From d040c1a1159923fc5db9915ae06c9187119848ea Mon Sep 17 00:00:00 2001 From: Simon Pilgrim Date: Sun, 24 Jul 2022 14:48:18 +0100 Subject: [PATCH] [X86] Regenerate hoist-and-by-const-from-lshr-in-eqcmp-zero.ll with common SSE2 prefix --- .../hoist-and-by-const-from-lshr-in-eqcmp-zero.ll | 74 ++++++++-------------- 1 file changed, 27 insertions(+), 47 deletions(-) diff --git a/llvm/test/CodeGen/X86/hoist-and-by-const-from-lshr-in-eqcmp-zero.ll b/llvm/test/CodeGen/X86/hoist-and-by-const-from-lshr-in-eqcmp-zero.ll index 2ec45dc..d79aff5 100644 --- a/llvm/test/CodeGen/X86/hoist-and-by-const-from-lshr-in-eqcmp-zero.ll +++ b/llvm/test/CodeGen/X86/hoist-and-by-const-from-lshr-in-eqcmp-zero.ll @@ -1,12 +1,12 @@ ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py -; RUN: llc -mtriple=i686-unknown-linux-gnu -mattr=+sse,sse2 < %s | FileCheck %s --check-prefixes=CHECK,X86,X86-SSE2,X86-BMI1 -; RUN: llc -mtriple=i686-unknown-linux-gnu -mattr=+sse,sse2,+bmi < %s | FileCheck %s --check-prefixes=CHECK,X86,X86-SSE2,X86-BMI1 -; RUN: llc -mtriple=i686-unknown-linux-gnu -mattr=+sse,sse2,+bmi,+bmi2 < %s | FileCheck %s --check-prefixes=CHECK,X86,X86-SSE2,X86-BMI2 -; RUN: llc -mtriple=i686-unknown-linux-gnu -mattr=+sse,sse2,+bmi,+bmi2,+avx2 < %s | FileCheck %s --check-prefixes=CHECK,X86,X86-BMI2,AVX2 -; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mattr=+sse,sse2 < %s | FileCheck %s --check-prefixes=CHECK,X64,X64-SSE2,X64-BMI1 -; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mattr=+sse,sse2,+bmi < %s | FileCheck %s --check-prefixes=CHECK,X64,X64-SSE2,X64-BMI1 -; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mattr=+sse,sse2,+bmi,+bmi2 < %s | FileCheck %s --check-prefixes=CHECK,X64,X64-SSE2,X64-BMI2 -; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mattr=+sse,sse2,+bmi,+bmi2,+avx2 < %s | FileCheck %s --check-prefixes=CHECK,X64,X64-BMI2,AVX2 +; RUN: llc -mtriple=i686-unknown-linux-gnu -mattr=+sse,sse2 < %s | FileCheck %s --check-prefixes=CHECK,SSE2,X86,X86-SSE2,X86-BMI1 +; RUN: llc -mtriple=i686-unknown-linux-gnu -mattr=+sse,sse2,+bmi < %s | FileCheck %s --check-prefixes=CHECK,SSE2,X86,X86-SSE2,X86-BMI1 +; RUN: llc -mtriple=i686-unknown-linux-gnu -mattr=+sse,sse2,+bmi,+bmi2 < %s | FileCheck %s --check-prefixes=CHECK,SSE2,X86,X86-SSE2,X86-BMI2 +; RUN: llc -mtriple=i686-unknown-linux-gnu -mattr=+sse,sse2,+bmi,+bmi2,+avx2 < %s | FileCheck %s --check-prefixes=CHECK,AVX2,X86,X86-BMI2 +; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mattr=+sse,sse2 < %s | FileCheck %s --check-prefixes=CHECK,SSE2,X64,X64-SSE2,X64-BMI1 +; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mattr=+sse,sse2,+bmi < %s | FileCheck %s --check-prefixes=CHECK,SSE2,X64,X64-SSE2,X64-BMI1 +; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mattr=+sse,sse2,+bmi,+bmi2 < %s | FileCheck %s --check-prefixes=CHECK,SSE2,X64,X64-SSE2,X64-BMI2 +; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mattr=+sse,sse2,+bmi,+bmi2,+avx2 < %s | FileCheck %s --check-prefixes=CHECK,AVX2,X64,X64-BMI2 ; We are looking for the following pattern here: ; (X & (C l>> Y)) ==/!