[llvm] 38ab47c - [DAG] Match USUBSAT patterns through zext/trunc
Simon Pilgrim via llvm-commits
llvm-commits at lists.llvm.org
Sun Feb 21 07:47:06 PST 2021
Author: Simon Pilgrim
Date: 2021-02-21T15:26:54Z
New Revision: 38ab47c8136d4f928cfd4babb89cb3c7ab783437
URL: https://github.com/llvm/llvm-project/commit/38ab47c8136d4f928cfd4babb89cb3c7ab783437
DIFF: https://github.com/llvm/llvm-project/commit/38ab47c8136d4f928cfd4babb89cb3c7ab783437.diff
LOG: [DAG] Match USUBSAT patterns through zext/trunc
This patch handles usubsat patterns hidden through zext/trunc and uses the getTruncatedUSUBSAT helper to determine if the USUBSAT can be correctly performed in the truncated form:
zext(x) >= y ? x - trunc(y) : 0 --> usubsat(x,trunc(umin(y,SatLimit)))
zext(x) > y ? x - trunc(y) : 0 --> usubsat(x,trunc(umin(y,SatLimit)))
Based on original examples:
void foo(unsigned short *p, int max, int n) {
int i;
unsigned m;
for (i = 0; i < n; i++) {
m = *--p;
*p = (unsigned short)(m >= max ? m-max : 0);
}
}
Differential Revision: https://reviews.llvm.org/D25987
Added:
Modified:
llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
llvm/test/CodeGen/X86/psubus.ll
Removed:
################################################################################
diff --git a/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp b/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
index f42cdde860ec..2b4ce5ff7ec7 100644
--- a/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
+++ b/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
@@ -9881,6 +9881,19 @@ SDValue DAGCombiner::visitVSELECT(SDNode *N) {
SDValue CondRHS = RHS;
SDValue OpLHS = Other.getOperand(0), OpRHS = Other.getOperand(1);
+ if (Other.getOpcode() == ISD::SUB &&
+ LHS.getOpcode() == ISD::ZERO_EXTEND && LHS.getOperand(0) == OpLHS &&
+ OpRHS.getOpcode() == ISD::TRUNCATE && OpRHS.getOperand(0) == RHS) {
+ // Look for a general sub with unsigned saturation first.
+ // zext(x) >= y ? x - trunc(y) : 0
+ // --> usubsat(x,trunc(umin(y,SatLimit)))
+ // zext(x) > y ? x - trunc(y) : 0
+ // --> usubsat(x,trunc(umin(y,SatLimit)))
+ if (SatCC == ISD::SETUGE || SatCC == ISD::SETUGT)
+ return getTruncatedUSUBSAT(VT, LHS.getValueType(), LHS, RHS, DAG,
+ DL);
+ }
+
if (OpLHS == LHS) {
// Look for a general sub with unsigned saturation first.
