECK-NEXT: store i8 [[CTTZ9]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 9), align 1
+; CHECK-NEXT: store i8 [[CTTZ10]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 10), align 1
+; CHECK-NEXT: store i8 [[CTTZ11]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 11), align 1
+; CHECK-NEXT: store i8 [[CTTZ12]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 12), align 1
+; CHECK-NEXT: store i8 [[CTTZ13]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 13), align 1
+; CHECK-NEXT: store i8 [[CTTZ14]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 14), align 1
+; CHECK-NEXT: store i8 [[CTTZ15]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 15), align 1
+; CHECK-NEXT: store i8 [[CTTZ16]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 16), align 1
+; CHECK-NEXT: store i8 [[CTTZ17]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 17), align 1
+; CHECK-NEXT: store i8 [[CTTZ18]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 18), align 1
+; CHECK-NEXT: store i8 [[CTTZ19]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 19), align 1
+; CHECK-NEXT: store i8 [[CTTZ20]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 20), align 1
+; CHECK-NEXT: store i8 [[CTTZ21]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 21), align 1
+; CHECK-NEXT: store i8 [[CTTZ22]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 22), align 1
+; CHECK-NEXT: store i8 [[CTTZ23]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 23), align 1
+; CHECK-NEXT: store i8 [[CTTZ24]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 24), align 1
+; CHECK-NEXT: store i8 [[CTTZ25]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 25), align 1
+; CHECK-NEXT: store i8 [[CTTZ26]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 26), align 1
+; CHECK-NEXT: store i8 [[CTTZ27]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 27), align 1
+; CHECK-NEXT: store i8 [[CTTZ28]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 28), align 1
+; CHECK-NEXT: store i8 [[CTTZ29]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 29), align 1
+; CHECK-NEXT: store i8 [[CTTZ30]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 30), align 1
+; CHECK-NEXT: store i8 [[CTTZ31]], 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
+ %cttz0 = call i8 @llvm.cttz.i8(i8 %ld0, i1 0)
+ %cttz1 = call i8 @llvm.cttz.i8(i8 %ld1, i1 0)
+ %cttz2 = call i8 @llvm.cttz.i8(i8 %ld2, i1 0)
+ %cttz3 = call i8 @llvm.cttz.i8(i8 %ld3, i1 0)
+ %cttz4 = call i8 @llvm.cttz.i8(i8 %ld4, i1 0)
+ %cttz5 = call i8 @llvm.cttz.i8(i8 %ld5, i1 0)
+ %cttz6 = call i8 @llvm.cttz.i8(i8 %ld6, i1 0)
+ %cttz7 = call i8 @llvm.cttz.i8(i8 %ld7, i1 0)
+ %cttz8 = call i8 @llvm.cttz.i8(i8 %ld8, i1 0)
+ %cttz9 = call i8 @llvm.cttz.i8(i8 %ld9, i1 0)
+ %cttz10 = call i8 @llvm.cttz.i8(i8 %ld10, i1 0)
+ %cttz11 = call i8 @llvm.cttz.i8(i8 %ld11, i1 0)
+ %cttz12 = call i8 @llvm.cttz.i8(i8 %ld12, i1 0)
+ %cttz13 = call i8 @llvm.cttz.i8(i8 %ld13, i1 0)
+ %cttz14 = call i8 @llvm.cttz.i8(i8 %ld14, i1 0)
+ %cttz15 = call i8 @llvm.cttz.i8(i8 %ld15, i1 0)
+ %cttz16 = call i8 @llvm.cttz.i8(i8 %ld16, i1 0)
+ %cttz17 = call i8 @llvm.cttz.i8(i8 %ld17, i1 0)
+ %cttz18 = call i8 @llvm.cttz.i8(i8 %ld18, i1 0)
+ %cttz19 = call i8 @llvm.cttz.i8(i8 %ld19, i1 0)
+ %cttz20 = call i8 @llvm.cttz.i8(i8 %ld20, i1 0)
+ %cttz21 = call i8 @llvm.cttz.i8(i8 %ld21, i1 0)
+ %cttz22 = call i8 @llvm.cttz.i8(i8 %ld22, i1 0)
+ %cttz23 = call i8 @llvm.cttz.i8(i8 %ld23, i1 0)
+ %cttz24 = call i8 @llvm.cttz.i8(i8 %ld24, i1 0)
+ %cttz25 = call i8 @llvm.cttz.i8(i8 %ld25, i1 0)
+ %cttz26 = call i8 @llvm.cttz.i8(i8 %ld26, i1 0)
+ %cttz27 = call i8 @llvm.cttz.i8(i8 %ld27, i1 0)
+ %cttz28 = call i8 @llvm.cttz.i8(i8 %ld28, i1 0)
+ %cttz29 = call i8 @llvm.cttz.i8(i8 %ld29, i1 0)
+ %cttz30 = call i8 @llvm.cttz.i8(i8 %ld30, i1 0)
+ %cttz31 = call i8 @llvm.cttz.i8(i8 %ld31, i1 0)
+ store i8 %cttz0 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 0), align 1
+ store i8 %cttz1 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 1), align 1
+ store i8 %cttz2 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 2), align 1
+ store i8 %cttz3 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 3), align 1
+ store i8 %cttz4 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 4), align 1
+ store i8 %cttz5 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 5), align 1
+ store i8 %cttz6 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 6), align 1
+ store i8 %cttz7 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 7), align 1
+ store i8 %cttz8 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 8), align 1
