maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity

[X86][AVX] Improve (i4 bitcast (v4i1 x)) handling for 256-bit vector compare results. 

As commented on D37849 and rL313547, AVX1 targets were missing a chance to use vmovmskpd for v4f64/v4i64 results for bool vector bitcasts

llvm-svn: 314293
This commit is contained in:
Simon Pilgrim 2017-09-27 10:10:17 +00:00
parent a932bfcc93
commit 3b0d9e789e
2 changed files with 10 additions and 20 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
llvm/lib/Target/X86/X86ISelLowering.cpp
View File

@ -29321,7 +29321,7 @@ static SDValue combineBitcastvxi1(SelectionDAG &DAG, SDValue BitCast,
// sign-extend to a 256-bit operation to avoid truncation.
if (N0->getOpcode() == ISD::SETCC &&
N0->getOperand(0)->getValueType(0).is256BitVector() &&
Subtarget.hasInt256()) {
Subtarget.hasAVX()) {
SExtVT = MVT::v4i64;
FPCastVT = MVT::v4f64;
}

28
llvm/test/CodeGen/X86/bitcast-setcc-256.ll
View File

@ -222,8 +222,8 @@ define i4 @v4i64(<4 x i64> %a, <4 x i64> %b) {
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm3
; AVX1-NEXT: vpcmpgtq %xmm2, %xmm3, %xmm2
; AVX1-NEXT: vpcmpgtq %xmm1, %xmm0, %xmm0
; AVX1-NEXT: vpacksswb %xmm2, %xmm0, %xmm0
; AVX1-NEXT: vmovmskps %xmm0, %eax
; AVX1-NEXT: vinsertf128 $1, %xmm2, %ymm0, %ymm0
; AVX1-NEXT: vmovmskpd %ymm0, %eax
; AVX1-NEXT: # kill: %AL<def> %AL<kill> %EAX<kill>
; AVX1-NEXT: vzeroupper
; AVX1-NEXT: retq
@ -259,23 +259,13 @@ define i4 @v4f64(<4 x double> %a, <4 x double> %b) {
; SSE2-SSSE3-NEXT: # kill: %AL<def> %AL<kill> %EAX<kill>
; SSE2-SSSE3-NEXT: retq
;
; AVX1-LABEL: v4f64:
; AVX1: # BB#0:
; AVX1-NEXT: vcmpltpd %ymm0, %ymm1, %ymm0
; AVX1-NEXT: vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT: vpacksswb %xmm1, %xmm0, %xmm0
; AVX1-NEXT: vmovmskps %xmm0, %eax
; AVX1-NEXT: # kill: %AL<def> %AL<kill> %EAX<kill>
; AVX1-NEXT: vzeroupper
; AVX1-NEXT: retq
;
; AVX2-LABEL: v4f64:
; AVX2: # BB#0:
; AVX2-NEXT: vcmpltpd %ymm0, %ymm1, %ymm0
; AVX2-NEXT: vmovmskpd %ymm0, %eax
; AVX2-NEXT: # kill: %AL<def> %AL<kill> %EAX<kill>
; AVX2-NEXT: vzeroupper
; AVX2-NEXT: retq
; AVX12-LABEL: v4f64:
; AVX12: # BB#0:
; AVX12-NEXT: vcmpltpd %ymm0, %ymm1, %ymm0
; AVX12-NEXT: vmovmskpd %ymm0, %eax
; AVX12-NEXT: # kill: %AL<def> %AL<kill> %EAX<kill>
; AVX12-NEXT: vzeroupper
; AVX12-NEXT: retq
;
; AVX512-LABEL: v4f64:
; AVX512: # BB#0: