[AVX-512][InstCombine] Teach InstCombine to optimize 512-bit packss/packus intrinsics like it does 128/256-bit.

llvm-svn: 295294

3 files changed, 132 insertions, 8 deletions

diff --git a/llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp b/llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp
index c26da03a89a..7f34a55638f 100644
--- a/llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstCombineCalls.cpp
@@ -2630,7 +2630,8 @@ Instruction *InstCombiner::visitCallInst(CallInst &CI) {
   case Intrinsic::x86_sse2_packsswb_128:
   case Intrinsic::x86_avx2_packssdw:
   case Intrinsic::x86_avx2_packsswb:
-  // TODO Add support for Intrinsic::x86_avx512_mask_packss*
+  case Intrinsic::x86_avx512_packssdw_512:
+  case Intrinsic::x86_avx512_packsswb_512:
     if (Value *V = simplifyX86pack(*II, *this, *Builder, true))
       return replaceInstUsesWith(*II, V);
     break;
@@ -2639,7 +2640,8 @@ Instruction *InstCombiner::visitCallInst(CallInst &CI) {
   case Intrinsic::x86_sse41_packusdw:
   case Intrinsic::x86_avx2_packusdw:
   case Intrinsic::x86_avx2_packuswb:
-  // TODO Add support for Intrinsic::x86_avx512_mask_packus*
+  case Intrinsic::x86_avx512_packusdw_512:
+  case Intrinsic::x86_avx512_packuswb_512:
     if (Value *V = simplifyX86pack(*II, *this, *Builder, false))
       return replaceInstUsesWith(*II, V);
     break;
diff --git a/llvm/lib/Transforms/InstCombine/InstCombineSimplifyDemanded.cpp b/llvm/lib/Transforms/InstCombine/InstCombineSimplifyDemanded.cpp
index 42ce76b7cc2..e641acbfbd9 100644
--- a/llvm/lib/Transforms/InstCombine/InstCombineSimplifyDemanded.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstCombineSimplifyDemanded.cpp
@@ -1482,8 +1482,11 @@ Value *InstCombiner::SimplifyDemandedVectorElts(Value *V, APInt DemandedElts,
     case Intrinsic::x86_avx2_packssdw:
     case Intrinsic::x86_avx2_packsswb:
     case Intrinsic::x86_avx2_packusdw:
-    case Intrinsic::x86_avx2_packuswb: {
-      // TODO Add support for Intrinsic::x86_avx512_mask_pack*
+    case Intrinsic::x86_avx2_packuswb:
+    case Intrinsic::x86_avx512_packssdw_512:
+    case Intrinsic::x86_avx512_packsswb_512:
+    case Intrinsic::x86_avx512_packusdw_512:
+    case Intrinsic::x86_avx512_packuswb_512: {
       auto *Ty0 = II->getArgOperand(0)->getType();
       unsigned InnerVWidth = Ty0->getVectorNumElements();
       assert(VWidth == (InnerVWidth * 2) && "Unexpected input size");
diff --git a/llvm/test/Transforms/InstCombine/x86-pack.ll b/llvm/test/Transforms/InstCombine/x86-pack.ll
index 68d5521d47a..f3c41a8aa47 100644
--- a/llvm/test/Transforms/InstCombine/x86-pack.ll
+++ b/llvm/test/Transforms/InstCombine/x86-pack.ll
@@ -69,6 +69,38 @@ define <32 x i8> @undef_packuswb_256() {
   ret <32 x i8> %1
 }
 
