[X86][CostModel] X86TTIImpl::getShuffleCost(): subvector insertions are cheap

This is similar to the subvector extractions,
except that the 0'th subvector isn't free to insert,
because we generally don't know whether or not
the upper elements need to be preserved:
https://godbolt.org/z/rsxP5W4sW

This is needed to avoid regressions in D100684

Reviewed By: RKSimon

Differential Revision: https://reviews.llvm.org/D100698

diff --git a/llvm/lib/Target/X86/X86TargetTransformInfo.cpp b/llvm/lib/Target/X86/X86TargetTransformInfo.cpp
index 9a17077..82de3ff 100644 (file)
--- a/llvm/lib/Target/X86/X86TargetTransformInfo.cpp
+++ b/llvm/lib/Target/X86/X86TargetTransformInfo.cpp
@@ -1025,6 +1025,19 @@ InstructionCost X86TTIImpl::getShuffleCost(TTI::ShuffleKind Kind,
     }
   }
 
+  // Subvector insertions are cheap if the subvectors are aligned.
+  // Note that in general, the insertion starting at the beginning of a vector
+  // isn't free, because we need to preserve the rest of the wide vector.
+  if (Kind == TTI::SK_InsertSubvector && LT.second.isVector()) {
+    int NumElts = LT.second.getVectorNumElements();
+    std::pair<int, MVT> SubLT = TLI->getTypeLegalizationCost(DL, SubTp);
+    if (SubLT.second.isVector()) {
+      int NumSubElts = SubLT.second.getVectorNumElements();
+      if ((Index % NumSubElts) == 0 && (NumElts % NumSubElts) == 0)
+        return SubLT.first;
+    }
+  }
+
   // Handle some common (illegal) sub-vector types as they are often very cheap
   // to shuffle even on targets without PSHUFB.
   EVT VT = TLI->getValueType(DL, BaseTp);

diff --git a/llvm/test/Analysis/CostModel/X86/masked-intrinsic-cost-inseltpoison.ll b/llvm/test/Analysis/CostModel/X86/masked-intrinsic-cost-inseltpoison.ll
index 75a094e..b853b2b 100644 (file)
--- a/llvm/test/Analysis/CostModel/X86/masked-intrinsic-cost-inseltpoison.ll
+++ b/llvm/test/Analysis/CostModel/X86/masked-intrinsic-cost-inseltpoison.ll
@@ -71,7 +71,7 @@ define i32 @masked_load() {
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0v16f32(<16 x float>* undef, i32 1, <16 x i1> undef, <16 x float> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0v8f32(<8 x float>* undef, i32 1, <8 x i1> undef, <8 x float> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0v4f32(<4 x float>* undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0v2f32(<2 x float>* undef, i32 1, <2 x i1> undef, <2 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0v2f32(<2 x float>* undef, i32 1, <2 x i1> undef, <2 x float> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0v8i64(<8 x i64>* undef, i32 1, <8 x i1> undef, <8 x i64> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0v4i64(<4 x i64>* undef, i32 1, <4 x i1> undef, <4 x i64> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0v2i64(<2 x i64>* undef, i32 1, <2 x i1> undef, <2 x i64> undef)
@@ -79,7 +79,7 @@ define i32 @masked_load() {
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0v16i32(<16 x i32>* undef, i32 1, <16 x i1> undef, <16 x i32> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0v8i32(<8 x i32>* undef, i32 1, <8 x i1> undef, <8 x i32> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0v4i32(<4 x i32>* undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0v2i32(<2 x i32>* undef, i32 1, <2 x i1> undef, <2 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0v2i32(<2 x i32>* undef, i32 1, <2 x i1> undef, <2 x i32> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 148 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0v32i16(<32 x i16>* undef, i32 1, <32 x i1> undef, <32 x i16> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0v16i16(<16 x i16>* undef, i32 1, <16 x i1> undef, <16 x i16> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0v8i16(<8 x i16>* undef, i32 1, <8 x i1> undef, <8 x i16> undef)
@@ -98,7 +98,7 @@ define i32 @masked_load() {
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0v16f32(<16 x float>* undef, i32 1, <16 x i1> undef, <16 x float> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0v8f32(<8 x float>* undef, i32 1, <8 x i1> undef, <8 x float> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0v4f32(<4 x float>* undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0v2f32(<2 x float>* undef, i32 1, <2 x i1> undef, <2 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0v2f32(<2 x float>* undef, i32 1, <2 x i1> undef, <2 x float> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0v8i64(<8 x i64>* undef, i32 1, <8 x i1> undef, <8 x i64> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0v4i64(<4 x i64>* undef, i32 1, <4 x i1> undef, <4 x i64> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0v2i64(<2 x i64>* undef, i32 1, <2 x i1> undef, <2 x i64> undef)
@@ -106,7 +106,7 @@ define i32 @masked_load() {