+ store i8 %cttz9 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 9), align 1
+ store i8 %cttz10, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 10), align 1
+ store i8 %cttz11, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 11), align 1
+ store i8 %cttz12, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 12), align 1
+ store i8 %cttz13, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 13), align 1
+ store i8 %cttz14, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 14), align 1
+ store i8 %cttz15, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 15), align 1
+ store i8 %cttz16, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 16), align 1
+ store i8 %cttz17, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 17), align 1
+ store i8 %cttz18, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 18), align 1
+ store i8 %cttz19, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 19), align 1
+ store i8 %cttz20, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 20), align 1
+ store i8 %cttz21, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 21), align 1
+ store i8 %cttz22, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 22), align 1
+ store i8 %cttz23, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 23), align 1
+ store i8 %cttz24, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 24), align 1
+ store i8 %cttz25, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 25), align 1
+ store i8 %cttz26, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 26), align 1
+ store i8 %cttz27, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 27), align 1
+ store i8 %cttz28, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 28), align 1
+ store i8 %cttz29, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 29), align 1
+ store i8 %cttz30, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 30), align 1
+ store i8 %cttz31, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 31), align 1
+ ret void
+}
+
+;
+; CTTZ_ZERO_UNDEF
+;
+
+define void @cttz_undef_2i64() #0 {
+; CHECK-LABEL: @cttz_undef_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: [[CTTZ0:%.*]] = call i64 @llvm.cttz.i64(i64 [[LD0]], i1 true)
+; CHECK-NEXT: [[CTTZ1:%.*]] = call i64 @llvm.cttz.i64(i64 [[LD1]], i1 true)
+; CHECK-NEXT: store i64 [[CTTZ0]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i32 0, i64 0), align 8
+; CHECK-NEXT: store i64 [[CTTZ1]], 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
+ %cttz0 = call i64 @llvm.cttz.i64(i64 %ld0, i1 -1)
+ %cttz1 = call i64 @llvm.cttz.i64(i64 %ld1, i1 -1)
+ store i64 %cttz0, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i32 0, i64 0), align 8
+ store i64 %cttz1, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i32 0, i64 1), align 8
+ ret void
+}
+
+define void @cttz_undef_4i64() #0 {
+; CHECK-LABEL: @cttz_undef_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: [[CTTZ0:%.*]] = call i64 @llvm.cttz.i64(i64 [[LD0]], i1 true)
+; CHECK-NEXT: [[CTTZ1:%.*]] = call i64 @llvm.cttz.i64(i64 [[LD1]], i1 true)
+; CHECK-NEXT: [[CTTZ2:%.*]] = call i64 @llvm.cttz.i64(i64 [[LD2]], i1 true)
+; CHECK-NEXT: [[CTTZ3:%.*]] = call i64 @llvm.cttz.i64(i64 [[LD3]], i1 true)
+; CHECK-NEXT: store i64 [[CTTZ0]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 0), align 4
+; CHECK-NEXT: store i64 [[CTTZ1]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 1), align 4
+; CHECK-NEXT: store i64 [[CTTZ2]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 2), align 4
+; CHECK-NEXT: store i64 [[CTTZ3]], 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
+ %cttz0 = call i64 @llvm.cttz.i64(i64 %ld0, i1 -1)
+ %cttz1 = call i64 @llvm.cttz.i64(i64 %ld1, i1 -1)
+ %cttz2 = call i64 @llvm.cttz.i64(i64 %ld2, i1 -1)
+ %cttz3 = call i64 @llvm.cttz.i64(i64 %ld3, i1 -1)
+ store i64 %cttz0, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 0), align 4
+ store i64 %cttz1, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 1), align 4
+ store i64 %cttz2, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 2), align 4
+ store i64 %cttz3, i64* getelementptr inbounds ([4 x i64], [4 x i64]* @dst64, i64 0, i64 3), align 4
+ ret void
+}
+
+define void @cttz_undef_4i32() #0 {
+; CHECK-LABEL: @cttz_undef_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: [[CTTZ0:%.*]] = call i32 @llvm.cttz.i32(i32 [[LD0]], i1 true)
+; CHECK-NEXT: [[CTTZ1:%.*]] = call i32 @llvm.cttz.i32(i32 [[LD1]], i1 true)
+; CHECK-NEXT: [[CTTZ2:%.*]] = call i32 @llvm.cttz.i32(i32 [[LD2]], i1 true)
+; CHECK-NEXT: [[CTTZ3:%.*]] = call i32 @llvm.cttz.i32(i32 [[LD3]], i1 true)