+define <32 x i16> @undef_packssdw_512() {
+; CHECK-LABEL: @undef_packssdw_512(
+; CHECK-NEXT:    ret <32 x i16> undef
+;
+  %1 = call <32 x i16> @llvm.x86.avx512.packssdw.512(<16 x i32> undef, <16 x i32> undef)
+  ret <32 x i16> %1
+}
+
+define <32 x i16> @undef_packusdw_512() {
+; CHECK-LABEL: @undef_packusdw_512(
+; CHECK-NEXT:    ret <32 x i16> undef
+;
+  %1 = call <32 x i16> @llvm.x86.avx512.packusdw.512(<16 x i32> undef, <16 x i32> undef)
+  ret <32 x i16> %1
+}
+
+define <64 x i8> @undef_packsswb_512() {
+; CHECK-LABEL: @undef_packsswb_512(
+; CHECK-NEXT:    ret <64 x i8> undef
+;
+  %1 = call <64 x i8> @llvm.x86.avx512.packsswb.512(<32 x i16> undef, <32 x i16> undef)
+  ret <64 x i8> %1
+}
+
+define <64 x i8> @undef_packuswb_512() {
+; CHECK-LABEL: @undef_packuswb_512(
+; CHECK-NEXT:    ret <64 x i8> undef
+;
+  %1 = call <64 x i8> @llvm.x86.avx512.packuswb.512(<32 x i16> undef, <32 x i16> undef)
+  ret <64 x i8> %1
+}
+
 ;
 ; Constant Folding
 ;
@@ -137,13 +169,45 @@ define <32 x i8> @fold_packuswb_256() {
   ret <32 x i8> %1
 }
 
+define <32 x i16> @fold_packssdw_512() {
+; CHECK-LABEL: @fold_packssdw_512(
+; CHECK-NEXT:    ret <32 x i16> <i16 0, i16 512, i16 32767, i16 -32768, i16 undef, i16 undef, i16 undef, i16 undef, i16 -127, i16 -32768, i16 -32767, i16 32767, i16 undef, i16 undef, i16 undef, i16 undef, i16 0, i16 512, i16 32767, i16 -32768, i16 undef, i16 undef, i16 undef, i16 undef, i16 -127, i16 -32768, i16 -32767, i16 32767, i16 undef, i16 undef, i16 undef, i16 undef>
+;
+  %1 = call <32 x i16> @llvm.x86.avx512.packssdw.512(<16 x i32> <i32 0, i32 512, i32 65535, i32 -65536, i32 -127, i32 -32768, i32 -32767, i32 32767, i32 0, i32 512, i32 65535, i32 -65536, i32 -127, i32 -32768, i32 -32767, i32 32767>, <16 x i32> undef)
+  ret <32 x i16> %1
+}
+
+define <32 x i16> @fold_packusdw_512() {
+; CHECK-LABEL: @fold_packusdw_512(
+; CHECK-NEXT:    ret <32 x i16> <i16 0, i16 0, i16 0, i16 -1, i16 0, i16 512, i16 -1, i16 0, i16 127, i16 -32768, i16 32767, i16 0, i16 0, i16 0, i16 0, i16 32767, i16 0, i16 0, i16 0, i16 -1, i16 0, i16 512, i16 -1, i16 0, i16 127, i16 -32768, i16 32767, i16 0, i16 0, i16 0, i16 0, i16 32767>
+;
+  %1 = call <32 x i16> @llvm.x86.avx512.packusdw.512(<16 x i32> <i32 0, i32 -512, i32 -65535, i32 65536, i32 127, i32 32768, i32 32767, i32 -32767, i32 0, i32 -512, i32 -65535, i32 65536, i32 127, i32 32768, i32 32767, i32 -32767>, <16 x i32> <i32 0, i32 512, i32 65535, i32 -65536, i32 -127, i32 -32768, i32 -32767, i32 32767, i32 0, i32 512, i32 65535, i32 -65536, i32 -127, i32 -32768, i32 -32767, i32 32767>)
+  ret <32 x i16> %1
+}
+
+define <64 x i8> @fold_packsswb_512() {
+; CHECK-LABEL: @fold_packsswb_512(
+; CHECK-NEXT:    ret <64 x i8> <i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0>
+;
+  %1 = call <64 x i8> @llvm.x86.avx512.packsswb.512(<32 x i16> undef, <32 x i16> zeroinitializer)
+  ret <64 x i8> %1
+}
+
+define <64 x i8> @fold_packuswb_512() {
+; CHECK-LABEL: @fold_packuswb_512(
+; CHECK-NEXT:    ret <64 x i8> <i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 -1, i8 -1, i8 -1, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 1, i8 2, i8 4, i8 8, i8 16, i8 32, i8 64, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 -1, i8 -1, i8 -1, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 1, i8 2, i8 4, i8 8, i8 16, i8 32, i8 64>
+;
+  %1 = call <64 x i8> @llvm.x86.avx512.packuswb.512(<32 x i16> zeroinitializer, <32 x i16> <i16 0, i16 -127, i16 -128, i16 -32768, i16 65536, i16 255, i16 512, i16 512, i16 -1, i16 1, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 0, i16 -127, i16 -128, i16 -32768, i16 65536, i16 255, i16 512, i16 512, i16 -1, i16 1, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64>)
+  ret <64 x i8> %1
+}
+
 ;
 ; Demanded Elts
 ;
 