// x >= y ? x-y : 0 --> usubsat x, y
diff --git a/llvm/test/CodeGen/X86/psubus.ll b/llvm/test/CodeGen/X86/psubus.ll
index 7c23305d6724..ea3a89fdef4c 100644
--- a/llvm/test/CodeGen/X86/psubus.ll
+++ b/llvm/test/CodeGen/X86/psubus.ll
@@ -465,130 +465,90 @@ vector.ph:
ret <32 x i8> %res
}
-; FIXME: match this to UMIN+TRUNC+PSUBUS
define <8 x i16> @test13(<8 x i16> %x, <8 x i32> %y) nounwind {
; SSE2-LABEL: test13:
; SSE2: # %bb.0: # %vector.ph
-; SSE2-NEXT: pxor %xmm3, %xmm3
-; SSE2-NEXT: movdqa %xmm0, %xmm4
-; SSE2-NEXT: punpcklwd {{.*#+}} xmm4 = xmm4[0],xmm3[0],xmm4[1],xmm3[1],xmm4[2],xmm3[2],xmm4[3],xmm3[3]
-; SSE2-NEXT: movdqa %xmm0, %xmm5
-; SSE2-NEXT: punpckhwd {{.*#+}} xmm5 = xmm5[4],xmm3[4],xmm5[5],xmm3[5],xmm5[6],xmm3[6],xmm5[7],xmm3[7]
-; SSE2-NEXT: movdqa {{.*#+}} xmm6 = [2147483648,2147483648,2147483648,2147483648]
-; SSE2-NEXT: movdqa %xmm2, %xmm7
-; SSE2-NEXT: pxor %xmm6, %xmm7
-; SSE2-NEXT: por %xmm6, %xmm5
-; SSE2-NEXT: pcmpgtd %xmm5, %xmm7
-; SSE2-NEXT: movdqa %xmm1, %xmm3
-; SSE2-NEXT: pxor %xmm6, %xmm3
-; SSE2-NEXT: por %xmm6, %xmm4
-; SSE2-NEXT: pcmpgtd %xmm4, %xmm3
-; SSE2-NEXT: packssdw %xmm7, %xmm3
-; SSE2-NEXT: pslld $16, %xmm2
-; SSE2-NEXT: psrad $16, %xmm2
-; SSE2-NEXT: pslld $16, %xmm1
-; SSE2-NEXT: psrad $16, %xmm1
-; SSE2-NEXT: packssdw %xmm2, %xmm1
-; SSE2-NEXT: psubw %xmm1, %xmm0
-; SSE2-NEXT: pandn %xmm0, %xmm3
-; SSE2-NEXT: movdqa %xmm3, %xmm0
+; SSE2-NEXT: movdqa {{.*#+}} xmm3 = [2147483648,2147483648,2147483648,2147483648]
+; SSE2-NEXT: movdqa %xmm2, %xmm4
+; SSE2-NEXT: pxor %xmm3, %xmm4
+; SSE2-NEXT: movdqa {{.*#+}} xmm5 = [2147549183,2147549183,2147549183,2147549183]
+; SSE2-NEXT: movdqa %xmm5, %xmm6
+; SSE2-NEXT: pcmpgtd %xmm4, %xmm6
+; SSE2-NEXT: pcmpeqd %xmm4, %xmm4
+; SSE2-NEXT: pand %xmm6, %xmm2
+; SSE2-NEXT: pxor %xmm4, %xmm6
+; SSE2-NEXT: por %xmm2, %xmm6
+; SSE2-NEXT: pslld $16, %xmm6
+; SSE2-NEXT: psrad $16, %xmm6
+; SSE2-NEXT: pxor %xmm1, %xmm3
+; SSE2-NEXT: pcmpgtd %xmm3, %xmm5
+; SSE2-NEXT: pxor %xmm5, %xmm4
+; SSE2-NEXT: pand %xmm1, %xmm5
+; SSE2-NEXT: por %xmm4, %xmm5
+; SSE2-NEXT: pslld $16, %xmm5
+; SSE2-NEXT: psrad $16, %xmm5
+; SSE2-NEXT: packssdw %xmm6, %xmm5
+; SSE2-NEXT: psubusw %xmm5, %xmm0
; SSE2-NEXT: retq
;
; SSSE3-LABEL: test13:
; SSSE3: # %bb.0: # %vector.ph