+; CHECK-NEXT: store i32 [[CTTZ0]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 0), align 4
+; CHECK-NEXT: store i32 [[CTTZ1]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 1), align 4
+; CHECK-NEXT: store i32 [[CTTZ2]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 2), align 4
+; CHECK-NEXT: store i32 [[CTTZ3]], 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
+ %cttz0 = call i32 @llvm.cttz.i32(i32 %ld0, i1 -1)
+ %cttz1 = call i32 @llvm.cttz.i32(i32 %ld1, i1 -1)
+ %cttz2 = call i32 @llvm.cttz.i32(i32 %ld2, i1 -1)
+ %cttz3 = call i32 @llvm.cttz.i32(i32 %ld3, i1 -1)
+ store i32 %cttz0, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 0), align 4
+ store i32 %cttz1, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 1), align 4
+ store i32 %cttz2, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 2), align 4
+ store i32 %cttz3, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 3), align 4
+ ret void
+}
+
+define void @cttz_undef_8i32() #0 {
+; CHECK-LABEL: @cttz_undef_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: [[CTTZ0:%.*]] = call i32 @llvm.cttz.i32(i32 [[LD0]], i1 true)
+; CHECK-NEXT: [[CTTZ1:%.*]] = call i32 @llvm.cttz.i32(i32 [[LD1]], i1 true)
+; CHECK-NEXT: [[CTTZ2:%.*]] = call i32 @llvm.cttz.i32(i32 [[LD2]], i1 true)
+; CHECK-NEXT: [[CTTZ3:%.*]] = call i32 @llvm.cttz.i32(i32 [[LD3]], i1 true)
+; CHECK-NEXT: [[CTTZ4:%.*]] = call i32 @llvm.cttz.i32(i32 [[LD4]], i1 true)
+; CHECK-NEXT: [[CTTZ5:%.*]] = call i32 @llvm.cttz.i32(i32 [[LD5]], i1 true)
+; CHECK-NEXT: [[CTTZ6:%.*]] = call i32 @llvm.cttz.i32(i32 [[LD6]], i1 true)
+; CHECK-NEXT: [[CTTZ7:%.*]] = call i32 @llvm.cttz.i32(i32 [[LD7]], i1 true)
+; CHECK-NEXT: store i32 [[CTTZ0]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 0), align 2
+; CHECK-NEXT: store i32 [[CTTZ1]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 1), align 2
+; CHECK-NEXT: store i32 [[CTTZ2]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 2), align 2
+; CHECK-NEXT: store i32 [[CTTZ3]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 3), align 2
+; CHECK-NEXT: store i32 [[CTTZ4]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 4), align 2
+; CHECK-NEXT: store i32 [[CTTZ5]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 5), align 2
+; CHECK-NEXT: store i32 [[CTTZ6]], i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 6), align 2
+; CHECK-NEXT: store i32 [[CTTZ7]], 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
+ %cttz0 = call i32 @llvm.cttz.i32(i32 %ld0, i1 -1)
+ %cttz1 = call i32 @llvm.cttz.i32(i32 %ld1, i1 -1)
+ %cttz2 = call i32 @llvm.cttz.i32(i32 %ld2, i1 -1)
+ %cttz3 = call i32 @llvm.cttz.i32(i32 %ld3, i1 -1)
+ %cttz4 = call i32 @llvm.cttz.i32(i32 %ld4, i1 -1)
+ %cttz5 = call i32 @llvm.cttz.i32(i32 %ld5, i1 -1)
+ %cttz6 = call i32 @llvm.cttz.i32(i32 %ld6, i1 -1)
+ %cttz7 = call i32 @llvm.cttz.i32(i32 %ld7, i1 -1)
+ store i32 %cttz0, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 0), align 2
+ store i32 %cttz1, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 1), align 2
+ store i32 %cttz2, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 2), align 2
+ store i32 %cttz3, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 3), align 2
+ store i32 %cttz4, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 4), align 2
+ store i32 %cttz5, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 5), align 2
+ store i32 %cttz6, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 6), align 2
+ store i32 %cttz7, i32* getelementptr inbounds ([8 x i32], [8 x i32]* @dst32, i32 0, i64 7), align 2
+ ret void
+}
+
+define void @cttz_undef_8i16() #0 {
+; CHECK-LABEL: @cttz_undef_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: [[CTTZ0:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD0]], i1 true)
+; CHECK-NEXT: [[CTTZ1:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD1]], i1 true)
+; CHECK-NEXT: [[CTTZ2:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD2]], i1 true)
+; CHECK-NEXT: [[CTTZ3:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD3]], i1 true)
+; CHECK-NEXT: [[CTTZ4:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD4]], i1 true)
+; CHECK-NEXT: [[CTTZ5:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD5]], i1 true)
+; CHECK-NEXT: [[CTTZ6:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD6]], i1 true)
+; CHECK-NEXT: [[CTTZ7:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD7]], i1 true)
+; CHECK-NEXT: store i16 [[CTTZ0]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 0), align 2
+; CHECK-NEXT: store i16 [[CTTZ1]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 1), align 2
