[X86] Add DAG combine to form saturating VTRUNCUS/VTRUNCS from VTRUNC

We already do this for ISD::TRUNCATE, but we can do the same for X86ISD::VTRUNC

Differential Revision: https://reviews.llvm.org/D68432

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@373765 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/Target/X86/X86ISelLowering.cpp b/lib/Target/X86/X86ISelLowering.cpp
index 407c4c8137d30af75ba62156c0b91142fce32c0a..09d90ff4e7ec5aef1bbe42dffd7acb24ad80441c 100644 (file)
--- a/lib/Target/X86/X86ISelLowering.cpp
+++ b/lib/Target/X86/X86ISelLowering.cpp
@@ -41145,6 +41145,19 @@ static SDValue combineTruncate(SDNode *N, SelectionDAG &DAG,
   return combineVectorTruncation(N, DAG, Subtarget);
 }
 
+static SDValue combineVTRUNC(SDNode *N, SelectionDAG &DAG) {
+  EVT VT = N->getValueType(0);
+  SDValue In = N->getOperand(0);
+  SDLoc DL(N);
+
+  if (auto SSatVal = detectSSatPattern(In, VT))
+    return DAG.getNode(X86ISD::VTRUNCS, DL, VT, SSatVal);
+  if (auto USatVal = detectUSatPattern(In, VT, DAG, DL))
+    return DAG.getNode(X86ISD::VTRUNCUS, DL, VT, USatVal);
+
+  return SDValue();
+}
+
 /// Returns the negated value if the node \p N flips sign of FP value.
 ///
 /// FP-negation node may have different forms: FNEG(x), FXOR (x, 0x80000000)
@@ -44503,6 +44516,7 @@ SDValue X86TargetLowering::PerformDAGCombine(SDNode *N,
   case ISD::FSUB:           return combineFaddFsub(N, DAG, Subtarget);
   case ISD::FNEG:           return combineFneg(N, DAG, Subtarget);
   case ISD::TRUNCATE:       return combineTruncate(N, DAG, Subtarget);
+  case X86ISD::VTRUNC:      return combineVTRUNC(N, DAG);
   case X86ISD::ANDNP:       return combineAndnp(N, DAG, DCI, Subtarget);
   case X86ISD::FAND:        return combineFAnd(N, DAG, Subtarget);
   case X86ISD::FANDN:       return combineFAndn(N, DAG, Subtarget);

diff --git a/test/CodeGen/X86/avx512-trunc.ll b/test/CodeGen/X86/avx512-trunc.ll
index 57cabca007661b60b5fb563b2148545b6e55e98e..065a55fe7c51bea91ad2493fda6e0f8398d21dd3 100644 (file)
--- a/test/CodeGen/X86/avx512-trunc.ll
+++ b/test/CodeGen/X86/avx512-trunc.ll
@@ -750,8 +750,7 @@ define <16 x i8> @usat_trunc_db_256(<8 x i32> %x) {
 ;
 ; SKX-LABEL: usat_trunc_db_256:
 ; SKX:       ## %bb.0:
-; SKX-NEXT:    vpminud {{.*}}(%rip){1to8}, %ymm0, %ymm0
-; SKX-NEXT:    vpmovdb %ymm0, %xmm0
+; SKX-NEXT:    vpmovusdb %ymm0, %xmm0
 ; SKX-NEXT:    vzeroupper
 ; SKX-NEXT:    retq
   %tmp1 = icmp ult <8 x i32> %x, <i32 255, i32 255, i32 255, i32 255, i32 255, i32 255, i32 255, i32 255>

diff --git a/test/CodeGen/X86/masked_store_trunc_ssat.ll b/test/CodeGen/X86/masked_store_trunc_ssat.ll
index af413e85be36f30acab70d6462be83df06a761f7..83bf33c4f7f1e7abfd8dfa8d35d482e048cdd91a 100644 (file)
--- a/test/CodeGen/X86/masked_store_trunc_ssat.ll
+++ b/test/CodeGen/X86/masked_store_trunc_ssat.ll
@@ -1432,9 +1432,7 @@ define void @truncstore_v8i64_v8i8(<8 x i64> %x, <8 x i8>* %p, <8 x i32> %mask)
 ; AVX512F:       # %bb.0:
 ; AVX512F-NEXT:    # kill: def $ymm1 killed $ymm1 def $zmm1
 ; AVX512F-NEXT:    vptestmd %zmm1, %zmm1, %k0
-; AVX512F-NEXT:    vpminsq {{.*}}(%rip){1to8}, %zmm0, %zmm0
-; AVX512F-NEXT:    vpmaxsq {{.*}}(%rip){1to8}, %zmm0, %zmm0
-; AVX512F-NEXT:    vpmovqb %zmm0, %xmm0
+; AVX512F-NEXT:    vpmovsqb %zmm0, %xmm0
 ; AVX512F-NEXT:    kmovw %k0, %eax
 ; AVX512F-NEXT:    testb $1, %al
 ; AVX512F-NEXT:    jne .LBB2_1

