[llvm] r304826 - [SLP] Add a test for fix of PR32164, NFC.

Alexey Bataev via llvm-commits llvm-commits at lists.llvm.org
Tue Jun 6 13:11:35 PDT 2017


Author: abataev
Date: Tue Jun  6 15:11:35 2017
New Revision: 304826

URL: http://llvm.org/viewvc/llvm-project?rev=304826&view=rev
Log:
[SLP] Add a test for fix of PR32164, NFC.

Added:
    llvm/trunk/test/Transforms/SLPVectorizer/X86/reverse_extract_elements.cpp

Added: llvm/trunk/test/Transforms/SLPVectorizer/X86/reverse_extract_elements.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Transforms/SLPVectorizer/X86/reverse_extract_elements.cpp?rev=304826&view=auto
==============================================================================
--- llvm/trunk/test/Transforms/SLPVectorizer/X86/reverse_extract_elements.cpp (added)
+++ llvm/trunk/test/Transforms/SLPVectorizer/X86/reverse_extract_elements.cpp Tue Jun  6 15:11:35 2017
@@ -0,0 +1,138 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt < %s -slp-vectorizer -instcombine -S -mtriple=x86_64-unknown-linux-gnu -mcpu=corei7 | FileCheck %s
+
+define float @dotf(<4 x float> %x, <4 x float> %y) {
+; CHECK-LABEL: @dotf(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = fmul fast <4 x float> [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <4 x float> [[TMP0]], <4 x float> undef, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
+; CHECK-NEXT:    [[RDX_SHUF:%.*]] = shufflevector <4 x float> [[TMP1]], <4 x float> undef, <4 x i32> <i32 2, i32 3, i32 undef, i32 undef>
+; CHECK-NEXT:    [[BIN_RDX:%.*]] = fadd fast <4 x float> [[TMP1]], [[RDX_SHUF]]
+; CHECK-NEXT:    [[RDX_SHUF1:%.*]] = shufflevector <4 x float> [[BIN_RDX]], <4 x float> undef, <4 x i32> <i32 1, i32 undef, i32 undef, i32 undef>
+; CHECK-NEXT:    [[BIN_RDX2:%.*]] = fadd fast <4 x float> [[BIN_RDX]], [[RDX_SHUF1]]
+; CHECK-NEXT:    [[TMP2:%.*]] = extractelement <4 x float> [[BIN_RDX2]], i32 0
+; CHECK-NEXT:    ret float [[TMP2]]
+;
+entry:
+  %vecext = extractelement <4 x float> %x, i32 0
+  %vecext1 = extractelement <4 x float> %y, i32 0
+  %mul = fmul fast float %vecext, %vecext1
+  %vecext.1 = extractelement <4 x float> %x, i32 1
+  %vecext1.1 = extractelement <4 x float> %y, i32 1
+  %mul.1 = fmul fast float %vecext.1, %vecext1.1
+  %add.1 = fadd fast float %mul.1, %mul
+  %vecext.2 = extractelement <4 x float> %x, i32 2
+  %vecext1.2 = extractelement <4 x float> %y, i32 2
+  %mul.2 = fmul fast float %vecext.2, %vecext1.2
+  %add.2 = fadd fast float %mul.2, %add.1
+  %vecext.3 = extractelement <4 x float> %x, i32 3
+  %vecext1.3 = extractelement <4 x float> %y, i32 3
+  %mul.3 = fmul fast float %vecext.3, %vecext1.3
+  %add.3 = fadd fast float %mul.3, %add.2
+  ret float %add.3
+}
+
+define double @dotd(<4 x double>* byval nocapture readonly align 32, <4 x double>* byval nocapture readonly align 32) {
+; CHECK-LABEL: @dotd(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[X:%.*]] = load <4 x double>, <4 x double>* [[TMP0:%.*]], align 32