+; CHECK-NEXT: store i16 [[CTTZ2]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 2), align 2
+; CHECK-NEXT: store i16 [[CTTZ3]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 3), align 2
+; CHECK-NEXT: store i16 [[CTTZ4]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 4), align 2
+; CHECK-NEXT: store i16 [[CTTZ5]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 5), align 2
+; CHECK-NEXT: store i16 [[CTTZ6]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 6), align 2
+; CHECK-NEXT: store i16 [[CTTZ7]], 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
+ %cttz0 = call i16 @llvm.cttz.i16(i16 %ld0, i1 -1)
+ %cttz1 = call i16 @llvm.cttz.i16(i16 %ld1, i1 -1)
+ %cttz2 = call i16 @llvm.cttz.i16(i16 %ld2, i1 -1)
+ %cttz3 = call i16 @llvm.cttz.i16(i16 %ld3, i1 -1)
+ %cttz4 = call i16 @llvm.cttz.i16(i16 %ld4, i1 -1)
+ %cttz5 = call i16 @llvm.cttz.i16(i16 %ld5, i1 -1)
+ %cttz6 = call i16 @llvm.cttz.i16(i16 %ld6, i1 -1)
+ %cttz7 = call i16 @llvm.cttz.i16(i16 %ld7, i1 -1)
+ store i16 %cttz0, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 0), align 2
+ store i16 %cttz1, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 1), align 2
+ store i16 %cttz2, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 2), align 2
+ store i16 %cttz3, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 3), align 2
+ store i16 %cttz4, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 4), align 2
+ store i16 %cttz5, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 5), align 2
+ store i16 %cttz6, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 6), align 2
+ store i16 %cttz7, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 7), align 2
+ ret void
+}
+
+define void @cttz_undef_16i16() #0 {
+; CHECK-LABEL: @cttz_undef_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: [[CTTZ0:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD0]], i1 true)
+; CHECK-NEXT: [[CTTZ1:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD1]], i1 true)
+; CHECK-NEXT: [[CTTZ2:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD2]], i1 true)
+; CHECK-NEXT: [[CTTZ3:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD3]], i1 true)
+; CHECK-NEXT: [[CTTZ4:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD4]], i1 true)
+; CHECK-NEXT: [[CTTZ5:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD5]], i1 true)
+; CHECK-NEXT: [[CTTZ6:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD6]], i1 true)
+; CHECK-NEXT: [[CTTZ7:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD7]], i1 true)
+; CHECK-NEXT: [[CTTZ8:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD8]], i1 true)
+; CHECK-NEXT: [[CTTZ9:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD9]], i1 true)
+; CHECK-NEXT: [[CTTZ10:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD10]], i1 true)
+; CHECK-NEXT: [[CTTZ11:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD11]], i1 true)
+; CHECK-NEXT: [[CTTZ12:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD12]], i1 true)
+; CHECK-NEXT: [[CTTZ13:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD13]], i1 true)
+; CHECK-NEXT: [[CTTZ14:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD14]], i1 true)
+; CHECK-NEXT: [[CTTZ15:%.*]] = call i16 @llvm.cttz.i16(i16 [[LD15]], i1 true)
+; CHECK-NEXT: store i16 [[CTTZ0]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 0), align 2
+; CHECK-NEXT: store i16 [[CTTZ1]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 1), align 2
+; CHECK-NEXT: store i16 [[CTTZ2]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 2), align 2
+; CHECK-NEXT: store i16 [[CTTZ3]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 3), align 2
+; CHECK-NEXT: store i16 [[CTTZ4]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 4), align 2
+; CHECK-NEXT: store i16 [[CTTZ5]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 5), align 2
+; CHECK-NEXT: store i16 [[CTTZ6]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 6), align 2
+; CHECK-NEXT: store i16 [[CTTZ7]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 7), align 2
+; CHECK-NEXT: store i16 [[CTTZ8]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 8), align 2
+; CHECK-NEXT: store i16 [[CTTZ9]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 9), align 2
+; CHECK-NEXT: store i16 [[CTTZ10]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 10), align 2
+; CHECK-NEXT: store i16 [[CTTZ11]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 11), align 2
+; CHECK-NEXT: store i16 [[CTTZ12]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 12), align 2
+; CHECK-NEXT: store i16 [[CTTZ13]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 13), align 2
