From: Simon Pilgrim Date: Sat, 14 Nov 2015 20:42:01 +0000 (+0000) Subject: [X86][SSE] Fixed arch/triple and regenerated results. X-Git-Url: http://demsky.eecs.uci.edu/git/?a=commitdiff_plain;h=2cfd2548d0d5fadf78a147b5d1c204b10b99f504;p=oota-llvm.git [X86][SSE] Fixed arch/triple and regenerated results. Tidyup before diffs from new patch. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@253144 91177308-0d34-0410-b5e6-96231b3b80d8 --- diff --git a/test/CodeGen/X86/2011-10-21-widen-cmp.ll b/test/CodeGen/X86/2011-10-21-widen-cmp.ll index 2fe645b0781..115d828d4cf 100644 --- a/test/CodeGen/X86/2011-10-21-widen-cmp.ll +++ b/test/CodeGen/X86/2011-10-21-widen-cmp.ll @@ -1,15 +1,22 @@ -; RUN: llc < %s -march=x86-64 -mcpu=corei7 | FileCheck %s - -target triple = "x86_64-unknown-linux-gnu" +; RUN: llc < %s -mtriple=x86_64-unknown-linux-gnu -mcpu=corei7 | FileCheck %s ; Check that a <4 x float> compare is generated and that we are ; not stuck in an endless loop. -; CHECK: cmp_2_floats -; CHECK: cmpordps -; CHECK: ret - define void @cmp_2_floats() { +; CHECK-LABEL: cmp_2_floats: +; CHECK: # BB#0: # %entry +; CHECK-NEXT: cmpordps %xmm0, %xmm0 +; CHECK-NEXT: pmovzxdq {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero +; CHECK-NEXT: psllq $32, %xmm0 +; CHECK-NEXT: pshufd {{.*#+}} xmm1 = xmm0[1,1,3,3] +; CHECK-NEXT: psrad $31, %xmm0 +; CHECK-NEXT: pblendw {{.*#+}} xmm0 = xmm1[0,1],xmm0[2,3],xmm1[4,5],xmm0[6,7] +; CHECK-NEXT: pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3] +; CHECK-NEXT: pslld $31, %xmm0 +; CHECK-NEXT: blendvps %xmm0, %xmm0 +; CHECK-NEXT: movlps %xmm0, (%rax) +; CHECK-NEXT: retq entry: %0 = fcmp oeq <2 x float> undef, undef %1 = select <2 x i1> %0, <2 x float> undef, <2 x float> undef @@ -17,11 +24,13 @@ entry: ret void } -; CHECK: cmp_2_doubles -; CHECK: cmpordpd -; CHECK: blendvpd -; CHECK: ret define void @cmp_2_doubles() { +; CHECK-LABEL: cmp_2_doubles: +; CHECK: # BB#0: # %entry +; CHECK-NEXT: cmpordpd %xmm0, %xmm0 +; CHECK-NEXT: blendvpd %xmm0, %xmm0 +; CHECK-NEXT: movapd %xmm0, (%rax) +; CHECK-NEXT: retq entry: %0 = fcmp oeq <2 x double> undef, undef %1 = select <2 x i1> %0, <2 x double> undef, <2 x double> undef @@ -29,11 +38,28 @@ entry: ret void } -; CHECK: mp_11193 -; CHECK: psraw $15 -; CHECK: ret -define void @mp_11193(<8 x float> * nocapture %aFOO, <8 x float>* nocapture %RET) -nounwind { +define void @mp_11193(<8 x float> * nocapture %aFOO, <8 x float>* nocapture %RET) nounwind { +; CHECK-LABEL: mp_11193: +; CHECK: # BB#0: # %allocas +; CHECK-NEXT: movaps {{.*#+}} xmm0 = [3.000000e+00,3.000000e+00,3.000000e+00,3.000000e+00] +; CHECK-NEXT: movaps {{.*#+}} xmm1 = [9.000000e+00,1.000000e+00,9.000000e+00,1.000000e+00] +; CHECK-NEXT: cmpltps %xmm0, %xmm1 +; CHECK-NEXT: movdqa {{.*#+}} xmm2 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15] +; CHECK-NEXT: pshufb %xmm2, %xmm1 +; CHECK-NEXT: movaps {{.*#+}} xmm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] +; CHECK-NEXT: cmpltps %xmm0, %xmm3 +; CHECK-NEXT: pshufb %xmm2, %xmm3 +; CHECK-NEXT: punpcklqdq {{.