-; SSSE3-NEXT: pxor %xmm3, %xmm3
-; SSSE3-NEXT: movdqa %xmm0, %xmm4
-; SSSE3-NEXT: punpcklwd {{.*#+}} xmm4 = xmm4[0],xmm3[0],xmm4[1],xmm3[1],xmm4[2],xmm3[2],xmm4[3],xmm3[3]
-; SSSE3-NEXT: movdqa %xmm0, %xmm5
-; SSSE3-NEXT: punpckhwd {{.*#+}} xmm5 = xmm5[4],xmm3[4],xmm5[5],xmm3[5],xmm5[6],xmm3[6],xmm5[7],xmm3[7]
-; SSSE3-NEXT: movdqa {{.*#+}} xmm6 = [2147483648,2147483648,2147483648,2147483648]
-; SSSE3-NEXT: movdqa %xmm2, %xmm7
-; SSSE3-NEXT: pxor %xmm6, %xmm7
-; SSSE3-NEXT: por %xmm6, %xmm5
-; SSSE3-NEXT: pcmpgtd %xmm5, %xmm7
-; SSSE3-NEXT: movdqa %xmm1, %xmm3
-; SSSE3-NEXT: pxor %xmm6, %xmm3
-; SSSE3-NEXT: por %xmm6, %xmm4
-; SSSE3-NEXT: pcmpgtd %xmm4, %xmm3
-; SSSE3-NEXT: packssdw %xmm7, %xmm3
-; SSSE3-NEXT: movdqa {{.*#+}} xmm4 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
-; SSSE3-NEXT: pshufb %xmm4, %xmm2
-; SSSE3-NEXT: pshufb %xmm4, %xmm1
-; SSSE3-NEXT: punpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm2[0]
-; SSSE3-NEXT: psubw %xmm1, %xmm0
-; SSSE3-NEXT: pandn %xmm0, %xmm3
-; SSSE3-NEXT: movdqa %xmm3, %xmm0
+; SSSE3-NEXT: movdqa {{.*#+}} xmm3 = [2147483648,2147483648,2147483648,2147483648]
+; SSSE3-NEXT: movdqa %xmm2, %xmm4
+; SSSE3-NEXT: pxor %xmm3, %xmm4
+; SSSE3-NEXT: movdqa {{.*#+}} xmm5 = [2147549183,2147549183,2147549183,2147549183]
+; SSSE3-NEXT: movdqa %xmm5, %xmm6
+; SSSE3-NEXT: pcmpgtd %xmm4, %xmm6
+; SSSE3-NEXT: movdqa {{.*#+}} xmm4 = [65535,65535,65535,65535]
+; SSSE3-NEXT: pand %xmm6, %xmm2
+; SSSE3-NEXT: pandn %xmm4, %xmm6
+; SSSE3-NEXT: por %xmm2, %xmm6
+; SSSE3-NEXT: movdqa {{.*#+}} xmm2 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
+; SSSE3-NEXT: pshufb %xmm2, %xmm6
+; SSSE3-NEXT: pxor %xmm1, %xmm3
+; SSSE3-NEXT: pcmpgtd %xmm3, %xmm5
+; SSSE3-NEXT: pand %xmm5, %xmm1
+; SSSE3-NEXT: pandn %xmm4, %xmm5
+; SSSE3-NEXT: por %xmm1, %xmm5
+; SSSE3-NEXT: pshufb %xmm2, %xmm5
+; SSSE3-NEXT: punpcklqdq {{.*#+}} xmm5 = xmm5[0],xmm6[0]
+; SSSE3-NEXT: psubusw %xmm5, %xmm0
; SSSE3-NEXT: retq
;
; SSE41-LABEL: test13:
; SSE41: # %bb.0: # %vector.ph
-; SSE41-NEXT: pxor %xmm4, %xmm4
-; SSE41-NEXT: movdqa %xmm0, %xmm3
-; SSE41-NEXT: punpckhwd {{.*#+}} xmm3 = xmm3[4],xmm4[4],xmm3[5],xmm4[5],xmm3[6],xmm4[6],xmm3[7],xmm4[7]
-; SSE41-NEXT: pmovzxwd {{.*#+}} xmm5 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