+; CHECK-NEXT: store i16 [[CTTZ14]], i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 14), align 2
+; CHECK-NEXT: store i16 [[CTTZ15]], 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
+ %cttz0 = call i16 @llvm.cttz.i16(i16 %ld0, i1 -1)
+ %cttz1 = call i16 @llvm.cttz.i16(i16 %ld1, i1 -1)
+ %cttz2 = call i16 @llvm.cttz.i16(i16 %ld2, i1 -1)
+ %cttz3 = call i16 @llvm.cttz.i16(i16 %ld3, i1 -1)
+ %cttz4 = call i16 @llvm.cttz.i16(i16 %ld4, i1 -1)
+ %cttz5 = call i16 @llvm.cttz.i16(i16 %ld5, i1 -1)
+ %cttz6 = call i16 @llvm.cttz.i16(i16 %ld6, i1 -1)
+ %cttz7 = call i16 @llvm.cttz.i16(i16 %ld7, i1 -1)
+ %cttz8 = call i16 @llvm.cttz.i16(i16 %ld8, i1 -1)
+ %cttz9 = call i16 @llvm.cttz.i16(i16 %ld9, i1 -1)
+ %cttz10 = call i16 @llvm.cttz.i16(i16 %ld10, i1 -1)
+ %cttz11 = call i16 @llvm.cttz.i16(i16 %ld11, i1 -1)
+ %cttz12 = call i16 @llvm.cttz.i16(i16 %ld12, i1 -1)
+ %cttz13 = call i16 @llvm.cttz.i16(i16 %ld13, i1 -1)
+ %cttz14 = call i16 @llvm.cttz.i16(i16 %ld14, i1 -1)
+ %cttz15 = call i16 @llvm.cttz.i16(i16 %ld15, i1 -1)
+ store i16 %cttz0 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 0), align 2
+ store i16 %cttz1 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 1), align 2
+ store i16 %cttz2 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 2), align 2
+ store i16 %cttz3 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 3), align 2
+ store i16 %cttz4 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 4), align 2
+ store i16 %cttz5 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 5), align 2
+ store i16 %cttz6 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 6), align 2
+ store i16 %cttz7 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 7), align 2
+ store i16 %cttz8 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 8), align 2
+ store i16 %cttz9 , i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 9), align 2
+ store i16 %cttz10, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 10), align 2
+ store i16 %cttz11, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 11), align 2
+ store i16 %cttz12, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 12), align 2
+ store i16 %cttz13, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 13), align 2
+ store i16 %cttz14, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 14), align 2
+ store i16 %cttz15, i16* getelementptr inbounds ([16 x i16], [16 x i16]* @dst16, i16 0, i64 15), align 2
+ ret void
+}
+
+define void @cttz_undef_16i8() #0 {
+; CHECK-LABEL: @cttz_undef_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: [[CTTZ0:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD0]], i1 true)
+; CHECK-NEXT: [[CTTZ1:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD1]], i1 true)
+; CHECK-NEXT: [[CTTZ2:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD2]], i1 true)
+; CHECK-NEXT: [[CTTZ3:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD3]], i1 true)
+; CHECK-NEXT: [[CTTZ4:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD4]], i1 true)
+; CHECK-NEXT: [[CTTZ5:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD5]], i1 true)
+; CHECK-NEXT: [[CTTZ6:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD6]], i1 true)
+; CHECK-NEXT: [[CTTZ7:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD7]], i1 true)
+; CHECK-NEXT: [[CTTZ8:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD8]], i1 true)
+; CHECK-NEXT: [[CTTZ9:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD9]], i1 true)
+; CHECK-NEXT: [[CTTZ10:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD10]], i1 true)
+; CHECK-NEXT: [[CTTZ11:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD11]], i1 true)
+; CHECK-NEXT: [[CTTZ12:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD12]], i1 true)
+; CHECK-NEXT: [[CTTZ13:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD13]], i1 true)
+; CHECK-NEXT: [[CTTZ14:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD14]], i1 true)
+; CHECK-NEXT: [[CTTZ15:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD15]], i1 true)
+; CHECK-NEXT: store i8 [[CTTZ0]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 0), align 1
+; CHECK-NEXT: store i8 [[CTTZ1]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 1), align 1
+; CHECK-NEXT: store i8 [[CTTZ2]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 2), align 1
+; CHECK-NEXT: store i8 [[CTTZ3]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 3), align 1
+; CHECK-NEXT: store i8 [[CTTZ4]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 4), align 1
+; CHECK-NEXT: store i8 [[CTTZ5]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 5), align 1
