From: Simon Pilgrim Date: Sat, 1 Oct 2016 17:00:26 +0000 (+0000) Subject: [SLPVectorizer][X86] Added fcopysign tests X-Git-Url: https://granicus.if.org/sourcecode?a=commitdiff_plain;h=a94243b44f81b57cd1b5d440ea8771aeedebf6e1;p=llvm [SLPVectorizer][X86] Added fcopysign tests git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@283046 91177308-0d34-0410-b5e6-96231b3b80d8 --- diff --git a/test/Transforms/SLPVectorizer/X86/fcopysign.ll b/test/Transforms/SLPVectorizer/X86/fcopysign.ll new file mode 100644 index 00000000000..9233ada291b --- /dev/null +++ b/test/Transforms/SLPVectorizer/X86/fcopysign.ll @@ -0,0 +1,404 @@ +; 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 --check-prefix=AVX256 +; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=bdver1 -basicaa -slp-vectorizer -S | FileCheck %s --check-prefix=CHECK --check-prefix=AVX --check-prefix=AVX256 +; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=core-avx2 -basicaa -slp-vectorizer -S | FileCheck %s --check-prefix=CHECK --check-prefix=AVX --check-prefix=AVX256 +; RUN: opt < %s -mtriple=x86_64-unknown -mcpu=skylake-avx512 -basicaa -slp-vectorizer -S | FileCheck %s --check-prefix=CHECK --check-prefix=AVX --check-prefix=AVX512 + +target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128" + +@srcA64 = common global [8 x double] zeroinitializer, align 64 +@srcB64 = common global [8 x double] zeroinitializer, align 64 +@srcC64 = common global [8 x double] zeroinitializer, align 64 +@srcA32 = common global [16 x float] zeroinitializer, align 64 +@srcB32 = common global [16 x float] zeroinitializer, align 64 +@srcC32 = common global [16 x float] zeroinitializer, align 64 +@dst64 = common global [8 x double] zeroinitializer, align 64 +@dst32 = common global [16 x float] zeroinitializer, align 64 + +declare float @llvm.copysign.f32(float, float) +declare double @llvm.copysign.f64(double, double) + +; +; CHECK +; + +define void @fcopysign_2f64() #0 { +; CHECK-LABEL: @fcopysign_2f64( +; CHECK-NEXT: [[A0:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcA64, i32 0, i64 0), align 8 +; CHECK-NEXT: [[A1:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcA64, i32 0, i64 1), align 8 +; CHECK-NEXT: [[B0:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcB64, i32 0, i64 0), align 8 +; CHECK-NEXT: [[B1:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcB64, i32 0, i64 1), align 8 +; CHECK-NEXT: [[FCOPYSIGN0:%.*]] = call double @llvm.copysign.f64(double [[A0]], double [[B0]]) +; CHECK-NEXT: [[FCOPYSIGN1:%.*]] = call double @llvm.copysign.f64(double [[A1]], double [[B1]]) +; CHECK-NEXT: store double [[FCOPYSIGN0]], double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 0), align 8 +; CHECK-NEXT: store double [[FCOPYSIGN1]], double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 1), align 8 +; CHECK-NEXT: ret void +; + %a0 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcA64, i32 0, i64 0), align 8 + %a1 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcA64, i32 0, i64 1), align 8 + %b0 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcB64, i32 0, i64 0), align 8 + %b1 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcB64, i32 0, i64 1), align 8 + %fcopysign0 = call double @llvm.copysign.f64(double %a0, double %b0) + %fcopysign1 = call double @llvm.copysign.f64(double %a1, double %b1) + store double %fcopysign0, double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 0), align 8 + store double %fcopysign1, double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 1), align 8 + ret void +} + +define void @fcopysign_4f64() #0 { +; CHECK-LABEL: @fcopysign_4f64( +; CHECK-NEXT: [[A0:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcA64, i32 0, i64 0), align 8 +; CHECK-NEXT: [[A1:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcA64, i32 0, i64 1), align 8 +; CHECK-NEXT: [[A2:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcA64, i32 0, i64 2), align 8 +; CHECK-NEXT: [[A3:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcA64, i32 0, i64 3), align 8 +; CHECK-NEXT: [[B0:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcB64, i32 0, i64 0), align 8 +; CHECK-NEXT: [[B1:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcB64, i32 0, i64 1), align 8 +; CHECK-NEXT: [[B2:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcB64, i32 0, i64 2), align 8 +; CHECK-NEXT: [[B3:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcB64, i32 0, i64 3), align 8 +; CHECK-NEXT: [[FCOPYSIGN0:%.*]] = call double @llvm.copysign.f64(double [[A0]], double [[B0]]) +; CHECK-NEXT: [[FCOPYSIGN1:%.*]] = call double @llvm.copysign.f64(double [[A1]], double [[B1]]) +; CHECK-NEXT: [[FCOPYSIGN2:%.