diff --git a/test/CodeGen/X86/masked_store_trunc_usat.ll b/test/CodeGen/X86/masked_store_trunc_usat.ll
index 1f47ed5d156debd36759ff80b9e07f1ceff5ea67..00fdbf26e70dbc2f49debc4174f42f6771ea717f 100644 (file)
--- a/test/CodeGen/X86/masked_store_trunc_usat.ll
+++ b/test/CodeGen/X86/masked_store_trunc_usat.ll
@@ -1215,8 +1215,7 @@ define void @truncstore_v8i64_v8i8(<8 x i64> %x, <8 x i8>* %p, <8 x i32> %mask)
 ; AVX512F:       # %bb.0:
 ; AVX512F-NEXT:    # kill: def $ymm1 killed $ymm1 def $zmm1
 ; AVX512F-NEXT:    vptestmd %zmm1, %zmm1, %k0
-; AVX512F-NEXT:    vpminuq {{.*}}(%rip){1to8}, %zmm0, %zmm0
-; AVX512F-NEXT:    vpmovqb %zmm0, %xmm0
+; AVX512F-NEXT:    vpmovusqb %zmm0, %xmm0
 ; AVX512F-NEXT:    kmovw %k0, %eax
 ; AVX512F-NEXT:    testb $1, %al
 ; AVX512F-NEXT:    jne .LBB2_1

diff --git a/test/CodeGen/X86/vector-trunc-packus.ll b/test/CodeGen/X86/vector-trunc-packus.ll
index a5fa41c521f801099f857f70c86ac3b4a437e8d8..8b777e9805a3572fafa9d328219805e6da55819c 100644 (file)
--- a/test/CodeGen/X86/vector-trunc-packus.ll
+++ b/test/CodeGen/X86/vector-trunc-packus.ll
@@ -1658,10 +1658,9 @@ define <8 x i8> @trunc_packus_v8i64_v8i8(<8 x i64> %a0) {
 ;
 ; AVX512-LABEL: trunc_packus_v8i64_v8i8:
 ; AVX512:       # %bb.0:
-; AVX512-NEXT:    vpminsq {{.*}}(%rip){1to8}, %zmm0, %zmm0
 ; AVX512-NEXT:    vpxor %xmm1, %xmm1, %xmm1
 ; AVX512-NEXT:    vpmaxsq %zmm1, %zmm0, %zmm0
-; AVX512-NEXT:    vpmovqb %zmm0, %xmm0
+; AVX512-NEXT:    vpmovusqb %zmm0, %xmm0
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = icmp slt <8 x i64> %a0, <i64 255, i64 255, i64 255, i64 255, i64 255, i64 255, i64 255, i64 255>
@@ -2790,10 +2789,9 @@ define <8 x i8> @trunc_packus_v8i32_v8i8(<8 x i32> %a0) {
 ;
 ; AVX512VL-LABEL: trunc_packus_v8i32_v8i8:
 ; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vpminsd {{.*}}(%rip){1to8}, %ymm0, %ymm0
 ; AVX512VL-NEXT:    vpxor %xmm1, %xmm1, %xmm1
 ; AVX512VL-NEXT:    vpmaxsd %ymm1, %ymm0, %ymm0
-; AVX512VL-NEXT:    vpmovdb %ymm0, %xmm0
+; AVX512VL-NEXT:    vpmovusdb %ymm0, %xmm0
 ; AVX512VL-NEXT:    vzeroupper
 ; AVX512VL-NEXT:    retq
 ;
@@ -2809,10 +2807,9 @@ define <8 x i8> @trunc_packus_v8i32_v8i8(<8 x i32> %a0) {
 ;
 ; AVX512BWVL-LABEL: trunc_packus_v8i32_v8i8:
 ; AVX512BWVL:       # %bb.0:
-; AVX512BWVL-NEXT:    vpminsd {{.*}}(%rip){1to8}, %ymm0, %ymm0
 ; AVX512BWVL-NEXT:    vpxor %xmm1, %xmm1, %xmm1
 ; AVX512BWVL-NEXT:    vpmaxsd %ymm1, %ymm0, %ymm0
-; AVX512BWVL-NEXT:    vpmovdb %ymm0, %xmm0
+; AVX512BWVL-NEXT:    vpmovusdb %ymm0, %xmm0
 ; AVX512BWVL-NEXT:    vzeroupper
 ; AVX512BWVL-NEXT:    retq
   %1 = icmp slt <8 x i32> %a0, <i32 255, i32 255, i32 255, i32 255, i32 255, i32 255, i32 255, i32 255>