-; SSE41-NEXT: movdqa %xmm3, %xmm6
-; SSE41-NEXT: pmaxud %xmm2, %xmm6
-; SSE41-NEXT: pcmpeqd %xmm3, %xmm6
-; SSE41-NEXT: pcmpeqd %xmm7, %xmm7
-; SSE41-NEXT: pxor %xmm7, %xmm6
-; SSE41-NEXT: movdqa %xmm5, %xmm3
-; SSE41-NEXT: pmaxud %xmm1, %xmm3
-; SSE41-NEXT: pcmpeqd %xmm5, %xmm3
-; SSE41-NEXT: pxor %xmm7, %xmm3
-; SSE41-NEXT: packssdw %xmm6, %xmm3
-; SSE41-NEXT: pblendw {{.*#+}} xmm2 = xmm2[0],xmm4[1],xmm2[2],xmm4[3],xmm2[4],xmm4[5],xmm2[6],xmm4[7]
-; SSE41-NEXT: pblendw {{.*#+}} xmm1 = xmm1[0],xmm4[1],xmm1[2],xmm4[3],xmm1[4],xmm4[5],xmm1[6],xmm4[7]
+; SSE41-NEXT: movdqa {{.*#+}} xmm3 = [65535,65535,65535,65535]
+; SSE41-NEXT: pminud %xmm3, %xmm2
+; SSE41-NEXT: pminud %xmm3, %xmm1
; SSE41-NEXT: packusdw %xmm2, %xmm1
-; SSE41-NEXT: psubw %xmm1, %xmm0
-; SSE41-NEXT: pandn %xmm0, %xmm3
-; SSE41-NEXT: movdqa %xmm3, %xmm0
+; SSE41-NEXT: psubusw %xmm1, %xmm0
; SSE41-NEXT: retq
;
; AVX1-LABEL: test13:
; AVX1: # %bb.0: # %vector.ph
-; AVX1-NEXT: vpxor %xmm2, %xmm2, %xmm2
-; AVX1-NEXT: vpunpckhwd {{.*#+}} xmm2 = xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
-; AVX1-NEXT: vpmovzxwd {{.*#+}} xmm3 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
-; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm4
-; AVX1-NEXT: vpmaxud %xmm4, %xmm2, %xmm5
-; AVX1-NEXT: vpcmpeqd %xmm5, %xmm2, %xmm2
-; AVX1-NEXT: vpcmpeqd %xmm5, %xmm5, %xmm5
-; AVX1-NEXT: vpxor %xmm5, %xmm2, %xmm2
-; AVX1-NEXT: vpmaxud %xmm1, %xmm3, %xmm6
-; AVX1-NEXT: vpcmpeqd %xmm6, %xmm3, %xmm3
-; AVX1-NEXT: vpxor %xmm5, %xmm3, %xmm3
-; AVX1-NEXT: vpackssdw %xmm2, %xmm3, %xmm2
-; AVX1-NEXT: vmovdqa {{.*#+}} xmm3 = <0,1,4,5,8,9,12,13,u,u,u,u,u,u,u,u>
-; AVX1-NEXT: vpshufb %xmm3, %xmm4, %xmm4
-; AVX1-NEXT: vpshufb %xmm3, %xmm1, %xmm1
-; AVX1-NEXT: vpunpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm4[0]
-; AVX1-NEXT: vpsubw %xmm1, %xmm0, %xmm0
-; AVX1-NEXT: vpandn %xmm0, %xmm2, %xmm0
+; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm2
+; AVX1-NEXT: vmovdqa {{.*#+}} xmm3 = [65535,65535,65535,65535]
+; AVX1-NEXT: vpminud %xmm3, %xmm2, %xmm2
+; AVX1-NEXT: vpminud %xmm3, %xmm1, %xmm1
+; AVX1-NEXT: vpackusdw %xmm2, %xmm1, %xmm1
+; AVX1-NEXT: vpsubusw %xmm1, %xmm0, %xmm0
; AVX1-NEXT: vzeroupper
; AVX1-NEXT: retq
;
; AVX2-LABEL: test13:
; AVX2: # %bb.0: # %vector.ph
-; AVX2-NEXT: vpmovzxwd {{.*#+}} ymm2 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
-; AVX2-NEXT: vpmaxud %ymm1, %ymm2, %ymm3
-; AVX2-NEXT: vpcmpeqd %ymm3, %ymm2, %ymm2
-; AVX2-NEXT: vpcmpeqd %ymm3, %ymm3, %ymm3
-; AVX2-NEXT: vpxor %ymm3, %ymm2, %ymm2
-; AVX2-NEXT: vextracti128 $1, %ymm2, %xmm3
-; AVX2-NEXT: vpackssdw %xmm3, %xmm2, %xmm2
-; AVX2-NEXT: vpshufb {{.*#+}} ymm1 = ymm1[0,1,4,5,8,9,12,13,u,u,u,u,u,u,u,u,16,17,20,21,24,25,28,29,u,u,u,u,u,u,u,u]
-; AVX2-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
-; AVX2-NEXT: vpsubw %xmm1, %xmm0, %xmm0
-; AVX2-NEXT: vpandn %xmm0, %xmm2, %xmm0
+; AVX2-NEXT: vpbroadcastd {{.*#+}} ymm2 = [65535,65535,65535,65535,65535,65535,65535,65535]
+; AVX2-NEXT: vpminud %ymm2, %ymm1, %ymm1
+; AVX2-NEXT: vextracti128 $1, %ymm1, %xmm2
+; AVX2-NEXT: vpackusdw %xmm2, %xmm1, %xmm1
+; AVX2-NEXT: vpsubusw %xmm1, %xmm0, %xmm0
; AVX2-NEXT: vzeroupper
; AVX2-NEXT: retq
;
; AVX512-LABEL: test13:
; AVX512: # %bb.0: # %vector.ph
-; AVX512-NEXT: vpmovzxwd {{.*#+}} ymm2 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
-; AVX512-NEXT: vpcmpnltud %ymm1, %ymm2, %k1
-; AVX512-NEXT: vpmovdw %ymm1, %xmm1
-; AVX512-NEXT: vpsubw %xmm1, %xmm0, %xmm0 {%k1} {z}
+; AVX512-NEXT: vpmovusdw %ymm1, %xmm1
+; AVX512-NEXT: vpsubusw %xmm1, %xmm0, %xmm0
; AVX512-NEXT: vzeroupper
; AVX512-NEXT: retq
vector.ph:
@@ -773,127 +733,90 @@ vector.ph:
ret <16 x i8> %res
}
-; FIXME: match this to UMIN+TRUNC+PSUBUS
define <8 x i16> @test15(<8 x i16> %x, <8 x i32> %y) nounwind {
; SSE2-LABEL: test15:
; SSE2: # %bb.0: # %vector.ph
-; SSE2-NEXT: pxor %xmm3, %xmm3
-; SSE2-NEXT: movdqa %xmm0, %xmm4
-; SSE2-NEXT: punpcklwd {{.*#+}} xmm4 = xmm4[0],xmm3[0],xmm4[1],xmm3[1],xmm4[2],xmm3[2],xmm4[3],xmm3[3]
-; SSE2-NEXT: movdqa %xmm0, %xmm5
-; SSE2-NEXT: punpckhwd {{.*#+}} xmm5 = xmm5[4],xmm3[4],xmm5[5],xmm3[5],xmm5[6],xmm3[6],xmm5[7],xmm3[7]
; SSE2-NEXT: movdqa {{.*#+}} xmm3 = [2147483648,2147483648,2147483648,2147483648]
-; SSE2-NEXT: movdqa %xmm2, %xmm6
-; SSE2-NEXT: pxor %xmm3, %xmm6
-; SSE2-NEXT: por %xmm3, %xmm5
-; SSE2-NEXT: pcmpgtd %xmm6, %xmm5
-; SSE2-NEXT: movdqa %xmm1, %xmm6
-; SSE2-NEXT: pxor %xmm3, %xmm6
-; SSE2-NEXT: por %xmm3, %xmm4
-; SSE2-NEXT: pcmpgtd %xmm6, %xmm4
-; SSE2-NEXT: packssdw %xmm5, %xmm4