*]] = call double @llvm.copysign.f64(double [[A2]], double [[B2]]) +; CHECK-NEXT: [[FCOPYSIGN3:%.*]] = call double @llvm.copysign.f64(double [[A3]], double [[B3]]) +; CHECK-NEXT: store double [[FCOPYSIGN0]], double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 0), align 8 +; CHECK-NEXT: store double [[FCOPYSIGN1]], double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 1), align 8 +; CHECK-NEXT: store double [[FCOPYSIGN2]], double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 2), align 8 +; CHECK-NEXT: store double [[FCOPYSIGN3]], double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 3), align 8 +; CHECK-NEXT: ret void +; + %a0 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcA64, i32 0, i64 0), align 8 + %a1 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcA64, i32 0, i64 1), align 8 + %a2 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcA64, i32 0, i64 2), align 8 + %a3 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcA64, i32 0, i64 3), align 8 + %b0 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcB64, i32 0, i64 0), align 8 + %b1 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcB64, i32 0, i64 1), align 8 + %b2 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcB64, i32 0, i64 2), align 8 + %b3 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcB64, i32 0, i64 3), align 8 + %fcopysign0 = call double @llvm.copysign.f64(double %a0, double %b0) + %fcopysign1 = call double @llvm.copysign.f64(double %a1, double %b1) + %fcopysign2 = call double @llvm.copysign.f64(double %a2, double %b2) + %fcopysign3 = call double @llvm.copysign.f64(double %a3, double %b3) + store double %fcopysign0, double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 0), align 8 + store double %fcopysign1, double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 1), align 8 + store double %fcopysign2, double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 2), align 8 + store double %fcopysign3, double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 3), align 8 + ret void +} + +define void @fcopysign_8f64() #0 { +; CHECK-LABEL: @fcopysign_8f64( +; CHECK-NEXT: [[A0:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcA64, i32 0, i64 0), align 4 +; CHECK-NEXT: [[A1:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcA64, i32 0, i64 1), align 4 +; CHECK-NEXT: [[A2:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcA64, i32 0, i64 2), align 4 +; CHECK-NEXT: [[A3:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcA64, i32 0, i64 3), align 4 +; CHECK-NEXT: [[A4:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcA64, i32 0, i64 4), align 4 +; CHECK-NEXT: [[A5:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcA64, i32 0, i64 5), align 4 +; CHECK-NEXT: [[A6:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcA64, i32 0, i64 6), align 4 +; CHECK-NEXT: [[A7:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcA64, i32 0, i64 7), align 4 +; CHECK-NEXT: [[B0:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcB64, i32 0, i64 0), align 4 +; CHECK-NEXT: [[B1:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcB64, i32 0, i64 1), align 4 +; CHECK-NEXT: [[B2:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcB64, i32 0, i64 2), align 4 +; CHECK-NEXT: [[B3:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcB64, i32 0, i64 3), align 4 +; CHECK-NEXT: [[B4:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcB64, i32 0, i64 4), align 4 +; CHECK-NEXT: [[B5:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcB64, i32 0, i64 5), align 4 +; CHECK-NEXT: [[B6:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcB64, i32 0, i64 6), align 4 +; CHECK-NEXT: [[B7:%.*]] = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcB64, i32 0, i64 7), align 4 +; CHECK-NEXT: [[FCOPYSIGN0:%.*]] = call double @llvm.copysign.f64(double [[A0]], double [[B0]]) +; CHECK-NEXT: [[FCOPYSIGN1:%.*]] = call double @llvm.copysign.f64(double [[A1]], double [[B1]]) +; CHECK-NEXT: [[FCOPYSIGN2:%.*]] = call double @llvm.copysign.f64(double [[A2]], double [[B2]]) +; CHECK-NEXT: [[FCOPYSIGN3:%.*]] = call double @llvm.copysign.f64(double [[A3]], double [[B3]]) +; CHECK-NEXT: [[FCOPYSIGN4:%.*]] = call double @llvm.copysign.f64(double [[A4]], double [[B4]]) +; CHECK-NEXT: [[FCOPYSIGN5:%.*]] = call double @llvm.copysign.f64(double [[A5]], double [[B5]]) +; CHECK-NEXT: [[FCOPYSIGN6:%.*]] = call double @llvm.copysign.f64(double [[A6]], double [[B6]]) +; CHECK-NEXT: [[FCOPYSIGN7:%.*]] = call double @llvm.copysign.f64(double [[A7]], double [[B7]]) +; CHECK-NEXT: store double [[FCOPYSIGN0]], double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 0), align 4 +; CHECK-NEXT: store double [[FCOPYSIGN1]], double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 1), align 4 +; CHECK-NEXT: store double [[FCOPYSIGN2]], double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 2), align 4 +; CHECK-NEXT: store double [[FCOPYSIGN3]], double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 3), align 4 +; CHECK-NEXT: store double [[FCOPYSIGN4]], double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 4), align 4 +; CHECK-NEXT: store double [[FCOPYSIGN5]], double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 5), align 4 +; CHECK-NEXT: store double [[FCOPYSIGN6]], double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 6), align 4 +; CHECK-NEXT: store double [[FCOPYSIGN7]], double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 7), align 4 +; CHECK-NEXT: ret void +; + %a0 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcA64, i32 0, i64 0), align 4 + %a1 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcA64, i32 0, i64 1), align 4 + %a2 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcA64, i32 0, i64 2), align 4 + %a3 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcA64, i32 0, i64 3), align 4 + %a4 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcA64, i32 0, i64 4), align 4 + %a5 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcA64, i32 0, i64 5), align 4 + %a6 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcA64, i32 0, i64 6), align 4 + %a7 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcA64, i32 0, i64 7), align 4 + %b0 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcB64, i32 0, i64 0), align 4 + %b1 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcB64, i32 0, i64 1), align 4 + %b2 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcB64, i32 0, i64 2), align 4 + %b3 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcB64, i32 0, i64 3), align 4 + %b4 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcB64, i32 0, i64 4), align 4 + %b5 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcB64, i32 0, i64 5), align 4 + %b6 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcB64, i32 0, i64 6), align 4 + %b7 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @srcB64, i32 0, i64 7), align 4 + %fcopysign0 = call double @llvm.copysign.f64(double %a0, double %b0) + %fcopysign1 = call double @llvm.copysign.f64(double %a1, double %b1) + %fcopysign2 = call double @llvm.copysign.f64(double %a2, double %b2) + %fcopysign3 = call double @llvm.copysign.f64(double %a3, double %b3) + %fcopysign4 = call double @llvm.copysign.f64(double %a4, double %b4) + %fcopysign5 = call double @llvm.copysign.f64(double %a5, double %b5) + %fcopysign6 = call double @llvm.copysign.f64(double %a6, double %b6) + %fcopysign7 = call double @llvm.copysign.f64(double %a7, double %b7) + store double %fcopysign0, double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 0), align 4 + store double %fcopysign1, double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 1), align 4 + store double %fcopysign2, double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 2), align 4 + store double %fcopysign3, double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 3), align 4 + store double %fcopysign4, double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 4), align 4 + store double %fcopysign5, double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 5), align 4 + store double %fcopysign6, double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 6), align 4 + store double %fcopysign7, double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 7), align 4 + ret void +} + +define void @fcopysign_4f32() #0 { +; CHECK-LABEL: @fcopysign_4f32( +; CHECK-NEXT: [[A0:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 0), align 4 +; CHECK-NEXT: [[A1:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 1), align 4 +; CHECK-NEXT: [[A2:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 2), align 4 +; CHECK-NEXT: [[A3:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 3), align 4 +; CHECK-NEXT: [[B0:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 0), align 4 +; CHECK-NEXT: [[B1:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 1), align 4 +; CHECK-NEXT: [[B2:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 2), align 4 +; CHECK-NEXT: [[B3:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 3), align 4 +; CHECK-NEXT: [[FCOPYSIGN0:%.*]] = call float @llvm.copysign.f32(float [[A0]], float [[B0]]) +; CHECK-NEXT: [[FCOPYSIGN1:%.*]] = call float @llvm.copysign.f32(float [[A1]], float [[B1]]) +; CHECK-NEXT: [[FCOPYSIGN2:%.*]] = call float @llvm.copysign.f32(float [[A2]], float [[B2]]) +; CHECK-NEXT: [[FCOPYSIGN3:%.