[llvm] r283045 - [SLPVectorizer][X86] Added fabs tests
Simon Pilgrim via llvm-commits
llvm-commits at lists.llvm.org
Sat Oct 1 09:54:01 PDT 2016
Author: rksimon
Date: Sat Oct 1 11:54:01 2016
New Revision: 283045
URL: http://llvm.org/viewvc/llvm-project?rev=283045&view=rev
Log:
[SLPVectorizer][X86] Added fabs tests
Added:
llvm/trunk/test/Transforms/SLPVectorizer/X86/fabs.ll
Added: llvm/trunk/test/Transforms/SLPVectorizer/X86/fabs.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Transforms/SLPVectorizer/X86/fabs.ll?rev=283045&view=auto
==============================================================================
--- llvm/trunk/test/Transforms/SLPVectorizer/X86/fabs.ll (added)
+++ llvm/trunk/test/Transforms/SLPVectorizer/X86/fabs.ll Sat Oct 1 11:54:01 2016
@@ -0,0 +1,274 @@
+; 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"
+
+ at src64 = common global [8 x double] zeroinitializer, align 64
+ at src32 = common global [16 x float] zeroinitializer, align 64
+ at dst64 = common global [8 x double] zeroinitializer, align 64
+ at dst32 = common global [16 x float] zeroinitializer, align 64
+
+declare float @llvm.fabs.f32(float)
+declare double @llvm.fabs.f64(double)
+
+;
+; FABS
+;
+
+define void @fabs_2f64() #0 {
+; CHECK-LABEL: @fabs_2f64(
+; CHECK-NEXT: [[TMP1:%.*]] = load <2 x double>, <2 x double>* bitcast ([8 x double]* @src64 to <2 x double>*), align 8
+; CHECK-NEXT: [[TMP2:%.*]] = call <2 x double> @llvm.fabs.v2f64(<2 x double> [[TMP1]])
+; CHECK-NEXT: store <2 x double> [[TMP2]], <2 x double>* bitcast ([8 x double]* @dst64 to <2 x double>*), align 8
+; CHECK-NEXT: ret void
+;
+ %a0 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 0), align 8
+ %a1 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 1), align 8
+ %fabs0 = call double @llvm.fabs.f64(double %a0)
+ %fabs1 = call double @llvm.fabs.f64(double %a1)
+ store double %fabs0, double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 0), align 8
+ store double %fabs1, double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 1), align 8
+ ret void
+}
+
+define void @fabs_4f64() #0 {
+; SSE-LABEL: @fabs_4f64(
+; SSE-NEXT: [[TMP1:%.*]] = load <2 x double>, <2 x double>* bitcast ([8 x double]* @src64 to <2 x double>*), align 8
+; SSE-NEXT: [[TMP2:%.*]] = load <2 x double>, <2 x double>* bitcast (double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 2) to <2 x double>*), align 8
+; SSE-NEXT: [[TMP3:%.*]] = call <2 x double> @llvm.fabs.v2f64(<2 x double> [[TMP1]])
+; SSE-NEXT: [[TMP4:%.*]] = call <2 x double> @llvm.fabs.v2f64(<2 x double> [[TMP2]])
+; SSE-NEXT: store <2 x double> [[TMP3]], <2 x double>* bitcast ([8 x double]* @dst64 to <2 x double>*), align 8
+; SSE-NEXT: store <2 x double> [[TMP4]], <2 x double>* bitcast (double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 2) to <2 x double>*), align 8
+; SSE-NEXT: ret void
+;
+; AVX-LABEL: @fabs_4f64(
+; AVX-NEXT: [[TMP1:%.*]] = load <4 x double>, <4 x double>* bitcast ([8 x double]* @src64 to <4 x double>*), align 8
+; AVX-NEXT: [[TMP2:%.*]] = call <4 x double> @llvm.fabs.v4f64(<4 x double> [[TMP1]])
+; AVX-NEXT: store <4 x double> [[TMP2]], <4 x double>* bitcast ([8 x double]* @dst64 to <4 x double>*), align 8
+; AVX-NEXT: ret void
+;
+ %a0 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 0), align 8
+ %a1 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 1), align 8
+ %a2 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 2), align 8
+ %a3 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 3), align 8
+ %fabs0 = call double @llvm.fabs.f64(double %a0)
+ %fabs1 = call double @llvm.fabs.f64(double %a1)
+ %fabs2 = call double @llvm.fabs.f64(double %a2)
+ %fabs3 = call double @llvm.fabs.f64(double %a3)
+ store double %fabs0, double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 0), align 8