-; SSE2-NEXT: pslld $16, %xmm2
-; SSE2-NEXT: psrad $16, %xmm2
-; SSE2-NEXT: pslld $16, %xmm1
-; SSE2-NEXT: psrad $16, %xmm1
-; SSE2-NEXT: packssdw %xmm2, %xmm1
-; SSE2-NEXT: psubw %xmm1, %xmm0
-; SSE2-NEXT: pand %xmm4, %xmm0
+; SSE2-NEXT: movdqa %xmm2, %xmm4
+; SSE2-NEXT: pxor %xmm3, %xmm4
+; SSE2-NEXT: movdqa {{.*#+}} xmm5 = [2147549183,2147549183,2147549183,2147549183]
+; SSE2-NEXT: movdqa %xmm5, %xmm6
+; SSE2-NEXT: pcmpgtd %xmm4, %xmm6
+; SSE2-NEXT: pcmpeqd %xmm4, %xmm4
+; SSE2-NEXT: pand %xmm6, %xmm2
+; SSE2-NEXT: pxor %xmm4, %xmm6
+; SSE2-NEXT: por %xmm2, %xmm6
+; SSE2-NEXT: pslld $16, %xmm6
+; SSE2-NEXT: psrad $16, %xmm6
+; SSE2-NEXT: pxor %xmm1, %xmm3
+; SSE2-NEXT: pcmpgtd %xmm3, %xmm5
+; SSE2-NEXT: pxor %xmm5, %xmm4
+; SSE2-NEXT: pand %xmm1, %xmm5
+; SSE2-NEXT: por %xmm4, %xmm5
+; SSE2-NEXT: pslld $16, %xmm5
+; SSE2-NEXT: psrad $16, %xmm5
+; SSE2-NEXT: packssdw %xmm6, %xmm5
+; SSE2-NEXT: psubusw %xmm5, %xmm0
; SSE2-NEXT: retq
;
; SSSE3-LABEL: test15:
; SSSE3: # %bb.0: # %vector.ph
-; SSSE3-NEXT: pxor %xmm3, %xmm3
-; SSSE3-NEXT: movdqa %xmm0, %xmm4
-; SSSE3-NEXT: punpcklwd {{.*#+}} xmm4 = xmm4[0],xmm3[0],xmm4[1],xmm3[1],xmm4[2],xmm3[2],xmm4[3],xmm3[3]
-; SSSE3-NEXT: movdqa %xmm0, %xmm5
-; SSSE3-NEXT: punpckhwd {{.*#+}} xmm5 = xmm5[4],xmm3[4],xmm5[5],xmm3[5],xmm5[6],xmm3[6],xmm5[7],xmm3[7]
; SSSE3-NEXT: movdqa {{.*#+}} xmm3 = [2147483648,2147483648,2147483648,2147483648]
-; SSSE3-NEXT: movdqa %xmm2, %xmm6
-; SSSE3-NEXT: pxor %xmm3, %xmm6
-; SSSE3-NEXT: por %xmm3, %xmm5
-; SSSE3-NEXT: pcmpgtd %xmm6, %xmm5
-; SSSE3-NEXT: movdqa %xmm1, %xmm6
-; SSSE3-NEXT: pxor %xmm3, %xmm6
-; SSSE3-NEXT: por %xmm3, %xmm4
-; SSSE3-NEXT: pcmpgtd %xmm6, %xmm4
-; SSSE3-NEXT: packssdw %xmm5, %xmm4
-; SSSE3-NEXT: movdqa {{.*#+}} xmm3 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
-; SSSE3-NEXT: pshufb %xmm3, %xmm2
-; SSSE3-NEXT: pshufb %xmm3, %xmm1
-; SSSE3-NEXT: punpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm2[0]
-; SSSE3-NEXT: psubw %xmm1, %xmm0
-; SSSE3-NEXT: pand %xmm4, %xmm0
+; SSSE3-NEXT: movdqa %xmm2, %xmm4
+; SSSE3-NEXT: pxor %xmm3, %xmm4
+; SSSE3-NEXT: movdqa {{.*#+}} xmm5 = [2147549183,2147549183,2147549183,2147549183]
+; SSSE3-NEXT: movdqa %xmm5, %xmm6
+; SSSE3-NEXT: pcmpgtd %xmm4, %xmm6