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0v16i32(<16 x i32>* undef, i32 1, <16 x i1> undef, <16 x i32> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0v8i32(<8 x i32>* undef, i32 1, <8 x i1> undef, <8 x i32> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0v4i32(<4 x i32>* undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0v2i32(<2 x i32>* undef, i32 1, <2 x i1> undef, <2 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0v2i32(<2 x i32>* undef, i32 1, <2 x i1> undef, <2 x i32> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 148 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0v32i16(<32 x i16>* undef, i32 1, <32 x i1> undef, <32 x i16> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0v16i16(<16 x i16>* undef, i32 1, <16 x i1> undef, <16 x i16> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0v8i16(<8 x i16>* undef, i32 1, <8 x i1> undef, <8 x i16> undef)
@@ -125,7 +125,7 @@ define i32 @masked_load() {
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0v16f32(<16 x float>* undef, i32 1, <16 x i1> undef, <16 x float> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0v8f32(<8 x float>* undef, i32 1, <8 x i1> undef, <8 x float> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0v4f32(<4 x float>* undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0v2f32(<2 x float>* undef, i32 1, <2 x i1> undef, <2 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0v2f32(<2 x float>* undef, i32 1, <2 x i1> undef, <2 x float> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0v8i64(<8 x i64>* undef, i32 1, <8 x i1> undef, <8 x i64> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0v4i64(<4 x i64>* undef, i32 1, <4 x i1> undef, <4 x i64> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0v2i64(<2 x i64>* undef, i32 1, <2 x i1> undef, <2 x i64> undef)
@@ -133,15 +133,15 @@ define i32 @masked_load() {
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0v16i32(<16 x i32>* undef, i32 1, <16 x i1> undef, <16 x i32> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0v8i32(<8 x i32>* undef, i32 1, <8 x i1> undef, <8 x i32> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0v4i32(<4 x i32>* undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0v2i32(<2 x i32>* undef, i32 1, <2 x i1> undef, <2 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0v2i32(<2 x i32>* undef, i32 1, <2 x i1> undef, <2 x i32> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0v32i16(<32 x i16>* undef, i32 1, <32 x i1> undef, <32 x i16> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0v16i16(<16 x i16>* undef, i32 1, <16 x i1> undef, <16 x i16> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0v8i16(<8 x i16>* undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0v4i16(<4 x i16>* undef, i32 1, <4 x i1> undef, <4 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0v4i16(<4 x i16>* undef, i32 1, <4 x i1> undef, <4 x i16> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0v64i8(<64 x i8>* undef, i32 1, <64 x i1> undef, <64 x i8> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0v32i8(<32 x i8>* undef, i32 1, <32 x i1> undef, <32 x i8> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0v16i8(<16 x i8>* undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0v8i8(<8 x i8>* undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0v8i8(<8 x i8>* undef, i32 1, <8 x i1> undef, <8 x i8> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
   %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0v8f64(<8 x double>* undef, i32 1, <8 x i1> undef, <8 x double> undef)
@@ -240,7 +240,7 @@ define i32 @masked_store() {
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v16f32.p0v16f32(<16 x float> undef, <16 x float>* undef, i32 1, <16 x i1> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v8f32.p0v8f32(<8 x float> undef, <8 x float>* undef, i32 1, <8 x i1> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4f32.p0v4f32(<4 x float> undef, <4 x float>* undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.store.v2f32.p0v2f32(<2 x float> undef, <2 x float>* undef, i32 1, <2 x i1> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2f32.p0v2f32(<2 x float> undef, <2 x float>* undef, i32 1, <2 x i1> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v8i64.p0v8i64(<8 x i64> undef, <8 x i64>* undef, i32 1, <8 x i1> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4i64.p0v4i64(<4 x i64> undef, <4 x i64>* undef, i32 1, <4 x i1> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2i64.p0v2i64(<2 x i64> undef, <2 x i64>* undef, i32 1, <2 x i1> undef)
@@ -248,7 +248,7 @@ define i32 @masked_store() {
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v16i32.p0v16i32(<16 x i32> undef, <16 x i32>* undef, i32 1, <16 x i1> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v8i32.p0v8i32(<8 x i32> undef, <8 x i32>* undef, i32 1, <8 x i1> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4i32.p0v4i32(<4 x i32> undef, <4 x i32>* undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.store.v2i32.p0v2i32(<2 x i32> undef, <2 x i32>* undef, i32 1, <2 x i1> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2i32.p0v2i32(<2 x i32> undef, <2 x i32>* undef, i32 1, <2 x i1> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: call void @llvm.masked.store.v32i16.p0v32i16(<32 x i16> undef, <32 x i16>* undef, i32 1, <32 x i1> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: call void @llvm.masked.store.v16i16.p0v16i16(<16 x i16> undef, <16 x i16>* undef, i32 1, <16 x i1> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i16.p0v8i16(<8 x i16> undef, <8 x i16>* undef, i32 1, <8 x i1> undef)