+ store double %fabs1, double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 1), align 8
+ store double %fabs2, double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 2), align 8
+ store double %fabs3, double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 3), align 8
+ ret void
+}
+
+define void @fabs_8f64() #0 {
+; SSE-LABEL: @fabs_8f64(
+; SSE-NEXT: [[TMP1:%.*]] = load <2 x double>, <2 x double>* bitcast ([8 x double]* @src64 to <2 x double>*), align 4
+; SSE-NEXT: [[TMP2:%.*]] = load <2 x double>, <2 x double>* bitcast (double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 2) to <2 x double>*), align 4
+; SSE-NEXT: [[TMP3:%.*]] = load <2 x double>, <2 x double>* bitcast (double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 4) to <2 x double>*), align 4
+; SSE-NEXT: [[TMP4:%.*]] = load <2 x double>, <2 x double>* bitcast (double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 6) to <2 x double>*), align 4
+; SSE-NEXT: [[TMP5:%.*]] = call <2 x double> @llvm.fabs.v2f64(<2 x double> [[TMP1]])
+; SSE-NEXT: [[TMP6:%.*]] = call <2 x double> @llvm.fabs.v2f64(<2 x double> [[TMP2]])
+; SSE-NEXT: [[TMP7:%.*]] = call <2 x double> @llvm.fabs.v2f64(<2 x double> [[TMP3]])
+; SSE-NEXT: [[TMP8:%.*]] = call <2 x double> @llvm.fabs.v2f64(<2 x double> [[TMP4]])
+; SSE-NEXT: store <2 x double> [[TMP5]], <2 x double>* bitcast ([8 x double]* @dst64 to <2 x double>*), align 4
+; SSE-NEXT: store <2 x double> [[TMP6]], <2 x double>* bitcast (double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 2) to <2 x double>*), align 4
+; SSE-NEXT: store <2 x double> [[TMP7]], <2 x double>* bitcast (double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 4) to <2 x double>*), align 4
+; SSE-NEXT: store <2 x double> [[TMP8]], <2 x double>* bitcast (double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 6) to <2 x double>*), align 4
+; SSE-NEXT: ret void
+;
+; AVX256-LABEL: @fabs_8f64(
+; AVX256-NEXT: [[TMP1:%.*]] = load <4 x double>, <4 x double>* bitcast ([8 x double]* @src64 to <4 x double>*), align 4
+; AVX256-NEXT: [[TMP2:%.*]] = load <4 x double>, <4 x double>* bitcast (double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 4) to <4 x double>*), align 4
+; AVX256-NEXT: [[TMP3:%.*]] = call <4 x double> @llvm.fabs.v4f64(<4 x double> [[TMP1]])
+; AVX256-NEXT: [[TMP4:%.*]] = call <4 x double> @llvm.fabs.v4f64(<4 x double> [[TMP2]])
+; AVX256-NEXT: store <4 x double> [[TMP3]], <4 x double>* bitcast ([8 x double]* @dst64 to <4 x double>*), align 4
+; AVX256-NEXT: store <4 x double> [[TMP4]], <4 x double>* bitcast (double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 4) to <4 x double>*), align 4
+; AVX256-NEXT: ret void
+;
+; AVX512-LABEL: @fabs_8f64(
+; AVX512-NEXT: [[TMP1:%.*]] = load <8 x double>, <8 x double>* bitcast ([8 x double]* @src64 to <8 x double>*), align 4
+; AVX512-NEXT: [[TMP2:%.*]] = call <8 x double> @llvm.fabs.v8f64(<8 x double> [[TMP1]])
+; AVX512-NEXT: store <8 x double> [[TMP2]], <8 x double>* bitcast ([8 x double]* @dst64 to <8 x double>*), align 4
+; AVX512-NEXT: ret void
+;
+ %a0 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 0), align 4
+ %a1 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 1), align 4
+ %a2 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 2), align 4
+ %a3 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 3), align 4
+ %a4 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 4), align 4
+ %a5 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 5), align 4
+ %a6 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 6), align 4
+ %a7 = load double, double* getelementptr inbounds ([8 x double], [8 x double]* @src64, i32 0, i64 7), align 4
+ %fabs0 = call double @llvm.fabs.f64(double %a0)
+ %fabs1 = call double @llvm.fabs.f64(double %a1)
+ %fabs2 = call double @llvm.fabs.f64(double %a2)