 define <8 x i16> @elts_packssdw_128(<4 x i32> %a0, <4 x i32> %a1) {
 ; CHECK-LABEL: @elts_packssdw_128(
-; CHECK-NEXT:    [[TMP1:%.*]] = call <8 x i16> @llvm.x86.sse2.packssdw.128(<4 x i32> %a0, <4 x i32> undef)
+; CHECK-NEXT:    [[TMP1:%.*]] = call <8 x i16> @llvm.x86.sse2.packssdw.128(<4 x i32> [[A0:%.*]], <4 x i32> undef)
 ; CHECK-NEXT:    [[TMP2:%.*]] = shufflevector <8 x i16> [[TMP1]], <8 x i16> undef, <8 x i32> <i32 1, i32 1, i32 1, i32 1, i32 undef, i32 undef, i32 undef, i32 undef>
 ; CHECK-NEXT:    ret <8 x i16> [[TMP2]]
 ;
@@ -156,7 +220,7 @@ define <8 x i16> @elts_packssdw_128(<4 x i32> %a0, <4 x i32> %a1) {
 
 define <8 x i16> @elts_packusdw_128(<4 x i32> %a0, <4 x i32> %a1) {
 ; CHECK-LABEL: @elts_packusdw_128(
-; CHECK-NEXT:    [[TMP1:%.*]] = call <8 x i16> @llvm.x86.sse41.packusdw(<4 x i32> %a0, <4 x i32> %a1)
+; CHECK-NEXT:    [[TMP1:%.*]] = call <8 x i16> @llvm.x86.sse41.packusdw(<4 x i32> [[A0:%.*]], <4 x i32> [[A1:%.*]])
 ; CHECK-NEXT:    ret <8 x i16> [[TMP1]]
 ;
   %1 = insertelement <4 x i32> %a0, i32 0, i32 0
@@ -190,7 +254,7 @@ define <16 x i8> @elts_packuswb_128(<8 x i16> %a0, <8 x i16> %a1) {
 
 define <16 x i16> @elts_packssdw_256(<8 x i32> %a0, <8 x i32> %a1) {
 ; CHECK-LABEL: @elts_packssdw_256(
-; CHECK-NEXT:    [[TMP1:%.*]] = call <16 x i16> @llvm.x86.avx2.packssdw(<8 x i32> %a0, <8 x i32> undef)
+; CHECK-NEXT:    [[TMP1:%.*]] = call <16 x i16> @llvm.x86.avx2.packssdw(<8 x i32> [[A0:%.*]], <8 x i32> undef)
 ; CHECK-NEXT:    ret <16 x i16> [[TMP1]]
 ;
   %1 = shufflevector <8 x i32> %a0, <8 x i32> undef, <8 x i32> <i32 1, i32 0, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
@@ -202,7 +266,7 @@ define <16 x i16> @elts_packssdw_256(<8 x i32> %a0, <8 x i32> %a1) {
 