@@ -267,7 +267,7 @@ define i32 @masked_store() {
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16f32.p0v16f32(<16 x float> undef, <16 x float>* undef, i32 1, <16 x i1> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f32.p0v8f32(<8 x float> undef, <8 x float>* undef, i32 1, <8 x i1> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f32.p0v4f32(<4 x float> undef, <4 x float>* undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.store.v2f32.p0v2f32(<2 x float> undef, <2 x float>* undef, i32 1, <2 x i1> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2f32.p0v2f32(<2 x float> undef, <2 x float>* undef, i32 1, <2 x i1> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i64.p0v8i64(<8 x i64> undef, <8 x i64>* undef, i32 1, <8 x i1> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i64.p0v4i64(<4 x i64> undef, <4 x i64>* undef, i32 1, <4 x i1> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2i64.p0v2i64(<2 x i64> undef, <2 x i64>* undef, i32 1, <2 x i1> undef)
@@ -275,7 +275,7 @@ define i32 @masked_store() {
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i32.p0v16i32(<16 x i32> undef, <16 x i32>* undef, i32 1, <16 x i1> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i32.p0v8i32(<8 x i32> undef, <8 x i32>* undef, i32 1, <8 x i1> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i32.p0v4i32(<4 x i32> undef, <4 x i32>* undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.store.v2i32.p0v2i32(<2 x i32> undef, <2 x i32>* undef, i32 1, <2 x i1> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2i32.p0v2i32(<2 x i32> undef, <2 x i32>* undef, i32 1, <2 x i1> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 168 for instruction: call void @llvm.masked.store.v32i16.p0v32i16(<32 x i16> undef, <32 x i16>* undef, i32 1, <32 x i1> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: call void @llvm.masked.store.v16i16.p0v16i16(<16 x i16> undef, <16 x i16>* undef, i32 1, <16 x i1> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i16.p0v8i16(<8 x i16> undef, <8 x i16>* undef, i32 1, <8 x i1> undef)
@@ -294,7 +294,7 @@ define i32 @masked_store() {
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16f32.p0v16f32(<16 x float> undef, <16 x float>* undef, i32 1, <16 x i1> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f32.p0v8f32(<8 x float> undef, <8 x float>* undef, i32 1, <8 x i1> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f32.p0v4f32(<4 x float> undef, <4 x float>* undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.store.v2f32.p0v2f32(<2 x float> undef, <2 x float>* undef, i32 1, <2 x i1> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2f32.p0v2f32(<2 x float> undef, <2 x float>* undef, i32 1, <2 x i1> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i64.p0v8i64(<8 x i64> undef, <8 x i64>* undef, i32 1, <8 x i1> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i64.p0v4i64(<4 x i64> undef, <4 x i64>* undef, i32 1, <4 x i1> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2i64.p0v2i64(<2 x i64> undef, <2 x i64>* undef, i32 1, <2 x i1> undef)
@@ -302,15 +302,15 @@ define i32 @masked_store() {
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i32.p0v16i32(<16 x i32> undef, <16 x i32>* undef, i32 1, <16 x i1> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i32.p0v8i32(<8 x i32> undef, <8 x i32>* undef, i32 1, <8 x i1> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i32.p0v4i32(<4 x i32> undef, <4 x i32>* undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.store.v2i32.p0v2i32(<2 x i32> undef, <2 x i32>* undef, i32 1, <2 x i1> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2i32.p0v2i32(<2 x i32> undef, <2 x i32>* undef, i32 1, <2 x i1> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v32i16.p0v32i16(<32 x i16> undef, <32 x i16>* undef, i32 1, <32 x i1> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i16.p0v16i16(<16 x i16> undef, <16 x i16>* undef, i32 1, <16 x i1> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i16.p0v8i16(<8 x i16> undef, <8 x i16>* undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v4i16.p0v4i16(<4 x i16> undef, <4 x i16>* undef, i32 1, <4 x i1> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v4i16.p0v4i16(<4 x i16> undef, <4 x i16>* undef, i32 1, <4 x i1> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v64i8.p0v64i8(<64 x i8> undef, <64 x i8>* undef, i32 1, <64 x i1> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v32i8.p0v32i8(<32 x i8> undef, <32 x i8>* undef, i32 1, <32 x i1> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i8.p0v16i8(<16 x i8> undef, <16 x i8>* undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v8i8.p0v8i8(<8 x i8> undef, <8 x i8>* undef, i32 1, <8 x i1> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v8i8.p0v8i8(<8 x i8> undef, <8 x i8>* undef, i32 1, <8 x i1> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
   call void @llvm.masked.store.v8f64.p0v8f64(<8 x double> undef, <8 x double>* undef, i32 1, <8 x i1> undef)
@@ -1149,12 +1149,12 @@ define void @test5(<2 x i32> %trigger, <2 x float>* %addr, <2 x float> %val) {
 ;
 ; AVX-LABEL: 'test5'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <2 x i32> %trigger, zeroinitializer
-; AVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.store.v2f32.p0v2f32(<2 x float> %val, <2 x float>* %addr, i32 4, <2 x i1> %mask)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2f32.p0v2f32(<2 x float> %val, <2 x float>* %addr, i32 4, <2 x i1> %mask)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512-LABEL: 'test5'
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <2 x i32> %trigger, zeroinitializer
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.store.v2f32.p0v2f32(<2 x float> %val, <2 x float>* %addr, i32 4, <2 x i1> %mask)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2f32.p0v2f32(<2 x float> %val, <2 x float>* %addr, i32 4, <2 x i1> %mask)
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
   %mask = icmp eq <2 x i32> %trigger, zeroinitializer
@@ -1175,12 +1175,12 @@ define void @test6(<2 x i32> %trigger, <2 x i32>* %addr, <2 x i32> %val) {
 ;
 ; AVX-LABEL: 'test6'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <2 x i32> %trigger, zeroinitializer
-; AVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.store.v2i32.p0v2i32(<2 x i32> %val, <2 x i32>* %addr, i32 4, <2 x i1> %mask)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2i32.p0v2i32(<2 x i32> %val, <2 x i32>* %addr, i32 4, <2 x i1> %mask)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512-LABEL: 'test6'
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <2 x i32> %trigger, zeroinitializer
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.store.v2i32.p0v2i32(<2 x i32> %val, <2 x i32>* %addr, i32 4, <2 x i1> %mask)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2i32.p0v2i32(<2 x i32> %val, <2 x i32>* %addr, i32 4, <2 x i1> %mask)
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
   %mask = icmp eq <2 x i32> %trigger, zeroinitializer
@@ -1201,12 +1201,12 @@ define <2 x float> @test7(<2 x i32> %trigger, <2 x float>* %addr, <2 x float> %d
 ;
 ; AVX-LABEL: 'test7'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <2 x i32> %trigger, zeroinitializer
-; AVX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %res = call <2 x float> @llvm.masked.load.v2f32.p0v2f32(<2 x float>* %addr, i32 4, <2 x i1> %mask, <2 x float> %dst)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %res = call <2 x float> @llvm.masked.load.v2f32.p0v2f32(<2 x float>* %addr, i32 4, <2 x i1> %mask, <2 x float> %dst)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x float> %res
 ;
 ; AVX512-LABEL: 'test7'
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <2 x i32> %trigger, zeroinitializer
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %res = call <2 x float> @llvm.masked.load.v2f32.p0v2f32(<2 x float>* %addr, i32 4, <2 x i1> %mask, <2 x float> %dst)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %res = call <2 x float> @llvm.masked.load.v2f32.p0v2f32(<2 x float>* %addr, i32 4, <2 x i1> %mask, <2 x float> %dst)
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x float> %res
 ;
   %mask = icmp eq <2 x i32> %trigger, zeroinitializer
@@ -1227,12 +1227,12 @@ define <2 x i32> @test8(<2 x i32> %trigger, <2 x i32>* %addr, <2 x i32> %dst) {
 ;
 ; AVX-LABEL: 'test8'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <2 x i32> %trigger, zeroinitializer
-; AVX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %res = call <2 x i32> @llvm.masked.load.v2i32.p0v2i32(<2 x i32>* %addr, i32 4, <2 x i1> %mask, <2 x i32> %dst)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %res = call <2 x i32> @llvm.masked.load.v2i32.p0v2i32(<2 x i32>* %addr, i32 4, <2 x i1> %mask, <2 x i32> %dst)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i32> %res
 ;
 ; AVX512-LABEL: 'test8'
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <2 x i32> %trigger, zeroinitializer
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %res = call <2 x i32> @llvm.masked.load.v2i32.p0v2i32(<2 x i32>* %addr, i32 4, <2 x i1> %mask, <2 x i32> %dst)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %res = call <2 x i32> @llvm.masked.load.v2i32.p0v2i32(<2 x i32>* %addr, i32 4, <2 x i1> %mask, <2 x i32> %dst)
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i32> %res
 ;
   %mask = icmp eq <2 x i32> %trigger, zeroinitializer

diff --git a/llvm/test/Analysis/CostModel/X86/masked-intrinsic-cost.ll b/llvm/test/Analysis/CostModel/X86/masked-intrinsic-cost.ll
index 53fbc19..def3534 100644 (file)
--- a/llvm/test/Analysis/CostModel/X86/masked-intrinsic-cost.ll
+++ b/llvm/test/Analysis/CostModel/X86/masked-intrinsic-cost.ll
@@ -71,7 +71,7 @@ define i32 @masked_load() {
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0v16f32(<16 x float>* undef, i32 1, <16 x i1> undef, <16 x float> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0v8f32(<8 x float>* undef, i32 1, <8 x i1> undef, <8 x float> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0v4f32(<4 x float>* undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0v2f32(<2 x float>* undef, i32 1, <2 x i1> undef, <2 x float> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0v2f32(<2 x float>* undef, i32 1, <2 x i1> undef, <2 x float> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0v8i64(<8 x i64>* undef, i32 1, <8 x i1> undef, <8 x i64> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0v4i64(<4 x i64>* undef, i32 1, <4 x i1> undef, <4 x i64> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0v2i64(<2 x i64>* undef, i32 1, <2 x i1> undef, <2 x i64> undef)