+ %fabs3 = call double @llvm.fabs.f64(double %a3)
+ %fabs4 = call double @llvm.fabs.f64(double %a4)
+ %fabs5 = call double @llvm.fabs.f64(double %a5)
+ %fabs6 = call double @llvm.fabs.f64(double %a6)
+ %fabs7 = call double @llvm.fabs.f64(double %a7)
+ store double %fabs0, double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 0), align 4
+ store double %fabs1, double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 1), align 4
+ store double %fabs2, double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 2), align 4
+ store double %fabs3, double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 3), align 4
+ store double %fabs4, double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 4), align 4
+ store double %fabs5, double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 5), align 4
+ store double %fabs6, double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 6), align 4
+ store double %fabs7, double* getelementptr inbounds ([8 x double], [8 x double]* @dst64, i32 0, i64 7), align 4
+ ret void
+}
+
+define void @fabs_4f32() #0 {
+; CHECK-LABEL: @fabs_4f32(
+; CHECK-NEXT: [[TMP1:%.*]] = load <4 x float>, <4 x float>* bitcast ([16 x float]* @src32 to <4 x float>*), align 4
+; CHECK-NEXT: [[TMP2:%.*]] = call <4 x float> @llvm.fabs.v4f32(<4 x float> [[TMP1]])
+; CHECK-NEXT: store <4 x float> [[TMP2]], <4 x float>* bitcast ([16 x float]* @dst32 to <4 x float>*), align 4
+; CHECK-NEXT: ret void
+;
+ %a0 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 0), align 4
+ %a1 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 1), align 4
+ %a2 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 2), align 4
+ %a3 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 3), align 4
+ %fabs0 = call float @llvm.fabs.f32(float %a0)
+ %fabs1 = call float @llvm.fabs.f32(float %a1)
+ %fabs2 = call float @llvm.fabs.f32(float %a2)
+ %fabs3 = call float @llvm.fabs.f32(float %a3)
+ store float %fabs0, float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 0), align 4
+ store float %fabs1, float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 1), align 4
+ store float %fabs2, float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 2), align 4
+ store float %fabs3, float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 3), align 4
+ ret void
+}
+
+define void @fabs_8f32() #0 {
+; SSE-LABEL: @fabs_8f32(
+; SSE-NEXT: [[TMP1:%.*]] = load <4 x float>, <4 x float>* bitcast ([16 x float]* @src32 to <4 x float>*), align 4
+; SSE-NEXT: [[TMP2:%.*]] = load <4 x float>, <4 x float>* bitcast (float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 4) to <4 x float>*), align 4
+; SSE-NEXT: [[TMP3:%.*]] = call <4 x float> @llvm.fabs.v4f32(<4 x float> [[TMP1]])
+; SSE-NEXT: [[TMP4:%.*]] = call <4 x float> @llvm.fabs.v4f32(<4 x float> [[TMP2]])
+; SSE-NEXT: store <4 x float> [[TMP3]], <4 x float>* bitcast ([16 x float]* @dst32 to <4 x float>*), align 4
+; SSE-NEXT: store <4 x float> [[TMP4]], <4 x float>* bitcast (float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 4) to <4 x float>*), align 4
+; SSE-NEXT: ret void
+;
+; AVX-LABEL: @fabs_8f32(
+; AVX-NEXT: [[TMP1:%.*]] = load <8 x float>, <8 x float>* bitcast ([16 x float]* @src32 to <8 x float>*), align 4
+; AVX-NEXT: [[TMP2:%.*]] = call <8 x float> @llvm.fabs.v8f32(<8 x float> [[TMP1]])
+; AVX-NEXT: store <8 x float> [[TMP2]], <8 x float>* bitcast ([16 x float]* @dst32 to <8 x float>*), align 4
+; AVX-NEXT: ret void
+;
+ %a0 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 0), align 4
+ %a1 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 1), align 4
+ %a2 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 2), align 4
+ %a3 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 3), align 4
+ %a4 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 4), align 4