diff --git a/test/CodeGen/X86/vector-trunc-ssat.ll b/test/CodeGen/X86/vector-trunc-ssat.ll
index a298382cfb0107cfc2c9c10e0a77a67b4fcbb9fc..8e99c0b64a0d40a954f555831f6502899414280b 100644 (file)
--- a/test/CodeGen/X86/vector-trunc-ssat.ll
+++ b/test/CodeGen/X86/vector-trunc-ssat.ll
@@ -1567,9 +1567,7 @@ define <8 x i8> @trunc_ssat_v8i64_v8i8(<8 x i64> %a0) {
 ;
 ; AVX512-LABEL: trunc_ssat_v8i64_v8i8:
 ; AVX512:       # %bb.0:
-; AVX512-NEXT:    vpminsq {{.*}}(%rip){1to8}, %zmm0, %zmm0
-; AVX512-NEXT:    vpmaxsq {{.*}}(%rip){1to8}, %zmm0, %zmm0
-; AVX512-NEXT:    vpmovqb %zmm0, %xmm0
+; AVX512-NEXT:    vpmovsqb %zmm0, %xmm0
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = icmp slt <8 x i64> %a0, <i64 127, i64 127, i64 127, i64 127, i64 127, i64 127, i64 127, i64 127>
@@ -2776,9 +2774,7 @@ define <8 x i8> @trunc_ssat_v8i32_v8i8(<8 x i32> %a0) {
 ;
 ; AVX512VL-LABEL: trunc_ssat_v8i32_v8i8:
 ; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vpminsd {{.*}}(%rip){1to8}, %ymm0, %ymm0
-; AVX512VL-NEXT:    vpmaxsd {{.*}}(%rip){1to8}, %ymm0, %ymm0
-; AVX512VL-NEXT:    vpmovdb %ymm0, %xmm0
+; AVX512VL-NEXT:    vpmovsdb %ymm0, %xmm0
 ; AVX512VL-NEXT:    vzeroupper
 ; AVX512VL-NEXT:    retq
 ;
@@ -2794,9 +2790,7 @@ define <8 x i8> @trunc_ssat_v8i32_v8i8(<8 x i32> %a0) {
 ;
 ; AVX512BWVL-LABEL: trunc_ssat_v8i32_v8i8:
 ; AVX512BWVL:       # %bb.0:
-; AVX512BWVL-NEXT:    vpminsd {{.*}}(%rip){1to8}, %ymm0, %ymm0
-; AVX512BWVL-NEXT:    vpmaxsd {{.*}}(%rip){1to8}, %ymm0, %ymm0
-; AVX512BWVL-NEXT:    vpmovdb %ymm0, %xmm0
+; AVX512BWVL-NEXT:    vpmovsdb %ymm0, %xmm0
 ; AVX512BWVL-NEXT:    vzeroupper
 ; AVX512BWVL-NEXT:    retq
   %1 = icmp slt <8 x i32> %a0, <i32 127, i32 127, i32 127, i32 127, i32 127, i32 127, i32 127, i32 127>

diff --git a/test/CodeGen/X86/vector-trunc-usat.ll b/test/CodeGen/X86/vector-trunc-usat.ll
index f1c19d91cd323986be1acb9610c649ffaf9677d4..1b38cc43ab114dc29a381113f08ee3dd931d9e88 100644 (file)
--- a/test/CodeGen/X86/vector-trunc-usat.ll
+++ b/test/CodeGen/X86/vector-trunc-usat.ll
@@ -1175,8 +1175,7 @@ define <8 x i8> @trunc_usat_v8i64_v8i8(<8 x i64> %a0) {
 ;
 ; AVX512-LABEL: trunc_usat_v8i64_v8i8:
 ; AVX512:       # %bb.0:
-; AVX512-NEXT:    vpminuq {{.*}}(%rip){1to8}, %zmm0, %zmm0
-; AVX512-NEXT:    vpmovqb %zmm0, %xmm0
+; AVX512-NEXT:    vpmovusqb %zmm0, %xmm0
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %1 = icmp ult <8 x i64> %a0, <i64 255, i64 255, i64 255, i64 255, i64 255, i64 255, i64 255, i64 255>
@@ -1952,8 +1951,7 @@ define <8 x i8> @trunc_usat_v8i32_v8i8(<8 x i32> %a0) {
 ;
 ; AVX512VL-LABEL: trunc_usat_v8i32_v8i8:
 ; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vpminud {{.*}}(%rip){1to8}, %ymm0, %ymm0
-; AVX512VL-NEXT:    vpmovdb %ymm0, %xmm0
+; AVX512VL-NEXT:    vpmovusdb %ymm0, %xmm0
 ; AVX512VL-NEXT:    vzeroupper
 ; AVX512VL-NEXT:    retq
 ;
@@ -1967,8 +1965,7 @@ define <8 x i8> @trunc_usat_v8i32_v8i8(<8 x i32> %a0) {
 ;
 ; AVX512BWVL-LABEL: trunc_usat_v8i32_v8i8:
 ; AVX512BWVL:       # %bb.0:
-; AVX512BWVL-NEXT:    vpminud {{.*}}(%rip){1to8}, %ymm0, %ymm0
-; AVX512BWVL-NEXT:    vpmovdb %ymm0, %xmm0
+; AVX512BWVL-NEXT:    vpmovusdb %ymm0, %xmm0
 ; AVX512BWVL-NEXT:    vzeroupper
 ; AVX512BWVL-NEXT:    retq
   %1 = icmp ult <8 x i32> %a0, <i32 255, i32 255, i32 255, i32 255, i32 255, i32 255, i32 255, i32 255>