+; CHECK-NEXT: store i8 [[CTTZ6]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 6), align 1
+; CHECK-NEXT: store i8 [[CTTZ7]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 7), align 1
+; CHECK-NEXT: store i8 [[CTTZ8]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 8), align 1
+; CHECK-NEXT: store i8 [[CTTZ9]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 9), align 1
+; CHECK-NEXT: store i8 [[CTTZ10]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 10), align 1
+; CHECK-NEXT: store i8 [[CTTZ11]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 11), align 1
+; CHECK-NEXT: store i8 [[CTTZ12]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 12), align 1
+; CHECK-NEXT: store i8 [[CTTZ13]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 13), align 1
+; CHECK-NEXT: store i8 [[CTTZ14]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 14), align 1
+; CHECK-NEXT: store i8 [[CTTZ15]], 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
+ %cttz0 = call i8 @llvm.cttz.i8(i8 %ld0, i1 -1)
+ %cttz1 = call i8 @llvm.cttz.i8(i8 %ld1, i1 -1)
+ %cttz2 = call i8 @llvm.cttz.i8(i8 %ld2, i1 -1)
+ %cttz3 = call i8 @llvm.cttz.i8(i8 %ld3, i1 -1)
+ %cttz4 = call i8 @llvm.cttz.i8(i8 %ld4, i1 -1)
+ %cttz5 = call i8 @llvm.cttz.i8(i8 %ld5, i1 -1)
+ %cttz6 = call i8 @llvm.cttz.i8(i8 %ld6, i1 -1)
+ %cttz7 = call i8 @llvm.cttz.i8(i8 %ld7, i1 -1)
+ %cttz8 = call i8 @llvm.cttz.i8(i8 %ld8, i1 -1)
+ %cttz9 = call i8 @llvm.cttz.i8(i8 %ld9, i1 -1)
+ %cttz10 = call i8 @llvm.cttz.i8(i8 %ld10, i1 -1)
+ %cttz11 = call i8 @llvm.cttz.i8(i8 %ld11, i1 -1)
+ %cttz12 = call i8 @llvm.cttz.i8(i8 %ld12, i1 -1)
+ %cttz13 = call i8 @llvm.cttz.i8(i8 %ld13, i1 -1)
+ %cttz14 = call i8 @llvm.cttz.i8(i8 %ld14, i1 -1)
+ %cttz15 = call i8 @llvm.cttz.i8(i8 %ld15, i1 -1)
+ store i8 %cttz0 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 0), align 1
+ store i8 %cttz1 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 1), align 1
+ store i8 %cttz2 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 2), align 1
+ store i8 %cttz3 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 3), align 1
+ store i8 %cttz4 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 4), align 1
+ store i8 %cttz5 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 5), align 1
+ store i8 %cttz6 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 6), align 1
+ store i8 %cttz7 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 7), align 1
+ store i8 %cttz8 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 8), align 1
+ store i8 %cttz9 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 9), align 1
+ store i8 %cttz10, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 10), align 1
+ store i8 %cttz11, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 11), align 1
+ store i8 %cttz12, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 12), align 1
+ store i8 %cttz13, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 13), align 1
+ store i8 %cttz14, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 14), align 1
+ store i8 %cttz15, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 15), align 1
+ ret void
+}
+
+define void @cttz_undef_32i8() #0 {
+; CHECK-LABEL: @cttz_undef_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: [[CTTZ0:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD0]], i1 true)
+; CHECK-NEXT: [[CTTZ1:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD1]], i1 true)
+; CHECK-NEXT: [[CTTZ2:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD2]], i1 true)
+; CHECK-NEXT: [[CTTZ3:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD3]], i1 true)
+; CHECK-NEXT: [[CTTZ4:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD4]], i1 true)
+; CHECK-NEXT: [[CTTZ5:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD5]], i1 true)
+; CHECK-NEXT: [[CTTZ6:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD6]], i1 true)
+; CHECK-NEXT: [[CTTZ7:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD7]], i1 true)
+; CHECK-NEXT: [[CTTZ8:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD8]], i1 true)
+; CHECK-NEXT: [[CTTZ9:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD9]], i1 true)
+; CHECK-NEXT: [[CTTZ10:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD10]], i1 true)
+; CHECK-NEXT: [[CTTZ11:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD11]], i1 true)
+; CHECK-NEXT: [[CTTZ12:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD12]], i1 true)
+; CHECK-NEXT: [[CTTZ13:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD13]], i1 true)
+; CHECK-NEXT: [[CTTZ14:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD14]], i1 true)