+ %a5 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 5), align 4
+ %a6 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 6), align 4
+ %a7 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 7), align 4
+ %fabs0 = call float @llvm.fabs.f32(float %a0)
+ %fabs1 = call float @llvm.fabs.f32(float %a1)
+ %fabs2 = call float @llvm.fabs.f32(float %a2)
+ %fabs3 = call float @llvm.fabs.f32(float %a3)
+ %fabs4 = call float @llvm.fabs.f32(float %a4)
+ %fabs5 = call float @llvm.fabs.f32(float %a5)
+ %fabs6 = call float @llvm.fabs.f32(float %a6)
+ %fabs7 = call float @llvm.fabs.f32(float %a7)
+ store float %fabs0, float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 0), align 4
+ store float %fabs1, float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 1), align 4
+ store float %fabs2, float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 2), align 4
+ store float %fabs3, float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 3), align 4
+ store float %fabs4, float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 4), align 4
+ store float %fabs5, float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 5), align 4
+ store float %fabs6, float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 6), align 4
+ store float %fabs7, float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 7), align 4
+ ret void
+}
+
+define void @fabs_16f32() #0 {
+; SSE-LABEL: @fabs_16f32(
+; SSE-NEXT: [[TMP1:%.*]] = load <4 x float>, <4 x float>* bitcast ([16 x float]* @src32 to <4 x float>*), align 4
+; SSE-NEXT: [[TMP2:%.*]] = load <4 x float>, <4 x float>* bitcast (float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 4) to <4 x float>*), align 4
+; SSE-NEXT: [[TMP3:%.*]] = load <4 x float>, <4 x float>* bitcast (float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 8) to <4 x float>*), align 4
+; SSE-NEXT: [[TMP4:%.*]] = load <4 x float>, <4 x float>* bitcast (float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 12) to <4 x float>*), align 4
+; SSE-NEXT: [[TMP5:%.*]] = call <4 x float> @llvm.fabs.v4f32(<4 x float> [[TMP1]])
+; SSE-NEXT: [[TMP6:%.*]] = call <4 x float> @llvm.fabs.v4f32(<4 x float> [[TMP2]])
+; SSE-NEXT: [[TMP7:%.*]] = call <4 x float> @llvm.fabs.v4f32(<4 x float> [[TMP3]])
+; SSE-NEXT: [[TMP8:%.*]] = call <4 x float> @llvm.fabs.v4f32(<4 x float> [[TMP4]])
+; SSE-NEXT: store <4 x float> [[TMP5]], <4 x float>* bitcast ([16 x float]* @dst32 to <4 x float>*), align 4
+; SSE-NEXT: store <4 x float> [[TMP6]], <4 x float>* bitcast (float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 4) to <4 x float>*), align 4
+; SSE-NEXT: store <4 x float> [[TMP7]], <4 x float>* bitcast (float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 8) to <4 x float>*), align 4
+; SSE-NEXT: store <4 x float> [[TMP8]], <4 x float>* bitcast (float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 12) to <4 x float>*), align 4
+; SSE-NEXT: ret void
+;
+; AVX256-LABEL: @fabs_16f32(
+; AVX256-NEXT: [[TMP1:%.*]] = load <8 x float>, <8 x float>* bitcast ([16 x float]* @src32 to <8 x float>*), align 4
+; AVX256-NEXT: [[TMP2:%.*]] = load <8 x float>, <8 x float>* bitcast (float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 8) to <8 x float>*), align 4
+; AVX256-NEXT: [[TMP3:%.*]] = call <8 x float> @llvm.fabs.v8f32(<8 x float> [[TMP1]])
+; AVX256-NEXT: [[TMP4:%.*]] = call <8 x float> @llvm.fabs.v8f32(<8 x float> [[TMP2]])
+; AVX256-NEXT: store <8 x float> [[TMP3]], <8 x float>* bitcast ([16 x float]* @dst32 to <8 x float>*), align 4
+; AVX256-NEXT: store <8 x float> [[TMP4]], <8 x float>* bitcast (float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 8) to <8 x float>*), align 4
+; AVX256-NEXT: ret void
+;
+; AVX512-LABEL: @fabs_16f32(
+; AVX512-NEXT: [[TMP1:%.*]] = load <16 x float>, <16 x float>* bitcast ([16 x float]* @src32 to <16 x float>*), align 4