*]] = call float @llvm.copysign.f32(float [[A3]], float [[B3]]) +; CHECK-NEXT: store float [[FCOPYSIGN0]], float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 0), align 4 +; CHECK-NEXT: store float [[FCOPYSIGN1]], float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 1), align 4 +; CHECK-NEXT: store float [[FCOPYSIGN2]], float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 2), align 4 +; CHECK-NEXT: store float [[FCOPYSIGN3]], float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 3), align 4 +; CHECK-NEXT: ret void +; + %a0 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 0), align 4 + %a1 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 1), align 4 + %a2 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 2), align 4 + %a3 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 3), align 4 + %b0 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 0), align 4 + %b1 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 1), align 4 + %b2 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 2), align 4 + %b3 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 3), align 4 + %fcopysign0 = call float @llvm.copysign.f32(float %a0, float %b0) + %fcopysign1 = call float @llvm.copysign.f32(float %a1, float %b1) + %fcopysign2 = call float @llvm.copysign.f32(float %a2, float %b2) + %fcopysign3 = call float @llvm.copysign.f32(float %a3, float %b3) + store float %fcopysign0, float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 0), align 4 + store float %fcopysign1, float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 1), align 4 + store float %fcopysign2, float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 2), align 4 + store float %fcopysign3, float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 3), align 4 + ret void +} + +define void @fcopysign_8f32() #0 { +; CHECK-LABEL: @fcopysign_8f32( +; CHECK-NEXT: [[A0:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 0), align 4 +; CHECK-NEXT: [[A1:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 1), align 4 +; CHECK-NEXT: [[A2:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 2), align 4 +; CHECK-NEXT: [[A3:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 3), align 4 +; CHECK-NEXT: [[A4:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 4), align 4 +; CHECK-NEXT: [[A5:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 5), align 4 +; CHECK-NEXT: [[A6:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 6), align 4 +; CHECK-NEXT: [[A7:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 7), align 4 +; CHECK-NEXT: [[B0:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 0), align 4 +; CHECK-NEXT: [[B1:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 1), align 4 +; CHECK-NEXT: [[B2:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 2), align 4 +; CHECK-NEXT: [[B3:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 3), align 4 +; CHECK-NEXT: [[B4:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 4), align 4 +; CHECK-NEXT: [[B5:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 5), align 4 +; CHECK-NEXT: [[B6:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 6), align 4 +; CHECK-NEXT: [[B7:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 7), align 4 +; CHECK-NEXT: [[FCOPYSIGN0:%.*]] = call float @llvm.copysign.f32(float [[A0]], float [[B0]]) +; CHECK-NEXT: [[FCOPYSIGN1:%.*]] = call float @llvm.copysign.f32(float [[A1]], float [[B1]]) +; CHECK-NEXT: [[FCOPYSIGN2:%.*]] = call float @llvm.copysign.f32(float [[A2]], float [[B2]]) +; CHECK-NEXT: [[FCOPYSIGN3:%.*]] = call float @llvm.copysign.f32(float [[A3]], float [[B3]]) +; CHECK-NEXT: [[FCOPYSIGN4:%.*]] = call float @llvm.copysign.f32(float [[A4]], float [[B4]]) +; CHECK-NEXT: [[FCOPYSIGN5:%.*]] = call float @llvm.copysign.f32(float [[A5]], float [[B5]]) +; CHECK-NEXT: [[FCOPYSIGN6:%.*]] = call float @llvm.copysign.f32(float [[A6]], float [[B6]]) +; CHECK-NEXT: [[FCOPYSIGN7:%.*]] = call float @llvm.copysign.f32(float [[A7]], float [[B7]]) +; CHECK-NEXT: store float [[FCOPYSIGN0]], float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 0), align 4 +; CHECK-NEXT: store float [[FCOPYSIGN1]], float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 1), align 4 +; CHECK-NEXT: store float [[FCOPYSIGN2]], float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 2), align 4 +; CHECK-NEXT: store float [[FCOPYSIGN3]], float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 3), align 4 +; CHECK-NEXT: store float [[FCOPYSIGN4]], float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 4), align 4 +; CHECK-NEXT: store float [[FCOPYSIGN5]], float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 5), align 4 +; CHECK-NEXT: store float [[FCOPYSIGN6]], float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 6), align 4 +; CHECK-NEXT: store float [[FCOPYSIGN7]], float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 7), align 4 +; CHECK-NEXT: ret void +; + %a0 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 0), align 4 + %a1 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 1), align 4 + %a2 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 2), align 4 + %a3 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 3), align 4 + %a4 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 4), align 4 + %a5 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 5), align 4 + %a6 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 6), align 4 + %a7 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 7), align 4 + %b0 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 0), align 4 + %b1 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 1), align 4 + %b2 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 2), align 4 + %b3 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 3), align 4 + %b4 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 4), align 4 + %b5 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 5), align 4 + %b6 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 6), align 4 + %b7 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 7), align 4 + %fcopysign0 = call float @llvm.copysign.f32(float %a0, float %b0) + %fcopysign1 = call float @llvm.copysign.f32(float %a1, float %b1) + %fcopysign2 = call float @llvm.copysign.f32(float %a2, float %b2) + %fcopysign3 = call float @llvm.copysign.f32(float %a3, float %b3) + %fcopysign4 = call float @llvm.copysign.f32(float %a4, float %b4) + %fcopysign5 = call float @llvm.copysign.f32(float %a5, float %b5) + %fcopysign6 = call float @llvm.copysign.f32(float %a6, float %b6) + %fcopysign7 = call float @llvm.copysign.f32(float %a7, float %b7) + store float %fcopysign0, float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 0), align 4 + store float %fcopysign1, float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 1), align 4 + store float %fcopysign2, float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 2), align 4 + store float %fcopysign3, float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 3), align 4 + store float %fcopysign4, float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 4), align 4 + store float %fcopysign5, float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 5), align 4 + store float %fcopysign6, float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 6), align 4 + store float %fcopysign7, float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 7), align 4 + ret void +} + +define void @fcopysign_16f32() #0 { +; CHECK-LABEL: @fcopysign_16f32( +; CHECK-NEXT: [[A0:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 0), align 4 +; CHECK-NEXT: [[A1:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 1), align 4 +; CHECK-NEXT: [[A2:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 2), align 4 +; CHECK-NEXT: [[A3:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 3), align 4 +; CHECK-NEXT: [[A4:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 4), align 4 +; CHECK-NEXT: [[A5:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 5), align 4 +; CHECK-NEXT: [[A6:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 6), align 4 +; CHECK-NEXT: [[A7:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 7), align 4 +; CHECK-NEXT: [[A8:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 8), align 4 +; CHECK-NEXT: [[A9:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 9), align 4 +; CHECK-NEXT: [[A10:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 10), align 4 +; CHECK-NEXT: [[A11:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 11), align 4 +; CHECK-NEXT: [[A12:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 12), align 4 +; CHECK-NEXT: [[A13:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 13), align 4 +; CHECK-NEXT: [[A14:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 14), align 4 +; CHECK-NEXT: [[A15:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 15), align 4 +; CHECK-NEXT: [[B0:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 0), align 4 +; CHECK-NEXT: [[B1:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 1), align 4 +; CHECK-NEXT: [[B2:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 2), align 4 +; CHECK-NEXT: [[B3:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 3), align 4 +; CHECK-NEXT: [[B4:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 4), align 4 +; CHECK-NEXT: [[B5:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 5), align 4 +; CHECK-NEXT: [[B6:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 6), align 4 +; CHECK-NEXT: [[B7:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 7), align 4 +; CHECK-NEXT: [[B8:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 8), align 4 +; CHECK-NEXT: [[B9:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 9), align 4 +; CHECK-NEXT: [[B10:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 10), align 4 +; CHECK-NEXT: [[B11:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 11), align 4 +; CHECK-NEXT: [[B12:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 12), align 4 +; CHECK-NEXT: [[B13:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 13), align 4 +; CHECK-NEXT: [[B14:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 14), align 4 +; CHECK-NEXT: [[B15:%.*]] = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 15), align 4 +; CHECK-NEXT: [[FCOPYSIGN0:%.*]] = call float @llvm.copysign.f32(float [[A0]], float [[B0]]) +; CHECK-NEXT: [[FCOPYSIGN1:%.*]] = call float @llvm.copysign.f32(float [[A1]], float [[B1]]) +; CHECK-NEXT: [[FCOPYSIGN2:%.*]] = call float @llvm.copysign.f32(float [[A2]], float [[B2]]) +; CHECK-NEXT: [[FCOPYSIGN3:%.*]] = call float @llvm.copysign.f32(float [[A3]], float [[B3]]) +; CHECK-NEXT: [[FCOPYSIGN4:%.*]] = call float @llvm.copysign.f32(float [[A4]], float [[B4]]) +; CHECK-NEXT: [[FCOPYSIGN5:%.*]] = call float @llvm.copysign.f32(float [[A5]], float [[B5]]) +; CHECK-NEXT: [[FCOPYSIGN6:%.*]] = call float @llvm.copysign.f32(float [[A6]], float [[B6]]) +; CHECK-NEXT: [[FCOPYSIGN7:%.*]] = call float @llvm.copysign.f32(float [[A7]], float [[B7]]) +; CHECK-NEXT: [[FCOPYSIGN8:%.*]] = call float @llvm.copysign.f32(float [[A8]], float [[B8]]) +; CHECK-NEXT: [[FCOPYSIGN9:%.*]] = call float @llvm.copysign.f32(float [[A9]], float [[B9]]) +; CHECK-NEXT: [[FCOPYSIGN10:%.*]] = call float @llvm.copysign.f32(float [[A10]], float [[B10]]) +; CHECK-NEXT: [[FCOPYSIGN11:%.*]] = call float @llvm.copysign.f32(float [[A11]], float [[B11]]) +; CHECK-NEXT: [[FCOPYSIGN12:%.*]] = call float @llvm.copysign.f32(float [[A12]], float [[B12]]) +; CHECK-NEXT: [[FCOPYSIGN13:%.*]] = call float @llvm.copysign.f32(float [[A13]], float [[B13]]) +; CHECK-NEXT: [[FCOPYSIGN14:%.*]] = call float @llvm.copysign.f32(float [[A14]], float [[B14]]) +; CHECK-NEXT: [[FCOPYSIGN15:%.*]] = call float @llvm.copysign.f32(float [[A15]], float [[B15]]) +; CHECK-NEXT: store float [[FCOPYSIGN0]], float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 0), align 4 +; CHECK-NEXT: store float [[FCOPYSIGN1]], float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 1), align 4 +; CHECK-NEXT: store float [[FCOPYSIGN2]], float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 2), align 4 +; CHECK-NEXT: store float [[FCOPYSIGN3]], float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 3), align 4 +; CHECK-NEXT: store float [[FCOPYSIGN4]], float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 4), align 4 +; CHECK-NEXT: store float [[FCOPYSIGN5]], float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 5), align 4 +; CHECK-NEXT: store float [[FCOPYSIGN6]], float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 6), align 4 +; CHECK-NEXT: store float [[FCOPYSIGN7]], float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 7), align 4 +; CHECK-NEXT: store float [[FCOPYSIGN8]], float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 8), align 4 +; CHECK-NEXT: store float [[FCOPYSIGN9]], float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 9), align 4 +; CHECK-NEXT: store float [[FCOPYSIGN10]], float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 10), align 4 +; CHECK-NEXT: store float [[FCOPYSIGN11]], float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 11), align 4 +; CHECK-NEXT: store float [[FCOPYSIGN12]], float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 12), align 4 +; CHECK-NEXT: store float [[FCOPYSIGN13]], float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 13), align 4 +; CHECK-NEXT: store float [[FCOPYSIGN14]], float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 14), align 4 +; CHECK-NEXT: store float [[FCOPYSIGN15]], float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 15), align 4 +; CHECK-NEXT: ret void +; + %a0 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 0), align 4 + %a1 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 1), align 4 + %a2 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 2), align 4 + %a3 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 3), align 4 + %a4 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 4), align 4 + %a5 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 5), align 4 + %a6 