[llvm] 24184db - [X86] Fold CONCAT(VPERMV3(X,Y,M0),VPERMV3(Z,W,M1)) -> VPERMV3(CONCAT(X,Z),CONCAT(Y,W),CONCAT(M0,M1))
Simon Pilgrim via llvm-commits
llvm-commits at lists.llvm.org
Wed Dec 9 06:34:56 PST 2020
Author: Simon Pilgrim
Date: 2020-12-09T14:29:32Z
New Revision: 24184dbb82f9eac76c4ffe077fb957a526b1b963
URL: https://github.com/llvm/llvm-project/commit/24184dbb82f9eac76c4ffe077fb957a526b1b963
DIFF: https://github.com/llvm/llvm-project/commit/24184dbb82f9eac76c4ffe077fb957a526b1b963.diff
LOG: [X86] Fold CONCAT(VPERMV3(X,Y,M0),VPERMV3(Z,W,M1)) -> VPERMV3(CONCAT(X,Z),CONCAT(Y,W),CONCAT(M0,M1))
Further prep work toward supporting different subvector sizes in combineX86ShufflesRecursively
Added:
Modified:
llvm/lib/Target/X86/X86ISelLowering.cpp
llvm/test/CodeGen/X86/vector-pack-512.ll
Removed:
################################################################################
diff --git a/llvm/lib/Target/X86/X86ISelLowering.cpp b/llvm/lib/Target/X86/X86ISelLowering.cpp
index 135b13b34f33..53f30af8d38b 100644
--- a/llvm/lib/Target/X86/X86ISelLowering.cpp
+++ b/llvm/lib/Target/X86/X86ISelLowering.cpp
@@ -48813,6 +48813,38 @@ static SDValue combineConcatVectorOps(const SDLoc &DL, MVT VT,
return DAG.getBitcast(VT, Res);
}
break;
+ case X86ISD::VPERMV3:
+ if (!IsSplat && NumOps == 2 && VT.is512BitVector()) {
+ MVT OpVT = Op0.getSimpleValueType();
+ int NumSrcElts = OpVT.getVectorNumElements();
+ SmallVector<int, 64> ConcatMask;
+ for (unsigned i = 0; i != NumOps; ++i) {
+ bool IsUnary;
+ SmallVector<int, 64> SubMask;
+ SmallVector<SDValue, 2> SubOps;
+ if (!getTargetShuffleMask(Ops[i].getNode(), OpVT, false, SubOps,
+ SubMask, IsUnary))
+ break;
+ for (int M : SubMask) {
+ if (0 <= M) {
+ M += M < NumSrcElts ? 0 : NumSrcElts;
+ M += i * NumSrcElts;
+ }
+ ConcatMask.push_back(M);
+ }
+ }
+ if (ConcatMask.size() == (NumOps * NumSrcElts)) {
+ SDValue Src0 = concatSubVectors(Ops[0].getOperand(0),
+ Ops[1].getOperand(0), DAG, DL);
+ SDValue Src1 = concatSubVectors(Ops[0].getOperand(2),
+ Ops[1].getOperand(2), DAG, DL);
+ MVT IntMaskSVT = MVT::getIntegerVT(VT.getScalarSizeInBits());
+ MVT IntMaskVT = MVT::getVectorVT(IntMaskSVT, NumOps * NumSrcElts);
+ SDValue Mask = getConstVector(ConcatMask, IntMaskVT, DAG, DL, true);
+ return DAG.getNode(X86ISD::VPERMV3, DL, VT, Src0, Mask, Src1);
+ }
+ }
+ break;
case X86ISD::VSHLI:
case X86ISD::VSRAI:
case X86ISD::VSRLI:
diff --git a/llvm/test/CodeGen/X86/vector-pack-512.ll b/llvm/test/CodeGen/X86/vector-pack-512.ll
index 0612e08e830a..432927d51061 100644
--- a/llvm/test/CodeGen/X86/vector-pack-512.ll
+++ b/llvm/test/CodeGen/X86/vector-pack-512.ll
@@ -145,11 +145,10 @@ define <32 x i16> @concat_trunc_packssdw_512(<16 x i32> %a0, <16 x i32> %a1) nou
; AVX512-NEXT: vpsrad $23, %zmm1, %zmm1
; AVX512-NEXT: vpmovdw %zmm0, %ymm0
; AVX512-NEXT: vpmovdw %zmm1, %ymm1
-; AVX512-NEXT: vmovdqa {{.*#+}} ymm2 = [2,6,3,7]
-; AVX512-NEXT: vpermi2q %ymm1, %ymm0, %ymm2
-; AVX512-NEXT: vmovdqa {{.*#+}} ymm3 = [0,4,1,5]
-; AVX512-NEXT: vpermi2q %ymm1, %ymm0, %ymm3
-; AVX512-NEXT: vinserti64x4 $1, %ymm2, %zmm3, %zmm0
+; AVX512-NEXT: vinserti64x4 $1, %ymm1, %zmm1, %zmm1
+; AVX512-NEXT: vinserti64x4 $1, %ymm0, %zmm0, %zmm2
+; AVX512-NEXT: vmovdqa64 {{.*#+}} zmm0 = [0,8,1,9,6,14,7,15]
+; AVX512-NEXT: vpermi2q %zmm1, %zmm2, %zmm0
; AVX512-NEXT: retq
%1 = ashr <16 x i32> %a0, <i32 17, i32 17, i32 17, i32 17, i32 17, i32 17, i32 17, i32 17, i32 17, i32 17, i32 17, i32 17, i32 17, i32 17, i32 17, i32 17>
%2 = ashr <16 x i32> %a1, <i32 23, i32 23, i32 23, i32 23, i32 23, i32 23, i32 23, i32 23, i32 23, i32 23, i32 23, i32 23, i32 23, i32 23, i32 23, i32 23>
@@ -166,11 +165,10 @@ define <32 x i16> @concat_trunc_packusdw_512(<16 x i32> %a0, <16 x i32> %a1) nou
; AVX512-NEXT: vpsrld $23, %zmm1, %zmm1
; AVX512-NEXT: vpmovdw %zmm0, %ymm0
; AVX512-NEXT: vpmovdw %zmm1, %ymm1
-; AVX512-NEXT: vmovdqa {{.*#+}} ymm2 = [2,6,3,7]
-; AVX512-NEXT: vpermi2q %ymm1, %ymm0, %ymm2
-; AVX512-NEXT: vmovdqa {{.*#+}} ymm3 = [0,4,1,5]
-; AVX512-NEXT: vpermi2q %ymm1, %ymm0, %ymm3
-; AVX512-NEXT: vinserti64x4 $1, %ymm2, %zmm3, %zmm0
+; AVX512-NEXT: vinserti64x4 $1, %ymm1, %zmm1, %zmm1
+; AVX512-NEXT: vinserti64x4 $1, %ymm0, %zmm0, %zmm2
+; AVX512-NEXT: vmovdqa64 {{.*#+}} zmm0 = [0,8,1,9,6,14,7,15]
+; AVX512-NEXT: vpermi2q %zmm1, %zmm2, %zmm0
; AVX512-NEXT: retq
%1 = lshr <16 x i32> %a0, <i32 17, i32 17, i32 17, i32 17, i32 17, i32 17, i32 17, i32 17, i32 17, i32 17, i32 17, i32 17, i32 17, i32 17, i32 17, i32 17>
%2 = lshr <16 x i32> %a1, <i32 23, i32 23, i32 23, i32 23, i32 23, i32 23, i32 23, i32 23, i32 23, i32 23, i32 23, i32 23, i32 23, i32 23, i32 23, i32 23>
@@ -183,26 +181,25 @@ define <32 x i16> @concat_trunc_packusdw_512(<16 x i32> %a0, <16 x i32> %a1) nou
define <64 x i8> @concat_trunc_packsswb_512(<32 x i16> %a0, <32 x i16> %a1) nounwind {