 define <16 x i16> @elts_packusdw_256(<8 x i32> %a0, <8 x i32> %a1) {
 ; CHECK-LABEL: @elts_packusdw_256(
-; CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <8 x i32> %a1, <8 x i32> undef, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+; CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <8 x i32> [[A1:%.*]], <8 x i32> undef, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
 ; CHECK-NEXT:    [[TMP2:%.*]] = call <16 x i16> @llvm.x86.avx2.packusdw(<8 x i32> undef, <8 x i32> [[TMP1]])
 ; CHECK-NEXT:    [[TMP3:%.*]] = shufflevector <16 x i16> [[TMP2]], <16 x i16> undef, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef>
 ; CHECK-NEXT:    ret <16 x i16> [[TMP3]]
@@ -236,6 +300,56 @@ define <32 x i8> @elts_packuswb_256(<16 x i16> %a0, <16 x i16> %a1) {
   ret <32 x i8> %4
 }
 
+define <32 x i16> @elts_packssdw_512(<16 x i32> %a0, <16 x i32> %a1) {
+; CHECK-LABEL: @elts_packssdw_512(
+; CHECK-NEXT:    [[TMP1:%.*]] = call <32 x i16> @llvm.x86.avx512.packssdw.512(<16 x i32> [[A0:%.*]], <16 x i32> undef)
+; CHECK-NEXT:    ret <32 x i16> [[TMP1]]
+;
+  %1 = shufflevector <16 x i32> %a0, <16 x i32> undef, <16 x i32> <i32 1, i32 0, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 9, i32 8, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+  %2 = shufflevector <16 x i32> %a1, <16 x i32> undef, <16 x i32> <i32 undef, i32 2, i32 1, i32 undef, i32 undef, i32 6, i32 5, i32 undef, i32 undef, i32 10, i32 9, i32 undef, i32 undef, i32 14, i32 13, i32 undef>
+  %3 = call <32 x i16> @llvm.x86.avx512.packssdw.512(<16 x i32> %1, <16 x i32> %2)
+  %4 = shufflevector <32 x i16> %3, <32 x i16> undef, <32 x i32> <i32 undef, i32 undef, i32 2, i32 3, i32 4, i32 undef, i32 undef, i32 7, i32 8, i32 undef, i32 undef, i32 11, i32 12, i32 undef, i32 undef, i32 15, i32 undef, i32 undef, i32 18, i32 19, i32 20, i32 undef, i32 undef, i32 23, i32 24, i32 undef, i32 undef, i32 27, i32 28, i32 undef, i32 undef, i32 31>
+  ret <32 x i16> %4
+}
+
+define <32 x i16> @elts_packusdw_512(<16 x i32> %a0, <16 x i32> %a1) {
+; CHECK-LABEL: @elts_packusdw_512(
+; CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <16 x i32> [[A1:%.*]], <16 x i32> undef, <16 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0, i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8>
+; CHECK-NEXT:    [[TMP2:%.*]] = call <32 x i16> @llvm.x86.avx512.packusdw.512(<16 x i32> undef, <16 x i32> [[TMP1]])
+; CHECK-NEXT:    [[TMP3:%.*]] = shufflevector <32 x i16> [[TMP2]], <32 x i16> undef, <32 x i32> <i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 20, i32 21, i32 22, i32 23, i32 undef, i32 undef, i32 undef, i32 undef, i32 28, i32 29, i32 30, i32 31, i32 undef, i32 undef, i32 undef, i32 undef>
+; CHECK-NEXT:    ret <32 x i16> [[TMP3]]
+;