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 6), align 4 + %a7 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 7), align 4 + %a8 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 8), align 4 + %a9 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 9), align 4 + %a10 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 10), align 4 + %a11 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 11), align 4 + %a12 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 12), align 4 + %a13 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 13), align 4 + %a14 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 14), align 4 + %a15 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcA32, i32 0, i64 15), align 4 + %b0 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 0), align 4 + %b1 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 1), align 4 + %b2 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 2), align 4 + %b3 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 3), align 4 + %b4 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 4), align 4 + %b5 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 5), align 4 + %b6 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 6), align 4 + %b7 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 7), align 4 + %b8 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 8), align 4 + %b9 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 9), align 4 + %b10 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 10), align 4 + %b11 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 11), align 4 + %b12 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 12), align 4 + %b13 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 13), align 4 + %b14 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 14), align 4 + %b15 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @srcB32, i32 0, i64 15), align 4 + %fcopysign0 = call float @llvm.copysign.f32(float %a0 , float %b0 ) + %fcopysign1 = call float @llvm.copysign.f32(float %a1 , float %b1 ) + %fcopysign2 = call float @llvm.copysign.f32(float %a2 , float %b2 ) + %fcopysign3 = call float @llvm.copysign.f32(float %a3 , float %b3 ) + %fcopysign4 = call float @llvm.copysign.f32(float %a4 , float %b4 ) + %fcopysign5 = call float @llvm.copysign.f32(float %a5 , float %b5 ) + %fcopysign6 = call float @llvm.copysign.f32(float %a6 , float %b6 ) + %fcopysign7 = call float @llvm.copysign.f32(float %a7 , float %b7 ) + %fcopysign8 = call float @llvm.copysign.f32(float %a8 , float %b8 ) + %fcopysign9 = call float @llvm.copysign.f32(float %a9 , float %b9 ) + %fcopysign10 = call float @llvm.copysign.f32(float %a10, float %b10) + %fcopysign11 = call float @llvm.copysign.f32(float %a11, float %b11) + %fcopysign12 = call float @llvm.copysign.f32(float %a12, float %b12) + %fcopysign13 = call float @llvm.copysign.f32(float %a13, float %b13) + %fcopysign14 = call float @llvm.copysign.f32(float %a14, float %b14) + %fcopysign15 = call float @llvm.copysign.f32(float %a15, float %b15) + store float %fcopysign0 , float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 0), align 4 + store float %fcopysign1 , float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 1), align 4 + store float %fcopysign2 , float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 2), align 4 + store float %fcopysign3 , float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 3), align 4 + store float %fcopysign4 , float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 4), align 4 + store float %fcopysign5 , float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 5), align 4 + store float %fcopysign6 , float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 6), align 4 + store float %fcopysign7 , float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 7), align 4 + store float %fcopysign8 , float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 8), align 4 + store float %fcopysign9 , float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 9), align 4 + store float %fcopysign10, float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 10), align 4 + store float %fcopysign11, float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 11), align 4 + store float %fcopysign12, float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 12), align 4 + store float %fcopysign13, float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 13), align 4 + store float %fcopysign14, float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 14), align 4 + store float %fcopysign15, float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 15), align 4 + ret void +} + +attributes #0 = { nounwind }