+; CHECK-NEXT: [[CTTZ15:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD15]], i1 true)
+; CHECK-NEXT: [[CTTZ16:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD16]], i1 true)
+; CHECK-NEXT: [[CTTZ17:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD17]], i1 true)
+; CHECK-NEXT: [[CTTZ18:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD18]], i1 true)
+; CHECK-NEXT: [[CTTZ19:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD19]], i1 true)
+; CHECK-NEXT: [[CTTZ20:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD20]], i1 true)
+; CHECK-NEXT: [[CTTZ21:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD21]], i1 true)
+; CHECK-NEXT: [[CTTZ22:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD22]], i1 true)
+; CHECK-NEXT: [[CTTZ23:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD23]], i1 true)
+; CHECK-NEXT: [[CTTZ24:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD24]], i1 true)
+; CHECK-NEXT: [[CTTZ25:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD25]], i1 true)
+; CHECK-NEXT: [[CTTZ26:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD26]], i1 true)
+; CHECK-NEXT: [[CTTZ27:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD27]], i1 true)
+; CHECK-NEXT: [[CTTZ28:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD28]], i1 true)
+; CHECK-NEXT: [[CTTZ29:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD29]], i1 true)
+; CHECK-NEXT: [[CTTZ30:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD30]], i1 true)
+; CHECK-NEXT: [[CTTZ31:%.*]] = call i8 @llvm.cttz.i8(i8 [[LD31]], i1 true)
+; CHECK-NEXT: store i8 [[CTTZ0]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 0), align 1
+; CHECK-NEXT: store i8 [[CTTZ1]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 1), align 1
+; CHECK-NEXT: store i8 [[CTTZ2]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 2), align 1
+; CHECK-NEXT: store i8 [[CTTZ3]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 3), align 1
+; CHECK-NEXT: store i8 [[CTTZ4]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 4), align 1
+; CHECK-NEXT: store i8 [[CTTZ5]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 5), align 1
+; CHECK-NEXT: store i8 [[CTTZ6]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 6), align 1
+; CHECK-NEXT: store i8 [[CTTZ7]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 7), align 1
+; CHECK-NEXT: store i8 [[CTTZ8]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 8), align 1
+; CHECK-NEXT: store i8 [[CTTZ9]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 9), align 1
+; CHECK-NEXT: store i8 [[CTTZ10]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 10), align 1
+; CHECK-NEXT: store i8 [[CTTZ11]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 11), align 1
+; CHECK-NEXT: store i8 [[CTTZ12]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 12), align 1
+; CHECK-NEXT: store i8 [[CTTZ13]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 13), align 1
+; CHECK-NEXT: store i8 [[CTTZ14]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 14), align 1
+; CHECK-NEXT: store i8 [[CTTZ15]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 15), align 1
+; CHECK-NEXT: store i8 [[CTTZ16]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 16), align 1
+; CHECK-NEXT: store i8 [[CTTZ17]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 17), align 1
+; CHECK-NEXT: store i8 [[CTTZ18]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 18), align 1
+; CHECK-NEXT: store i8 [[CTTZ19]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 19), align 1
+; CHECK-NEXT: store i8 [[CTTZ20]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 20), align 1
+; CHECK-NEXT: store i8 [[CTTZ21]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 21), align 1
+; CHECK-NEXT: store i8 [[CTTZ22]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 22), align 1
+; CHECK-NEXT: store i8 [[CTTZ23]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 23), align 1
+; CHECK-NEXT: store i8 [[CTTZ24]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 24), align 1
+; CHECK-NEXT: store i8 [[CTTZ25]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 25), align 1
+; CHECK-NEXT: store i8 [[CTTZ26]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 26), align 1
+; CHECK-NEXT: store i8 [[CTTZ27]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 27), align 1
+; CHECK-NEXT: store i8 [[CTTZ28]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 28), align 1
+; CHECK-NEXT: store i8 [[CTTZ29]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 29), align 1