= 0 @@ -544,25 +544,25 @@ define <4 x i1> @vec_4xi32_splat_eq(<4 x i32> %x, <4 x i32> %y) nounwind { } define <4 x i1> @vec_4xi32_nonsplat_eq(<4 x i32> %x, <4 x i32> %y) nounwind { -; X86-SSE2-LABEL: vec_4xi32_nonsplat_eq: -; X86-SSE2: # %bb.0: -; X86-SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm1[2,3,2,3] -; X86-SSE2-NEXT: pshuflw {{.*#+}} xmm3 = xmm2[2,3,3,3,4,5,6,7] -; X86-SSE2-NEXT: movdqa {{.*#+}} xmm4 = [0,1,16776960,2147483648] -; X86-SSE2-NEXT: movdqa %xmm4, %xmm5 -; X86-SSE2-NEXT: psrld %xmm3, %xmm5 -; X86-SSE2-NEXT: pshuflw {{.*#+}} xmm2 = xmm2[0,1,1,1,4,5,6,7] -; X86-SSE2-NEXT: movdqa %xmm4, %xmm3 -; X86-SSE2-NEXT: psrld %xmm2, %xmm3 -; X86-SSE2-NEXT: punpckhqdq {{.*#+}} xmm3 = xmm3[1],xmm5[1] -; X86-SSE2-NEXT: pshuflw {{.*#+}} xmm1 = xmm1[2,3,3,3,4,5,6,7] -; X86-SSE2-NEXT: psrld %xmm1, %xmm4 -; X86-SSE2-NEXT: pslldq {{.*#+}} xmm4 = zero,zero,zero,zero,zero,zero,zero,zero,xmm4[0,1,2,3,4,5,6,7] -; X86-SSE2-NEXT: shufps {{.*#+}} xmm4 = xmm4[0,3],xmm3[0,3] -; X86-SSE2-NEXT: andps %xmm4, %xmm0 -; X86-SSE2-NEXT: pxor %xmm1, %xmm1 -; X86-SSE2-NEXT: pcmpeqd %xmm1, %xmm0 -; X86-SSE2-NEXT: retl +; SSE2-LABEL: vec_4xi32_nonsplat_eq: +; SSE2: # %bb.0: +; SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm1[2,3,2,3] +; SSE2-NEXT: pshuflw {{.*#+}} xmm3 = xmm2[2,3,3,3,4,5,6,7] +; SSE2-NEXT: movdqa {{.*#+}} xmm4 = [0,1,16776960,2147483648] +; SSE2-NEXT: movdqa %xmm4, %xmm5 +; SSE2-NEXT: psrld %xmm3, %xmm5 +; SSE2-NEXT: pshuflw {{.*#+}} xmm2 = xmm2[0,1,1,1,4,5,6,7] +; SSE2-NEXT: movdqa %xmm4, %xmm3 +; SSE2-NEXT: psrld %xmm2, %xmm3 +; SSE2-NEXT: punpckhqdq {{.*#+}} xmm3 = xmm3[1],xmm5[1] +; SSE2-NEXT: pshuflw {{.*#+}} xmm1 = xmm1[2,3,3,3,4,5,6,7] +; SSE2-NEXT: psrld %xmm1, %xmm4 +; SSE2-NEXT: pslldq {{.*#+}} xmm4 = zero,zero,zero,zero,zero,zero,zero,zero,xmm4[0,1,2,3,4,5,6,7] +; SSE2-NEXT: shufps {{.*#+}} xmm4 = xmm4[0,3],xmm3[0,3] +; SSE2-NEXT: andps %xmm4, %xmm0 +; SSE2-NEXT: pxor %xmm1, %xmm1 +; SSE2-NEXT: pcmpeqd %xmm1, %xmm0 +; SSE2-NEXT: ret{{[l|q]}} ; ; AVX2-LABEL: vec_4xi32_nonsplat_eq: ; AVX2: # %bb.0: @@ -572,26 +572,6 @@ define <4 x i1> @vec_4xi32_nonsplat_eq(<4 x i32> %x, <4 x i32> %y) nounwind { ; AVX2-NEXT: vpxor %xmm1, %xmm1, %xmm1 ; AVX2-NEXT: vpcmpeqd %xmm1, %xmm0, %xmm0 ; AVX2-NEXT: ret{{[l|q]}} -; -; X64-SSE2-LABEL: vec_4xi32_nonsplat_eq: -; X64-SSE2: # %bb.0: -; X64-SSE2-NEXT: pshufd {{.*#+}} xmm2 = xmm1[2,3,2,3] -; X64-SSE2-NEXT: pshuflw {{.*#+}} xmm3 = xmm2[2,3,3,3,4,5,6,7] -; X64-SSE2-NEXT: movdqa {{.*#+}} xmm4 = [0,1,16776960,2147483648] -; X64-SSE2-NEXT: movdqa %xmm4, %xmm5 -; X64-SSE2-NEXT: psrld %xmm3, %xmm5 -; X64-SSE2-NEXT: pshuflw {{.*#+}} xmm2 = xmm2[0,1,1,1,4,5,6,7] -; X64-SSE2-NEXT: movdqa %xmm4, %xmm3 -; X64-SSE2-NEXT: psrld %xmm2, %xmm3 -; X64-SSE2-NEXT: punpckhqdq {{.*#+}} xmm3 = xmm3[1],xmm5[1] -; X64-SSE2-NEXT: pshuflw {{.*#+}} xmm1 = xmm1[2,3,3,3,4,5,6,7] -; X64-SSE2-NEXT: psrld %xmm1, %xmm4 -; X64-SSE2-NEXT: pslldq {{.*#+}} xmm4 = zero,zero,zero,zero,zero,zero,zero,zero,xmm4[0,1,2,3,4,5,6,7] -; X64-SSE2-NEXT: shufps {{.*#+}} xmm4 = xmm4[0,3],xmm3[0,3] -; X64-SSE2-NEXT: andps %xmm4, %xmm0 -; X64-SSE2-NEXT: pxor %xmm1, %xmm1 -; X64-SSE2-NEXT: pcmpeqd %xmm1, %xmm0 -; X64-SSE2-NEXT: retq %t0 = lshr <4 x i32> , %y %t1 = and <4 x i32> %t0, %x %res = icmp eq <4 x i32> %t1, -- 2.7.4