+; SSSE3-NEXT: movdqa {{.*#+}} xmm4 = [65535,65535,65535,65535]
+; SSSE3-NEXT: pand %xmm6, %xmm2
+; SSSE3-NEXT: pandn %xmm4, %xmm6
+; SSSE3-NEXT: por %xmm2, %xmm6
+; SSSE3-NEXT: movdqa {{.*#+}} xmm2 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
+; SSSE3-NEXT: pshufb %xmm2, %xmm6
+; SSSE3-NEXT: pxor %xmm1, %xmm3
+; SSSE3-NEXT: pcmpgtd %xmm3, %xmm5
+; SSSE3-NEXT: pand %xmm5, %xmm1
+; SSSE3-NEXT: pandn %xmm4, %xmm5
+; SSSE3-NEXT: por %xmm1, %xmm5
+; SSSE3-NEXT: pshufb %xmm2, %xmm5
+; SSSE3-NEXT: punpcklqdq {{.*#+}} xmm5 = xmm5[0],xmm6[0]
+; SSSE3-NEXT: psubusw %xmm5, %xmm0
; SSSE3-NEXT: retq
;
; SSE41-LABEL: test15:
; SSE41: # %bb.0: # %vector.ph
-; SSE41-NEXT: pxor %xmm3, %xmm3
-; SSE41-NEXT: movdqa %xmm0, %xmm4
-; SSE41-NEXT: punpckhwd {{.*#+}} xmm4 = xmm4[4],xmm3[4],xmm4[5],xmm3[5],xmm4[6],xmm3[6],xmm4[7],xmm3[7]
-; SSE41-NEXT: pmovzxwd {{.*#+}} xmm5 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
-; SSE41-NEXT: movdqa %xmm4, %xmm6
-; SSE41-NEXT: pminud %xmm2, %xmm6
-; SSE41-NEXT: pcmpeqd %xmm4, %xmm6
-; SSE41-NEXT: pcmpeqd %xmm4, %xmm4
-; SSE41-NEXT: pxor %xmm4, %xmm6
-; SSE41-NEXT: movdqa %xmm5, %xmm7
-; SSE41-NEXT: pminud %xmm1, %xmm7
-; SSE41-NEXT: pcmpeqd %xmm5, %xmm7
-; SSE41-NEXT: pxor %xmm4, %xmm7
-; SSE41-NEXT: packssdw %xmm6, %xmm7
-; SSE41-NEXT: pblendw {{.*#+}} xmm2 = xmm2[0],xmm3[1],xmm2[2],xmm3[3],xmm2[4],xmm3[5],xmm2[6],xmm3[7]
-; SSE41-NEXT: pblendw {{.*#+}} xmm1 = xmm1[0],xmm3[1],xmm1[2],xmm3[3],xmm1[4],xmm3[5],xmm1[6],xmm3[7]
+; SSE41-NEXT: movdqa {{.*#+}} xmm3 = [65535,65535,65535,65535]
+; SSE41-NEXT: pminud %xmm3, %xmm2
+; SSE41-NEXT: pminud %xmm3, %xmm1
; SSE41-NEXT: packusdw %xmm2, %xmm1
-; SSE41-NEXT: psubw %xmm1, %xmm0
-; SSE41-NEXT: pand %xmm7, %xmm0
+; SSE41-NEXT: psubusw %xmm1, %xmm0
; SSE41-NEXT: retq
;
; AVX1-LABEL: test15:
; AVX1: # %bb.0: # %vector.ph
-; AVX1-NEXT: vpxor %xmm2, %xmm2, %xmm2
-; AVX1-NEXT: vpunpckhwd {{.*#+}} xmm2 = xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
-; AVX1-NEXT: vpmovzxwd {{.*#+}} xmm3 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
-; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm4
-; AVX1-NEXT: vpminud %xmm4, %xmm2, %xmm5
-; AVX1-NEXT: vpcmpeqd %xmm5, %xmm2, %xmm2
-; AVX1-NEXT: vpcmpeqd %xmm5, %xmm5, %xmm5
-; AVX1-NEXT: vpxor %xmm5, %xmm2, %xmm2