+; AVX512-NEXT: [[TMP2:%.*]] = call <16 x float> @llvm.fabs.v16f32(<16 x float> [[TMP1]])
+; AVX512-NEXT: store <16 x float> [[TMP2]], <16 x float>* bitcast ([16 x float]* @dst32 to <16 x float>*), align 4
+; AVX512-NEXT: ret void
+;
+ %a0 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 0), align 4
+ %a1 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 1), align 4
+ %a2 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 2), align 4
+ %a3 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 3), align 4
+ %a4 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 4), align 4
+ %a5 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 5), align 4
+ %a6 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 6), align 4
+ %a7 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 7), align 4
+ %a8 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 8), align 4
+ %a9 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 9), align 4
+ %a10 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 10), align 4
+ %a11 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 11), align 4
+ %a12 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 12), align 4
+ %a13 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 13), align 4
+ %a14 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 14), align 4
+ %a15 = load float, float* getelementptr inbounds ([16 x float], [16 x float]* @src32, i32 0, i64 15), align 4
+ %fabs0 = call float @llvm.fabs.f32(float %a0 )
+ %fabs1 = call float @llvm.fabs.f32(float %a1 )
+ %fabs2 = call float @llvm.fabs.f32(float %a2 )
+ %fabs3 = call float @llvm.fabs.f32(float %a3 )
+ %fabs4 = call float @llvm.fabs.f32(float %a4 )
+ %fabs5 = call float @llvm.fabs.f32(float %a5 )
+ %fabs6 = call float @llvm.fabs.f32(float %a6 )
+ %fabs7 = call float @llvm.fabs.f32(float %a7 )
+ %fabs8 = call float @llvm.fabs.f32(float %a8 )
+ %fabs9 = call float @llvm.fabs.f32(float %a9 )
+ %fabs10 = call float @llvm.fabs.f32(float %a10)
+ %fabs11 = call float @llvm.fabs.f32(float %a11)
+ %fabs12 = call float @llvm.fabs.f32(float %a12)
+ %fabs13 = call float @llvm.fabs.f32(float %a13)
+ %fabs14 = call float @llvm.fabs.f32(float %a14)
+ %fabs15 = call float @llvm.fabs.f32(float %a15)
+ store float %fabs0 , float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 0), align 4
+ store float %fabs1 , float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 1), align 4
+ store float %fabs2 , float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 2), align 4
+ store float %fabs3 , float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 3), align 4
+ store float %fabs4 , float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 4), align 4
+ store float %fabs5 , float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 5), align 4
+ store float %fabs6 , float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 6), align 4
+ store float %fabs7 , float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 7), align 4
+ store float %fabs8 , float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 8), align 4
+ store float %fabs9 , float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 9), align 4
+ store float %fabs10, float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 10), align 4
+ store float %fabs11, float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 11), align 4
+ store float %fabs12, float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 12), align 4
+ store float %fabs13, float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 13), align 4
+ store float %fabs14, float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 14), align 4
+ store float %fabs15, float* getelementptr inbounds ([16 x float], [16 x float]* @dst32, i32 0, i64 15), align 4
+ ret void
+}
+
+attributes #0 = { nounwind }
More information about the llvm-commits
mailing list