*#+}} xmm3 = xmm3[0],xmm1[0] +; CHECK-NEXT: psllw $15, %xmm3 +; CHECK-NEXT: psraw $15, %xmm3 +; CHECK-NEXT: pextrb $0, %xmm3, %eax +; CHECK-NEXT: shlb $7, %al +; CHECK-NEXT: sarb $7, %al +; CHECK-NEXT: movsbl %al, %eax +; CHECK-NEXT: xorps %xmm0, %xmm0 +; CHECK-NEXT: cvtsi2ssl %eax, %xmm0 +; CHECK-NEXT: movss %xmm0, (%rsi) +; CHECK-NEXT: retq allocas: %bincmp = fcmp olt <8 x float> , %t = extractelement <8 x i1> %bincmp, i32 0 diff --git a/test/CodeGen/X86/2011-20-21-zext-ui2fp.ll b/test/CodeGen/X86/2011-20-21-zext-ui2fp.ll index 78cdfcf0e1f..dd8c3efa83b 100644 --- a/test/CodeGen/X86/2011-20-21-zext-ui2fp.ll +++ b/test/CodeGen/X86/2011-20-21-zext-ui2fp.ll @@ -1,13 +1,41 @@ -; RUN: llc < %s -march=x86-64 -mcpu=corei7 | FileCheck %s -target triple = "x86_64-unknown-linux-gnu" +; RUN: llc < %s -mtriple=x86_64-unknown-linux-gnu -mcpu=corei7 | FileCheck %s ; Check that the booleans are converted using zext and not via sext. ; 0x1 means that we only look at the first bit. -;CHECK: 0x1 -;CHECK-LABEL: ui_to_fp_conv: -;CHECK: ret define void @ui_to_fp_conv(<8 x float> * nocapture %aFOO, <8 x float>* nocapture %RET) nounwind { +; CHECK: 0x1 +; CHECK-LABEL: ui_to_fp_conv: +; CHECK: # BB#0: # %allocas +; CHECK-NEXT: movaps {{.*#+}} xmm0 = [1.000000e+00,1.000000e+00,3.000000e+00,3.000000e+00] +; CHECK-NEXT: cmpltps {{.*}}(%rip), %xmm0 +; CHECK-NEXT: pshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15] +; CHECK-NEXT: pxor %xmm1, %xmm1 +; CHECK-NEXT: punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0] +; CHECK-NEXT: psllw $15, %xmm0 +; CHECK-NEXT: psraw $15, %xmm0 +; CHECK-NEXT: pmovzxwd {{.*#+}} xmm1 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero +; CHECK-NEXT: movdqa {{.*#+}} xmm2 = [1,1,1,1] +; CHECK-NEXT: pand %xmm2, %xmm1 +; CHECK-NEXT: movdqa {{.*#+}} xmm3 = [1258291200,1258291200,1258291200,1258291200] +; CHECK-NEXT: movdqa %xmm1, %xmm4 +; CHECK-NEXT: pblendw {{.*#+}} xmm4 = xmm4[0],xmm3[1],xmm4[2],xmm3[3],xmm4[4],xmm3[5],xmm4[6],xmm3[7] +; CHECK-NEXT: psrld $16, %xmm1 +; CHECK-NEXT: movdqa {{.*#+}} xmm5 = [1392508928,1392508928,1392508928,1392508928] +; CHECK-NEXT: pblendw {{.*#+}} xmm1 = xmm1[0],xmm5[1],xmm1[2],xmm5[3],xmm1[4],xmm5[5],xmm1[6],xmm5[7] +; CHECK-NEXT: movaps {{.*#+}} xmm6 = [-5.497642e+11,-5.497642e+11,-5.497642e+11,-5.497642e+11] +; CHECK-NEXT: addps %xmm6, %xmm1 +; CHECK-NEXT: addps %xmm4, %xmm1 +; CHECK-NEXT: punpckhwd {{.*#+}} xmm0 = xmm0[4,4,5,5,6,6,7,7] +; CHECK-NEXT: pand %xmm2, %xmm0 +; CHECK-NEXT: pblendw {{.*#+}} xmm3 = xmm0[0],xmm3[1],xmm0[2],xmm3[3],xmm0[4],xmm3[5],xmm0[6],xmm3[7] +; CHECK-NEXT: psrld $16, %xmm0 +; CHECK-NEXT: pblendw {{.*#+}} xmm0 = xmm0[0],xmm5[1],xmm0[2],xmm5[3],xmm0[4],xmm5[5],xmm0[6],xmm5[7] +; CHECK-NEXT: addps %xmm6, %xmm0 +; CHECK-NEXT: addps %xmm3, %xmm0 +; CHECK-NEXT: movups %xmm0, 16(%rsi) +; CHECK-NEXT: movups %xmm1, (%rsi) +; CHECK-NEXT: retq allocas: %bincmp = fcmp olt <8 x float> , %bool2float = uitofp <8 x i1> %bincmp to <8 x float>