; AVX512F-LABEL: concat_trunc_packsswb_512:
; AVX512F: # %bb.0:
-; AVX512F-NEXT: vextracti64x4 $1, %zmm0, %ymm2
-; AVX512F-NEXT: vpsraw $15, %ymm2, %ymm2
+; AVX512F-NEXT: vpsraw $15, %ymm0, %ymm2
+; AVX512F-NEXT: vextracti64x4 $1, %zmm0, %ymm0
; AVX512F-NEXT: vpsraw $15, %ymm0, %ymm0
; AVX512F-NEXT: vpmovzxwd {{.*#+}} zmm0 = ymm0[0],zero,ymm0[1],zero,ymm0[2],zero,ymm0[3],zero,ymm0[4],zero,ymm0[5],zero,ymm0[6],zero,ymm0[7],zero,ymm0[8],zero,ymm0[9],zero,ymm0[10],zero,ymm0[11],zero,ymm0[12],zero,ymm0[13],zero,ymm0[14],zero,ymm0[15],zero
; AVX512F-NEXT: vpmovdb %zmm0, %xmm0
+; AVX512F-NEXT: vinserti128 $1, %xmm0, %ymm0, %ymm0
; AVX512F-NEXT: vpmovzxwd {{.*#+}} zmm2 = ymm2[0],zero,ymm2[1],zero,ymm2[2],zero,ymm2[3],zero,ymm2[4],zero,ymm2[5],zero,ymm2[6],zero,ymm2[7],zero,ymm2[8],zero,ymm2[9],zero,ymm2[10],zero,ymm2[11],zero,ymm2[12],zero,ymm2[13],zero,ymm2[14],zero,ymm2[15],zero
; AVX512F-NEXT: vpmovdb %zmm2, %xmm2
-; AVX512F-NEXT: vinserti128 $1, %xmm2, %ymm0, %ymm0
-; AVX512F-NEXT: vpmovzxwd {{.*#+}} zmm2 = ymm1[0],zero,ymm1[1],zero,ymm1[2],zero,ymm1[3],zero,ymm1[4],zero,ymm1[5],zero,ymm1[6],zero,ymm1[7],zero,ymm1[8],zero,ymm1[9],zero,ymm1[10],zero,ymm1[11],zero,ymm1[12],zero,ymm1[13],zero,ymm1[14],zero,ymm1[15],zero
-; AVX512F-NEXT: vpmovdb %zmm2, %xmm2
+; AVX512F-NEXT: vpmovzxwd {{.*#+}} zmm3 = ymm1[0],zero,ymm1[1],zero,ymm1[2],zero,ymm1[3],zero,ymm1[4],zero,ymm1[5],zero,ymm1[6],zero,ymm1[7],zero,ymm1[8],zero,ymm1[9],zero,ymm1[10],zero,ymm1[11],zero,ymm1[12],zero,ymm1[13],zero,ymm1[14],zero,ymm1[15],zero
+; AVX512F-NEXT: vpmovdb %zmm3, %xmm3
; AVX512F-NEXT: vextracti64x4 $1, %zmm1, %ymm1
; AVX512F-NEXT: vpmovzxwd {{.*#+}} zmm1 = ymm1[0],zero,ymm1[1],zero,ymm1[2],zero,ymm1[3],zero,ymm1[4],zero,ymm1[5],zero,ymm1[6],zero,ymm1[7],zero,ymm1[8],zero,ymm1[9],zero,ymm1[10],zero,ymm1[11],zero,ymm1[12],zero,ymm1[13],zero,ymm1[14],zero,ymm1[15],zero
; AVX512F-NEXT: vpmovdb %zmm1, %xmm1
-; AVX512F-NEXT: vinserti128 $1, %xmm1, %ymm2, %ymm1
+; AVX512F-NEXT: vinserti128 $1, %xmm1, %ymm3, %ymm1
; AVX512F-NEXT: vpand {{.*}}(%rip), %ymm1, %ymm1
-; AVX512F-NEXT: vmovdqa {{.*#+}} ymm2 = [2,6,3,7]
-; AVX512F-NEXT: vpermi2q %ymm1, %ymm0, %ymm2
-; AVX512F-NEXT: vmovdqa {{.*#+}} ymm3 = [0,4,1,5]
-; AVX512F-NEXT: vpermi2q %ymm1, %ymm0, %ymm3
-; AVX512F-NEXT: vinserti64x4 $1, %ymm2, %zmm3, %zmm0
+; AVX512F-NEXT: vinserti64x4 $1, %ymm0, %zmm2, %zmm2
+; AVX512F-NEXT: vinserti64x4 $1, %ymm1, %zmm1, %zmm1