+; CHECK-NEXT:    [[Y:%.*]] = load <4 x double>, <4 x double>* [[TMP1:%.*]], align 32
+; CHECK-NEXT:    [[TMP2:%.*]] = fmul fast <4 x double> [[X]], [[Y]]
+; CHECK-NEXT:    [[TMP3:%.*]] = shufflevector <4 x double> [[TMP2]], <4 x double> undef, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
+; CHECK-NEXT:    [[RDX_SHUF:%.*]] = shufflevector <4 x double> [[TMP3]], <4 x double> undef, <4 x i32> <i32 2, i32 3, i32 undef, i32 undef>
+; CHECK-NEXT:    [[BIN_RDX:%.*]] = fadd fast <4 x double> [[TMP3]], [[RDX_SHUF]]
+; CHECK-NEXT:    [[RDX_SHUF1:%.*]] = shufflevector <4 x double> [[BIN_RDX]], <4 x double> undef, <4 x i32> <i32 1, i32 undef, i32 undef, i32 undef>
+; CHECK-NEXT:    [[BIN_RDX2:%.*]] = fadd fast <4 x double> [[BIN_RDX]], [[RDX_SHUF1]]
+; CHECK-NEXT:    [[TMP4:%.*]] = extractelement <4 x double> [[BIN_RDX2]], i32 0
+; CHECK-NEXT:    ret double [[TMP4]]
+;
+entry:
+  %x = load <4 x double>, <4 x double>* %0, align 32
+  %y = load <4 x double>, <4 x double>* %1, align 32
+  %vecext = extractelement <4 x double> %x, i32 0
+  %vecext1 = extractelement <4 x double> %y, i32 0
+  %mul = fmul fast double %vecext, %vecext1
+  %vecext.1 = extractelement <4 x double> %x, i32 1
+  %vecext1.1 = extractelement <4 x double> %y, i32 1
+  %mul.1 = fmul fast double %vecext.1, %vecext1.1
+  %add.1 = fadd fast double %mul.1, %mul
+  %vecext.2 = extractelement <4 x double> %x, i32 2
+  %vecext1.2 = extractelement <4 x double> %y, i32 2
+  %mul.2 = fmul fast double %vecext.2, %vecext1.2
+  %add.2 = fadd fast double %mul.2, %add.1
+  %vecext.3 = extractelement <4 x double> %x, i32 3
+  %vecext1.3 = extractelement <4 x double> %y, i32 3
+  %mul.3 = fmul fast double %vecext.3, %vecext1.3
+  %add.3 = fadd fast double %mul.3, %add.2
+  ret double %add.3
+}
+
+define float @dotfq(<4 x float>* nocapture readonly %x, <4 x float>* nocapture readonly %y) {
+; CHECK-LABEL: @dotfq(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load <4 x float>, <4 x float>* [[X:%.*]], align 16
+; CHECK-NEXT:    [[TMP1:%.*]] = load <4 x float>, <4 x float>* [[Y:%.*]], align 16
+; CHECK-NEXT:    [[TMP2:%.*]] = fmul fast <4 x float> [[TMP1]], [[TMP0]]
+; CHECK-NEXT:    [[TMP3:%.*]] = shufflevector <4 x float> [[TMP2]], <4 x float> undef, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
+; CHECK-NEXT:    [[RDX_SHUF:%.*]] = shufflevector <4 x float> [[TMP3]], <4 x float> undef, <4 x i32> <i32 2, i32 3, i32 undef, i32 undef>
+; CHECK-NEXT:    [[BIN_RDX:%.*]] = fadd fast <4 x float> [[TMP3]], [[RDX_SHUF]]
+; CHECK-NEXT:    [[RDX_SHUF1:%.*]] = shufflevector <4 x float> [[BIN_RDX]], <4 x float> undef, <4 x i32> <i32 1, i32 undef, i32 undef, i32 undef>
+; CHECK-NEXT:    [[BIN_RDX2:%.*]] = fadd fast <4 x float> [[BIN_RDX]], [[RDX_SHUF1]]
+; CHECK-NEXT:    [[TMP4:%.*]] = extractelement <4 x float> [[BIN_RDX2]], i32 0
+; CHECK-NEXT:    ret float [[TMP4]]
+;
+entry:
+  %0 = load <4 x float>, <4 x float>* %x, align 16
+  %1 = load <4 x float>, <4 x float>* %y, align 16
+  %vecext = extractelement <4 x float> %0, i32 0
+  %vecext1 = extractelement <4 x float> %1, i32 0
+  %mul = fmul fast float %vecext1, %vecext
+  %vecext.1 = extractelement <4 x float> %0, i32 1
+  %vecext1.1 = extractelement <4 x float> %1, i32 1
+  %mul.1 = fmul fast float %vecext1.1, %vecext.1
+  %add.1 = fadd fast float %mul.1, %mul
+  %vecext.2 = extractelement <4 x float> %0, i32 2
+  %vecext1.2 = extractelement <4 x float> %1, i32 2
+  %mul.2 = fmul fast float %vecext1.2, %vecext.2
+  %add.2 = fadd fast float %mul.2, %add.1
+  %vecext.3 = extractelement <4 x float> %0, i32 3
+  %vecext1.3 = extractelement <4 x float> %1, i32 3
+  %mul.3 = fmul fast float %vecext1.3, %vecext.3
+  %add.3 = fadd fast float %mul.3, %add.2
+  ret float %add.3
+}
+
+define double @dotdq(<4 x double>* nocapture readonly %x, <4 x double>* nocapture readonly %y) {
+; CHECK-LABEL: @dotdq(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load <4 x double>, <4 x double>* [[X:%.*]], align 32
+; CHECK-NEXT:    [[TMP1:%.*]] = load <4 x double>, <4 x double>* [[Y:%.*]], align 32
+; CHECK-NEXT:    [[TMP2:%.*]] = fmul fast <4 x double> [[TMP1]], [[TMP0]]
+; CHECK-NEXT:    [[TMP3:%.*]] = shufflevector <4 x double> [[TMP2]], <4 x double> undef, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
+; CHECK-NEXT:    [[RDX_SHUF:%.*]] = shufflevector <4 x double> [[TMP3]], <4 x double> undef, <4 x i32> <i32 2, i32 3, i32 undef, i32 undef>
+; CHECK-NEXT:    [[BIN_RDX:%.*]] = fadd fast <4 x double> [[TMP3]], [[RDX_SHUF]]
+; CHECK-NEXT:    [[RDX_SHUF1:%.*]] = shufflevector <4 x double> [[BIN_RDX]], <4 x double> undef, <4 x i32> <i32 1, i32 undef, i32 undef, i32 undef>
+; CHECK-NEXT:    [[BIN_RDX2:%.*]] = fadd fast <4 x double> [[BIN_RDX]], [[RDX_SHUF1]]
+; CHECK-NEXT:    [[TMP4:%.*]] = extractelement <4 x double> [[BIN_RDX2]], i32 0
+; CHECK-NEXT:    ret double [[TMP4]]
+;
+entry:
+  %0 = load <4 x double>, <4 x double>* %x, align 32
+  %1 = load <4 x double>, <4 x double>* %y, align 32
+  %vecext = extractelement <4 x double> %0, i32 0
+  %vecext1 = extractelement <4 x double> %1, i32 0
+  %mul = fmul fast double %vecext1, %vecext
+  %vecext.1 = extractelement <4 x double> %0, i32 1
+  %vecext1.1 = extractelement <4 x double> %1, i32 1
+  %mul.1 = fmul fast double %vecext1.1, %vecext.1
+  %add.1 = fadd fast double %mul.1, %mul
+  %vecext.2 = extractelement <4 x double> %0, i32 2
+  %vecext1.2 = extractelement <4 x double> %1, i32 2
+  %mul.2 = fmul fast double %vecext1.2, %vecext.2
+  %add.2 = fadd fast double %mul.2, %add.1
+  %vecext.3 = extractelement <4 x double> %0, i32 3
+  %vecext1.3 = extractelement <4 x double> %1, i32 3
+  %mul.3 = fmul fast double %vecext1.3, %vecext.3
+  %add.3 = fadd fast double %mul.3, %add.2
+  ret double %add.3
+}




More information about the llvm-commits mailing list