@@ -79,7 +79,7 @@ define i32 @masked_load() {
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0v16i32(<16 x i32>* undef, i32 1, <16 x i1> undef, <16 x i32> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0v8i32(<8 x i32>* undef, i32 1, <8 x i1> undef, <8 x i32> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0v4i32(<4 x i32>* undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0v2i32(<2 x i32>* undef, i32 1, <2 x i1> undef, <2 x i32> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0v2i32(<2 x i32>* undef, i32 1, <2 x i1> undef, <2 x i32> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 148 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0v32i16(<32 x i16>* undef, i32 1, <32 x i1> undef, <32 x i16> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0v16i16(<16 x i16>* undef, i32 1, <16 x i1> undef, <16 x i16> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0v8i16(<8 x i16>* undef, i32 1, <8 x i1> undef, <8 x i16> undef)
@@ -98,7 +98,7 @@ define i32 @masked_load() {
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0v16f32(<16 x float>* undef, i32 1, <16 x i1> undef, <16 x float> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0v8f32(<8 x float>* undef, i32 1, <8 x i1> undef, <8 x float> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0v4f32(<4 x float>* undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0v2f32(<2 x float>* undef, i32 1, <2 x i1> undef, <2 x float> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0v2f32(<2 x float>* undef, i32 1, <2 x i1> undef, <2 x float> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0v8i64(<8 x i64>* undef, i32 1, <8 x i1> undef, <8 x i64> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0v4i64(<4 x i64>* undef, i32 1, <4 x i1> undef, <4 x i64> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0v2i64(<2 x i64>* undef, i32 1, <2 x i1> undef, <2 x i64> undef)
@@ -106,7 +106,7 @@ define i32 @masked_load() {
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0v16i32(<16 x i32>* undef, i32 1, <16 x i1> undef, <16 x i32> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0v8i32(<8 x i32>* undef, i32 1, <8 x i1> undef, <8 x i32> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0v4i32(<4 x i32>* undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0v2i32(<2 x i32>* undef, i32 1, <2 x i1> undef, <2 x i32> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0v2i32(<2 x i32>* undef, i32 1, <2 x i1> undef, <2 x i32> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 148 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0v32i16(<32 x i16>* undef, i32 1, <32 x i1> undef, <32 x i16> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0v16i16(<16 x i16>* undef, i32 1, <16 x i1> undef, <16 x i16> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0v8i16(<8 x i16>* undef, i32 1, <8 x i1> undef, <8 x i16> undef)
@@ -125,7 +125,7 @@ define i32 @masked_load() {
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x float> @llvm.masked.load.v16f32.p0v16f32(<16 x float>* undef, i32 1, <16 x i1> undef, <16 x float> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x float> @llvm.masked.load.v8f32.p0v8f32(<8 x float>* undef, i32 1, <8 x i1> undef, <8 x float> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x float> @llvm.masked.load.v4f32.p0v4f32(<4 x float>* undef, i32 1, <4 x i1> undef, <4 x float> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0v2f32(<2 x float>* undef, i32 1, <2 x i1> undef, <2 x float> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F32 = call <2 x float> @llvm.masked.load.v2f32.p0v2f32(<2 x float>* undef, i32 1, <2 x i1> undef, <2 x float> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = call <8 x i64> @llvm.masked.load.v8i64.p0v8i64(<8 x i64>* undef, i32 1, <8 x i1> undef, <8 x i64> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = call <4 x i64> @llvm.masked.load.v4i64.p0v4i64(<4 x i64>* undef, i32 1, <4 x i1> undef, <4 x i64> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = call <2 x i64> @llvm.masked.load.v2i64.p0v2i64(<2 x i64>* undef, i32 1, <2 x i1> undef, <2 x i64> undef)
@@ -133,15 +133,15 @@ define i32 @masked_load() {
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I32 = call <16 x i32> @llvm.masked.load.v16i32.p0v16i32(<16 x i32>* undef, i32 1, <16 x i1> undef, <16 x i32> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.masked.load.v8i32.p0v8i32(<8 x i32>* undef, i32 1, <8 x i1> undef, <8 x i32> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.masked.load.v4i32.p0v4i32(<4 x i32>* undef, i32 1, <4 x i1> undef, <4 x i32> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0v2i32(<2 x i32>* undef, i32 1, <2 x i1> undef, <2 x i32> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I32 = call <2 x i32> @llvm.masked.load.v2i32.p0v2i32(<2 x i32>* undef, i32 1, <2 x i1> undef, <2 x i32> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I16 = call <32 x i16> @llvm.masked.load.v32i16.p0v32i16(<32 x i16>* undef, i32 1, <32 x i1> undef, <32 x i16> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = call <16 x i16> @llvm.masked.load.v16i16.p0v16i16(<16 x i16>* undef, i32 1, <16 x i1> undef, <16 x i16> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.masked.load.v8i16.p0v8i16(<8 x i16>* undef, i32 1, <8 x i1> undef, <8 x i16> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0v4i16(<4 x i16>* undef, i32 1, <4 x i1> undef, <4 x i16> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I16 = call <4 x i16> @llvm.masked.load.v4i16.p0v4i16(<4 x i16>* undef, i32 1, <4 x i1> undef, <4 x i16> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V64I8 = call <64 x i8> @llvm.masked.load.v64i8.p0v64i8(<64 x i8>* undef, i32 1, <64 x i1> undef, <64 x i8> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V32I8 = call <32 x i8> @llvm.masked.load.v32i8.p0v32i8(<32 x i8>* undef, i32 1, <32 x i1> undef, <32 x i8> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = call <16 x i8> @llvm.masked.load.v16i8.p0v16i8(<16 x i8>* undef, i32 1, <16 x i1> undef, <16 x i8> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0v8i8(<8 x i8>* undef, i32 1, <8 x i1> undef, <8 x i8> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I8 = call <8 x i8> @llvm.masked.load.v8i8.p0v8i8(<8 x i8>* undef, i32 1, <8 x i1> undef, <8 x i8> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
   %V8F64 = call <8 x double> @llvm.masked.load.v8f64.p0v8f64(<8 x double>* undef, i32 1, <8 x i1> undef, <8 x double> undef)
@@ -240,7 +240,7 @@ define i32 @masked_store() {
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v16f32.p0v16f32(<16 x float> undef, <16 x float>* undef, i32 1, <16 x i1> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v8f32.p0v8f32(<8 x float> undef, <8 x float>* undef, i32 1, <8 x i1> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4f32.p0v4f32(<4 x float> undef, <4 x float>* undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.store.v2f32.p0v2f32(<2 x float> undef, <2 x float>* undef, i32 1, <2 x i1> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2f32.p0v2f32(<2 x float> undef, <2 x float>* undef, i32 1, <2 x i1> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v8i64.p0v8i64(<8 x i64> undef, <8 x i64>* undef, i32 1, <8 x i1> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4i64.p0v4i64(<4 x i64> undef, <4 x i64>* undef, i32 1, <4 x i1> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v2i64.p0v2i64(<2 x i64> undef, <2 x i64>* undef, i32 1, <2 x i1> undef)
@@ -248,7 +248,7 @@ define i32 @masked_store() {
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: call void @llvm.masked.store.v16i32.p0v16i32(<16 x i32> undef, <16 x i32>* undef, i32 1, <16 x i1> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v8i32.p0v8i32(<8 x i32> undef, <8 x i32>* undef, i32 1, <8 x i1> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: call void @llvm.masked.store.v4i32.p0v4i32(<4 x i32> undef, <4 x i32>* undef, i32 1, <4 x i1> undef)
-; AVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.store.v2i32.p0v2i32(<2 x i32> undef, <2 x i32>* undef, i32 1, <2 x i1> undef)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2i32.p0v2i32(<2 x i32> undef, <2 x i32>* undef, i32 1, <2 x i1> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: call void @llvm.masked.store.v32i16.p0v32i16(<32 x i16> undef, <32 x i16>* undef, i32 1, <32 x i1> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: call void @llvm.masked.store.v16i16.p0v16i16(<16 x i16> undef, <16 x i16>* undef, i32 1, <16 x i1> undef)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i16.p0v8i16(<8 x i16> undef, <8 x i16>* undef, i32 1, <8 x i1> undef)
@@ -267,7 +267,7 @@ define i32 @masked_store() {
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16f32.p0v16f32(<16 x float> undef, <16 x float>* undef, i32 1, <16 x i1> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f32.p0v8f32(<8 x float> undef, <8 x float>* undef, i32 1, <8 x i1> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f32.p0v4f32(<4 x float> undef, <4 x float>* undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.store.v2f32.p0v2f32(<2 x float> undef, <2 x float>* undef, i32 1, <2 x i1> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2f32.p0v2f32(<2 x float> undef, <2 x float>* undef, i32 1, <2 x i1> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i64.p0v8i64(<8 x i64> undef, <8 x i64>* undef, i32 1, <8 x i1> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i64.p0v4i64(<4 x i64> undef, <4 x i64>* undef, i32 1, <4 x i1> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2i64.p0v2i64(<2 x i64> undef, <2 x i64>* undef, i32 1, <2 x i1> undef)
@@ -275,7 +275,7 @@ define i32 @masked_store() {
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i32.p0v16i32(<16 x i32> undef, <16 x i32>* undef, i32 1, <16 x i1> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i32.p0v8i32(<8 x i32> undef, <8 x i32>* undef, i32 1, <8 x i1> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i32.p0v4i32(<4 x i32> undef, <4 x i32>* undef, i32 1, <4 x i1> undef)
-; KNL-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.store.v2i32.p0v2i32(<2 x i32> undef, <2 x i32>* undef, i32 1, <2 x i1> undef)