+  %1 = shufflevector <16 x i32> %a0, <16 x i32> undef, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+  %2 = shufflevector <16 x i32> %a1, <16 x i32> undef, <16 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0, i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8>
+  %3 = call <32 x i16> @llvm.x86.avx512.packusdw.512(<16 x i32> %1, <16 x i32> %2)
+  %4 = shufflevector <32 x i16> %3, <32 x i16> undef, <32 x i32> <i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 20, i32 21, i32 22, i32 23, i32 undef, i32 undef, i32 undef, i32 undef, i32 28, i32 29, i32 30, i32 31, i32 undef, i32 undef, i32 undef, i32 undef>
+  ret <32 x i16> %4
+}
+
+define <64 x i8> @elts_packsswb_512(<32 x i16> %a0, <32 x i16> %a1) {
+; CHECK-LABEL: @elts_packsswb_512(
+; CHECK-NEXT:    ret <64 x i8> zeroinitializer
+;
+  %1 = insertelement <32 x i16> %a0, i16 0, i32 0
+  %2 = insertelement <32 x i16> %a1, i16 0, i32 8
+  %3 = insertelement <32 x i16> %1, i16 0, i32 16
+  %4 = insertelement <32 x i16> %2, i16 0, i32 24
+  %5 = call <64 x i8> @llvm.x86.avx512.packsswb.512(<32 x i16> %3, <32 x i16> %4)
+  %6 = shufflevector <64 x i8> %5, <64 x i8> undef, <64 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 24, i32 24, i32 24, i32 24, i32 24, i32 24, i32 24, i32 24, i32 24, i32 24, i32 24, i32 24, i32 24, i32 24, i32 24, i32 24, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 56, i32 56, i32 56, i32 56, i32 56, i32 56, i32 56, i32 56, i32 56, i32 56, i32 56, i32 56, i32 56, i32 56, i32 56, i32 56>
+  ret <64 x i8> %6
+}
+
+define <64 x i8> @elts_packuswb_512(<32 x i16> %a0, <32 x i16> %a1) {
+; CHECK-LABEL: @elts_packuswb_512(
+; CHECK-NEXT:    ret <64 x i8> undef
+;
+  %1 = insertelement <32 x i16> undef, i16 0, i32 1
+  %2 = insertelement <32 x i16> undef, i16 0, i32 0
+  %3 = call <64 x i8> @llvm.x86.avx512.packuswb.512(<32 x i16> %1, <32 x i16> %2)
+  %4 = shufflevector <64 x i8> %3, <64 x i8> undef, <64 x i32> zeroinitializer
+  ret <64 x i8> %4
+}
+
 declare <8 x i16> @llvm.x86.sse2.packssdw.128(<4 x i32>, <4 x i32>) nounwind readnone
 declare <16 x i8> @llvm.x86.sse2.packsswb.128(<8 x i16>, <8 x i16>) nounwind readnone
 declare <16 x i8> @llvm.x86.sse2.packuswb.128(<8 x i16>, <8 x i16>) nounwind readnone
@@ -245,3 +359,8 @@ declare <16 x i16> @llvm.x86.avx2.packssdw(<8 x i32>, <8 x i32>) nounwind readno
 declare <16 x i16> @llvm.x86.avx2.packusdw(<8 x i32>, <8 x i32>) nounwind readnone
 declare <32 x i8> @llvm.x86.avx2.packsswb(<16 x i16>, <16 x i16>) nounwind readnone
 declare <32 x i8> @llvm.x86.avx2.packuswb(<16 x i16>, <16 x i16>) nounwind readnone
+
+declare <32 x i16> @llvm.x86.avx512.packssdw.512(<16 x i32>, <16 x i32>) nounwind readnone
+declare <32 x i16> @llvm.x86.avx512.packusdw.512(<16 x i32>, <16 x i32>) nounwind readnone
+declare <64 x i8> @llvm.x86.avx512.packsswb.512(<32 x i16>, <32 x i16>) nounwind readnone
+declare <64 x i8> @llvm.x86.avx512.packuswb.512(<32 x i16>, <32 x i16>) nounwind readnone