+; AVX512F-NEXT: vmovdqa64 {{.*#+}} zmm0 = [0,8,1,9,6,14,7,15]
+; AVX512F-NEXT: vpermi2q %zmm1, %zmm2, %zmm0
; AVX512F-NEXT: retq
;
; AVX512BW-LABEL: concat_trunc_packsswb_512:
@@ -211,11 +208,10 @@ define <64 x i8> @concat_trunc_packsswb_512(<32 x i16> %a0, <32 x i16> %a1) noun
; AVX512BW-NEXT: vpmovwb %zmm0, %ymm0
; AVX512BW-NEXT: vpmovwb %zmm1, %ymm1
; AVX512BW-NEXT: vpand {{.*}}(%rip), %ymm1, %ymm1
-; AVX512BW-NEXT: vmovdqa {{.*#+}} ymm2 = [2,6,3,7]
-; AVX512BW-NEXT: vpermi2q %ymm1, %ymm0, %ymm2
-; AVX512BW-NEXT: vmovdqa {{.*#+}} ymm3 = [0,4,1,5]
-; AVX512BW-NEXT: vpermi2q %ymm1, %ymm0, %ymm3
-; AVX512BW-NEXT: vinserti64x4 $1, %ymm2, %zmm3, %zmm0
+; AVX512BW-NEXT: vinserti64x4 $1, %ymm1, %zmm1, %zmm1
+; AVX512BW-NEXT: vinserti64x4 $1, %ymm0, %zmm0, %zmm2
+; AVX512BW-NEXT: vmovdqa64 {{.*#+}} zmm0 = [0,8,1,9,6,14,7,15]
+; AVX512BW-NEXT: vpermi2q %zmm1, %zmm2, %zmm0
; AVX512BW-NEXT: retq
%1 = ashr <32 x i16> %a0, <i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15>
%2 = and <32 x i16> %a1, <i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1>
@@ -228,26 +224,25 @@ define <64 x i8> @concat_trunc_packsswb_512(<32 x i16> %a0, <32 x i16> %a1) noun
define <64 x i8> @concat_trunc_packuswb_512(<32 x i16> %a0, <32 x i16> %a1) nounwind {
; AVX512F-LABEL: concat_trunc_packuswb_512:
; AVX512F: # %bb.0:
-; AVX512F-NEXT: vextracti64x4 $1, %zmm0, %ymm2
-; AVX512F-NEXT: vpsrlw $15, %ymm2, %ymm2
+; AVX512F-NEXT: vpsrlw $15, %ymm0, %ymm2
+; AVX512F-NEXT: vextracti64x4 $1, %zmm0, %ymm0
; AVX512F-NEXT: vpsrlw $15, %ymm0, %ymm0
; AVX512F-NEXT: vpmovzxwd {{.*#+}} zmm0 = ymm0[0],zero,ymm0[1],zero,ymm0[2],zero,ymm0[3],zero,ymm0[4],zero,ymm0[5],zero,ymm0[6],zero,ymm0[7],zero,ymm0[8],zero,ymm0[9],zero,ymm0[10],zero,ymm0[11],zero,ymm0[12],zero,ymm0[13],zero,ymm0[14],zero,ymm0[15],zero
; AVX512F-NEXT: vpmovdb %zmm0, %xmm0
+; AVX512F-NEXT: vinserti128 $1, %xmm0, %ymm0, %ymm0
; AVX512F-NEXT: vpmovzxwd {{.*#+}} zmm2 = ymm2[0],zero,ymm2[1],zero,ymm2[2],zero,ymm2[3],zero,ymm2[4],zero,ymm2[5],zero,ymm2[6],zero,ymm2[7],zero,ymm2[8],zero,ymm2[9],zero,ymm2[10],zero,ymm2[11],zero,ymm2[12],zero,ymm2[13],zero,ymm2[14],zero,ymm2[15],zero
; AVX512F-NEXT: vpmovdb %zmm2, %xmm2
-; AVX512F-NEXT: vinserti128 $1, %xmm2, %ymm0, %ymm0
-; AVX512F-NEXT: vpmovzxwd {{.*#+}} zmm2 = ymm1[0],zero,ymm1[1],zero,ymm1[2],zero,ymm1[3],zero,ymm1[4],zero,ymm1[5],zero,ymm1[6],zero,ymm1[7],zero,ymm1[8],zero,ymm1[9],zero,ymm1[10],zero,ymm1[11],zero,ymm1[12],zero,ymm1[13],zero,ymm1[14],zero,ymm1[15],zero