+; KNL-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2i32.p0v2i32(<2 x i32> undef, <2 x i32>* undef, i32 1, <2 x i1> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 168 for instruction: call void @llvm.masked.store.v32i16.p0v32i16(<32 x i16> undef, <32 x i16>* undef, i32 1, <32 x i1> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: call void @llvm.masked.store.v16i16.p0v16i16(<16 x i16> undef, <16 x i16>* undef, i32 1, <16 x i1> undef)
 ; KNL-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: call void @llvm.masked.store.v8i16.p0v8i16(<8 x i16> undef, <8 x i16>* undef, i32 1, <8 x i1> undef)
@@ -294,7 +294,7 @@ define i32 @masked_store() {
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16f32.p0v16f32(<16 x float> undef, <16 x float>* undef, i32 1, <16 x i1> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8f32.p0v8f32(<8 x float> undef, <8 x float>* undef, i32 1, <8 x i1> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4f32.p0v4f32(<4 x float> undef, <4 x float>* undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.store.v2f32.p0v2f32(<2 x float> undef, <2 x float>* undef, i32 1, <2 x i1> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2f32.p0v2f32(<2 x float> undef, <2 x float>* undef, i32 1, <2 x i1> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i64.p0v8i64(<8 x i64> undef, <8 x i64>* undef, i32 1, <8 x i1> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i64.p0v4i64(<4 x i64> undef, <4 x i64>* undef, i32 1, <4 x i1> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v2i64.p0v2i64(<2 x i64> undef, <2 x i64>* undef, i32 1, <2 x i1> undef)
@@ -302,15 +302,15 @@ define i32 @masked_store() {
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i32.p0v16i32(<16 x i32> undef, <16 x i32>* undef, i32 1, <16 x i1> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i32.p0v8i32(<8 x i32> undef, <8 x i32>* undef, i32 1, <8 x i1> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v4i32.p0v4i32(<4 x i32> undef, <4 x i32>* undef, i32 1, <4 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.store.v2i32.p0v2i32(<2 x i32> undef, <2 x i32>* undef, i32 1, <2 x i1> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2i32.p0v2i32(<2 x i32> undef, <2 x i32>* undef, i32 1, <2 x i1> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v32i16.p0v32i16(<32 x i16> undef, <32 x i16>* undef, i32 1, <32 x i1> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i16.p0v16i16(<16 x i16> undef, <16 x i16>* undef, i32 1, <16 x i1> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v8i16.p0v8i16(<8 x i16> undef, <8 x i16>* undef, i32 1, <8 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v4i16.p0v4i16(<4 x i16> undef, <4 x i16>* undef, i32 1, <4 x i1> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v4i16.p0v4i16(<4 x i16> undef, <4 x i16>* undef, i32 1, <4 x i1> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v64i8.p0v64i8(<64 x i8> undef, <64 x i8>* undef, i32 1, <64 x i1> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v32i8.p0v32i8(<32 x i8> undef, <32 x i8>* undef, i32 1, <32 x i1> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.masked.store.v16i8.p0v16i8(<16 x i8> undef, <16 x i8>* undef, i32 1, <16 x i1> undef)
-; SKX-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: call void @llvm.masked.store.v8i8.p0v8i8(<8 x i8> undef, <8 x i8>* undef, i32 1, <8 x i1> undef)
+; SKX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v8i8.p0v8i8(<8 x i8> undef, <8 x i8>* undef, i32 1, <8 x i1> undef)
 ; SKX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 0
 ;
   call void @llvm.masked.store.v8f64.p0v8f64(<8 x double> undef, <8 x double>* undef, i32 1, <8 x i1> undef)
@@ -1149,12 +1149,12 @@ define void @test5(<2 x i32> %trigger, <2 x float>* %addr, <2 x float> %val) {
 ;
 ; AVX-LABEL: 'test5'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <2 x i32> %trigger, zeroinitializer
-; AVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.store.v2f32.p0v2f32(<2 x float> %val, <2 x float>* %addr, i32 4, <2 x i1> %mask)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2f32.p0v2f32(<2 x float> %val, <2 x float>* %addr, i32 4, <2 x i1> %mask)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512-LABEL: 'test5'
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <2 x i32> %trigger, zeroinitializer
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.store.v2f32.p0v2f32(<2 x float> %val, <2 x float>* %addr, i32 4, <2 x i1> %mask)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2f32.p0v2f32(<2 x float> %val, <2 x float>* %addr, i32 4, <2 x i1> %mask)
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
   %mask = icmp eq <2 x i32> %trigger, zeroinitializer
@@ -1175,12 +1175,12 @@ define void @test6(<2 x i32> %trigger, <2 x i32>* %addr, <2 x i32> %val) {
 ;
 ; AVX-LABEL: 'test6'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <2 x i32> %trigger, zeroinitializer
-; AVX-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: call void @llvm.masked.store.v2i32.p0v2i32(<2 x i32> %val, <2 x i32>* %addr, i32 4, <2 x i1> %mask)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: call void @llvm.masked.store.v2i32.p0v2i32(<2 x i32> %val, <2 x i32>* %addr, i32 4, <2 x i1> %mask)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512-LABEL: 'test6'