+; CHECK-NEXT: store i8 [[CTTZ30]], i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 30), align 1
+; CHECK-NEXT: store i8 [[CTTZ31]], 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
+ %cttz0 = call i8 @llvm.cttz.i8(i8 %ld0, i1 -1)
+ %cttz1 = call i8 @llvm.cttz.i8(i8 %ld1, i1 -1)
+ %cttz2 = call i8 @llvm.cttz.i8(i8 %ld2, i1 -1)
+ %cttz3 = call i8 @llvm.cttz.i8(i8 %ld3, i1 -1)
+ %cttz4 = call i8 @llvm.cttz.i8(i8 %ld4, i1 -1)
+ %cttz5 = call i8 @llvm.cttz.i8(i8 %ld5, i1 -1)
+ %cttz6 = call i8 @llvm.cttz.i8(i8 %ld6, i1 -1)
+ %cttz7 = call i8 @llvm.cttz.i8(i8 %ld7, i1 -1)
+ %cttz8 = call i8 @llvm.cttz.i8(i8 %ld8, i1 -1)
+ %cttz9 = call i8 @llvm.cttz.i8(i8 %ld9, i1 -1)
+ %cttz10 = call i8 @llvm.cttz.i8(i8 %ld10, i1 -1)
+ %cttz11 = call i8 @llvm.cttz.i8(i8 %ld11, i1 -1)
+ %cttz12 = call i8 @llvm.cttz.i8(i8 %ld12, i1 -1)
+ %cttz13 = call i8 @llvm.cttz.i8(i8 %ld13, i1 -1)
+ %cttz14 = call i8 @llvm.cttz.i8(i8 %ld14, i1 -1)
+ %cttz15 = call i8 @llvm.cttz.i8(i8 %ld15, i1 -1)
+ %cttz16 = call i8 @llvm.cttz.i8(i8 %ld16, i1 -1)
+ %cttz17 = call i8 @llvm.cttz.i8(i8 %ld17, i1 -1)
+ %cttz18 = call i8 @llvm.cttz.i8(i8 %ld18, i1 -1)
+ %cttz19 = call i8 @llvm.cttz.i8(i8 %ld19, i1 -1)
+ %cttz20 = call i8 @llvm.cttz.i8(i8 %ld20, i1 -1)
+ %cttz21 = call i8 @llvm.cttz.i8(i8 %ld21, i1 -1)
+ %cttz22 = call i8 @llvm.cttz.i8(i8 %ld22, i1 -1)
+ %cttz23 = call i8 @llvm.cttz.i8(i8 %ld23, i1 -1)
+ %cttz24 = call i8 @llvm.cttz.i8(i8 %ld24, i1 -1)
+ %cttz25 = call i8 @llvm.cttz.i8(i8 %ld25, i1 -1)
+ %cttz26 = call i8 @llvm.cttz.i8(i8 %ld26, i1 -1)
+ %cttz27 = call i8 @llvm.cttz.i8(i8 %ld27, i1 -1)
+ %cttz28 = call i8 @llvm.cttz.i8(i8 %ld28, i1 -1)
+ %cttz29 = call i8 @llvm.cttz.i8(i8 %ld29, i1 -1)
+ %cttz30 = call i8 @llvm.cttz.i8(i8 %ld30, i1 -1)
+ %cttz31 = call i8 @llvm.cttz.i8(i8 %ld31, i1 -1)
+ store i8 %cttz0 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 0), align 1
+ store i8 %cttz1 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 1), align 1
+ store i8 %cttz2 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 2), align 1
+ store i8 %cttz3 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 3), align 1
+ store i8 %cttz4 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 4), align 1
+ store i8 %cttz5 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 5), align 1
+ store i8 %cttz6 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 6), align 1
+ store i8 %cttz7 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 7), align 1
+ store i8 %cttz8 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 8), align 1
+ store i8 %cttz9 , i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 9), align 1
+ store i8 %cttz10, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 10), align 1
+ store i8 %cttz11, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 11), align 1
+ store i8 %cttz12, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 12), align 1
+ store i8 %cttz13, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 13), align 1
+ store i8 %cttz14, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 14), align 1
+ store i8 %cttz15, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 15), align 1
+ store i8 %cttz16, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 16), align 1
+ store i8 %cttz17, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 17), align 1
+ store i8 %cttz18, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 18), align 1
+ store i8 %cttz19, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 19), align 1
+ store i8 %cttz20, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 20), align 1
+ store i8 %cttz21, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 21), align 1
+ store i8 %cttz22, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 22), align 1
+ store i8 %cttz23, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 23), align 1
+ store i8 %cttz24, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 24), align 1
+ store i8 %cttz25, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 25), align 1
+ store i8 %cttz26, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 26), align 1
+ store i8 %cttz27, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 27), align 1
+ store i8 %cttz28, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 28), align 1
+ store i8 %cttz29, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 29), align 1
+ store i8 %cttz30, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 30), align 1
+ store i8 %cttz31, i8* getelementptr inbounds ([32 x i8], [32 x i8]* @dst8, i8 0, i64 31), align 1
+ ret void
+}
+
+attributes #0 = { nounwind }
More information about the llvm-commits
mailing list