-; AVX512F-NEXT: vpmovdb %zmm2, %xmm2
+; AVX512F-NEXT: vpmovzxwd {{.*#+}} zmm3 = ymm1[0],zero,ymm1[1],zero,ymm1[2],zero,ymm1[3],zero,ymm1[4],zero,ymm1[5],zero,ymm1[6],zero,ymm1[7],zero,ymm1[8],zero,ymm1[9],zero,ymm1[10],zero,ymm1[11],zero,ymm1[12],zero,ymm1[13],zero,ymm1[14],zero,ymm1[15],zero
+; AVX512F-NEXT: vpmovdb %zmm3, %xmm3
; AVX512F-NEXT: vextracti64x4 $1, %zmm1, %ymm1
; AVX512F-NEXT: vpmovzxwd {{.*#+}} zmm1 = ymm1[0],zero,ymm1[1],zero,ymm1[2],zero,ymm1[3],zero,ymm1[4],zero,ymm1[5],zero,ymm1[6],zero,ymm1[7],zero,ymm1[8],zero,ymm1[9],zero,ymm1[10],zero,ymm1[11],zero,ymm1[12],zero,ymm1[13],zero,ymm1[14],zero,ymm1[15],zero
; AVX512F-NEXT: vpmovdb %zmm1, %xmm1
-; AVX512F-NEXT: vinserti128 $1, %xmm1, %ymm2, %ymm1
+; AVX512F-NEXT: vinserti128 $1, %xmm1, %ymm3, %ymm1
; AVX512F-NEXT: vpand {{.*}}(%rip), %ymm1, %ymm1
-; AVX512F-NEXT: vmovdqa {{.*#+}} ymm2 = [2,6,3,7]
-; AVX512F-NEXT: vpermi2q %ymm1, %ymm0, %ymm2
-; AVX512F-NEXT: vmovdqa {{.*#+}} ymm3 = [0,4,1,5]
-; AVX512F-NEXT: vpermi2q %ymm1, %ymm0, %ymm3
-; AVX512F-NEXT: vinserti64x4 $1, %ymm2, %zmm3, %zmm0
+; AVX512F-NEXT: vinserti64x4 $1, %ymm0, %zmm2, %zmm2
+; AVX512F-NEXT: vinserti64x4 $1, %ymm1, %zmm1, %zmm1
+; AVX512F-NEXT: vmovdqa64 {{.*#+}} zmm0 = [0,8,1,9,6,14,7,15]
+; AVX512F-NEXT: vpermi2q %zmm1, %zmm2, %zmm0
; AVX512F-NEXT: retq
;
; AVX512BW-LABEL: concat_trunc_packuswb_512:
@@ -256,11 +251,10 @@ define <64 x i8> @concat_trunc_packuswb_512(<32 x i16> %a0, <32 x i16> %a1) noun
; AVX512BW-NEXT: vpmovwb %zmm0, %ymm0
; AVX512BW-NEXT: vpmovwb %zmm1, %ymm1
; AVX512BW-NEXT: vpand {{.*}}(%rip), %ymm1, %ymm1
-; AVX512BW-NEXT: vmovdqa {{.*#+}} ymm2 = [2,6,3,7]
-; AVX512BW-NEXT: vpermi2q %ymm1, %ymm0, %ymm2
-; AVX512BW-NEXT: vmovdqa {{.*#+}} ymm3 = [0,4,1,5]
-; AVX512BW-NEXT: vpermi2q %ymm1, %ymm0, %ymm3
-; AVX512BW-NEXT: vinserti64x4 $1, %ymm2, %zmm3, %zmm0
+; AVX512BW-NEXT: vinserti64x4 $1, %ymm1, %zmm1, %zmm1
+; AVX512BW-NEXT: vinserti64x4 $1, %ymm0, %zmm0, %zmm2
+; AVX512BW-NEXT: vmovdqa64 {{.*#+}} zmm0 = [0,8,1,9,6,14,7,15]
+; AVX512BW-NEXT: vpermi2q %zmm1, %zmm2, %zmm0
; AVX512BW-NEXT: retq
%1 = lshr <32 x i16> %a0, <i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15, i16 15>
%2 = and <32 x i16> %a1, <i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1>
More information about the llvm-commits
mailing list