-; AVX1-NEXT: vpminud %xmm1, %xmm3, %xmm6
-; AVX1-NEXT: vpcmpeqd %xmm6, %xmm3, %xmm3
-; AVX1-NEXT: vpxor %xmm5, %xmm3, %xmm3
-; AVX1-NEXT: vpackssdw %xmm2, %xmm3, %xmm2
-; AVX1-NEXT: vmovdqa {{.*#+}} xmm3 = <0,1,4,5,8,9,12,13,u,u,u,u,u,u,u,u>
-; AVX1-NEXT: vpshufb %xmm3, %xmm4, %xmm4
-; AVX1-NEXT: vpshufb %xmm3, %xmm1, %xmm1
-; AVX1-NEXT: vpunpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm4[0]
-; AVX1-NEXT: vpsubw %xmm1, %xmm0, %xmm0
-; AVX1-NEXT: vpand %xmm0, %xmm2, %xmm0
+; AVX1-NEXT: vextractf128 $1, %ymm1, %xmm2
+; AVX1-NEXT: vmovdqa {{.*#+}} xmm3 = [65535,65535,65535,65535]
+; AVX1-NEXT: vpminud %xmm3, %xmm2, %xmm2
+; AVX1-NEXT: vpminud %xmm3, %xmm1, %xmm1
+; AVX1-NEXT: vpackusdw %xmm2, %xmm1, %xmm1
+; AVX1-NEXT: vpsubusw %xmm1, %xmm0, %xmm0
; AVX1-NEXT: vzeroupper
; AVX1-NEXT: retq
;
; AVX2-LABEL: test15:
; AVX2: # %bb.0: # %vector.ph
-; AVX2-NEXT: vpmovzxwd {{.*#+}} ymm2 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
-; AVX2-NEXT: vpminud %ymm1, %ymm2, %ymm3
-; AVX2-NEXT: vpcmpeqd %ymm3, %ymm2, %ymm2
-; AVX2-NEXT: vpcmpeqd %ymm3, %ymm3, %ymm3
-; AVX2-NEXT: vpxor %ymm3, %ymm2, %ymm2
-; AVX2-NEXT: vextracti128 $1, %ymm2, %xmm3
-; AVX2-NEXT: vpackssdw %xmm3, %xmm2, %xmm2
-; AVX2-NEXT: vpshufb {{.*#+}} ymm1 = ymm1[0,1,4,5,8,9,12,13,u,u,u,u,u,u,u,u,16,17,20,21,24,25,28,29,u,u,u,u,u,u,u,u]
-; AVX2-NEXT: vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
-; AVX2-NEXT: vpsubw %xmm1, %xmm0, %xmm0
-; AVX2-NEXT: vpand %xmm0, %xmm2, %xmm0
+; AVX2-NEXT: vpbroadcastd {{.*#+}} ymm2 = [65535,65535,65535,65535,65535,65535,65535,65535]
+; AVX2-NEXT: vpminud %ymm2, %ymm1, %ymm1
+; AVX2-NEXT: vextracti128 $1, %ymm1, %xmm2
+; AVX2-NEXT: vpackusdw %xmm2, %xmm1, %xmm1
+; AVX2-NEXT: vpsubusw %xmm1, %xmm0, %xmm0
; AVX2-NEXT: vzeroupper
; AVX2-NEXT: retq
;
; AVX512-LABEL: test15:
; AVX512: # %bb.0: # %vector.ph
-; AVX512-NEXT: vpmovzxwd {{.*#+}} ymm2 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
-; AVX512-NEXT: vpcmpnleud %ymm1, %ymm2, %k1
-; AVX512-NEXT: vpmovdw %ymm1, %xmm1
-; AVX512-NEXT: vpsubw %xmm1, %xmm0, %xmm0 {%k1} {z}
+; AVX512-NEXT: vpmovusdw %ymm1, %xmm1
+; AVX512-NEXT: vpsubusw %xmm1, %xmm0, %xmm0
; AVX512-NEXT: vzeroupper
; AVX512-NEXT: retq
vector.ph:
More information about the llvm-commits
mailing list