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <2 x i32> %trigger, zeroinitializer
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: call void @llvm.masked.store.v2i32.p0v2i32(<2 x i32> %val, <2 x i32>* %addr, i32 4, <2 x i1> %mask)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: call void @llvm.masked.store.v2i32.p0v2i32(<2 x i32> %val, <2 x i32>* %addr, i32 4, <2 x i1> %mask)
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
   %mask = icmp eq <2 x i32> %trigger, zeroinitializer
@@ -1201,12 +1201,12 @@ define <2 x float> @test7(<2 x i32> %trigger, <2 x float>* %addr, <2 x float> %d
 ;
 ; AVX-LABEL: 'test7'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <2 x i32> %trigger, zeroinitializer
-; AVX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %res = call <2 x float> @llvm.masked.load.v2f32.p0v2f32(<2 x float>* %addr, i32 4, <2 x i1> %mask, <2 x float> %dst)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %res = call <2 x float> @llvm.masked.load.v2f32.p0v2f32(<2 x float>* %addr, i32 4, <2 x i1> %mask, <2 x float> %dst)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x float> %res
 ;
 ; AVX512-LABEL: 'test7'
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <2 x i32> %trigger, zeroinitializer
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %res = call <2 x float> @llvm.masked.load.v2f32.p0v2f32(<2 x float>* %addr, i32 4, <2 x i1> %mask, <2 x float> %dst)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %res = call <2 x float> @llvm.masked.load.v2f32.p0v2f32(<2 x float>* %addr, i32 4, <2 x i1> %mask, <2 x float> %dst)
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x float> %res
 ;
   %mask = icmp eq <2 x i32> %trigger, zeroinitializer
@@ -1227,12 +1227,12 @@ define <2 x i32> @test8(<2 x i32> %trigger, <2 x i32>* %addr, <2 x i32> %dst) {
 ;
 ; AVX-LABEL: 'test8'
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <2 x i32> %trigger, zeroinitializer
-; AVX-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %res = call <2 x i32> @llvm.masked.load.v2i32.p0v2i32(<2 x i32>* %addr, i32 4, <2 x i1> %mask, <2 x i32> %dst)
+; AVX-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %res = call <2 x i32> @llvm.masked.load.v2i32.p0v2i32(<2 x i32>* %addr, i32 4, <2 x i1> %mask, <2 x i32> %dst)
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i32> %res
 ;
 ; AVX512-LABEL: 'test8'
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %mask = icmp eq <2 x i32> %trigger, zeroinitializer
-; AVX512-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %res = call <2 x i32> @llvm.masked.load.v2i32.p0v2i32(<2 x i32>* %addr, i32 4, <2 x i1> %mask, <2 x i32> %dst)
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %res = call <2 x i32> @llvm.masked.load.v2i32.p0v2i32(<2 x i32>* %addr, i32 4, <2 x i1> %mask, <2 x i32> %dst)
 ; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i32> %res
 ;
   %mask = icmp eq <2 x i32> %trigger, zeroinitializer

diff --git a/llvm/test/Transforms/LoopVectorize/X86/gather_scatter.ll b/llvm/test/Transforms/LoopVectorize/X86/gather_scatter.ll
index 9584dd7..23b3e1e 100644 (file)
--- a/llvm/test/Transforms/LoopVectorize/X86/gather_scatter.ll
+++ b/llvm/test/Transforms/LoopVectorize/X86/gather_scatter.ll
@@ -95,39 +95,66 @@ define void @foo1(float* noalias %in, float* noalias %out, i32* noalias %trigger
 ; FVW2-NEXT:  entry:
 ; FVW2-NEXT:    br label [[VECTOR_BODY:%.*]]
 ; FVW2:       vector.body:
-; FVW2-NEXT:    [[INDEX6:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT_1:%.*]], [[VECTOR_BODY]] ]
+; FVW2-NEXT:    [[INDEX6:%.*]] = phi i64 [ 0, [[ENTRY:%.*]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
 ; FVW2-NEXT:    [[TMP0:%.*]] = getelementptr inbounds i32, i32* [[TRIGGER:%.*]], i64 [[INDEX6]]
 ; FVW2-NEXT:    [[TMP1:%.*]] = bitcast i32* [[TMP0]] to <2 x i32>*
 ; FVW2-NEXT:    [[WIDE_LOAD:%.*]] = load <2 x i32>, <2 x i32>* [[TMP1]], align 4
-; FVW2-NEXT:    [[TMP2:%.*]] = icmp sgt <2 x i32> [[WIDE_LOAD]], zeroinitializer
-; FVW2-NEXT:    [[TMP3:%.*]] = getelementptr inbounds i32, i32* [[INDEX:%.*]], i64 [[INDEX6]]
-; FVW2-NEXT:    [[TMP4:%.*]] = bitcast i32* [[TMP3]] to <2 x i32>*
-; FVW2-NEXT:    [[WIDE_MASKED_LOAD:%.*]] = call <2 x i32> @llvm.masked.load.v2i32.p0v2i32(<2 x i32>* [[TMP4]], i32 4, <2 x i1> [[TMP2]], <2 x i32> poison)
-; FVW2-NEXT:    [[TMP5:%.*]] = sext <2 x i32> [[WIDE_MASKED_LOAD]] to <2 x i64>
-; FVW2-NEXT:    [[TMP6:%.*]] = getelementptr inbounds float, float* [[IN:%.*]], <2 x i64> [[TMP5]]
-; FVW2-NEXT:    [[WIDE_MASKED_GATHER:%.*]] = call <2 x float> @llvm.masked.gather.v2f32.v2p0f32(<2 x float*> [[TMP6]], i32 4, <2 x i1> [[TMP2]], <2 x float> undef)
-; FVW2-NEXT:    [[TMP7:%.*]] = fadd <2 x float> [[WIDE_MASKED_GATHER]], <float 5.000000e-01, float 5.000000e-01>
-; FVW2-NEXT:    [[TMP8:%.*]] = getelementptr inbounds float, float* [[OUT:%.*]], i64 [[INDEX6]]
-; FVW2-NEXT:    [[TMP9:%.*]] = bitcast float* [[TMP8]] to <2 x float>*
-; FVW2-NEXT:    call void @llvm.masked.store.v2f32.p0v2f32(<2 x float> [[TMP7]], <2 x float>* [[TMP9]], i32 4, <2 x i1> [[TMP2]])
-; FVW2-NEXT:    [[INDEX_NEXT:%.*]] = or i64 [[INDEX6]], 2
-; FVW2-NEXT:    [[TMP10:%.*]] = getelementptr inbounds i32, i32* [[TRIGGER]], i64 [[INDEX_NEXT]]
-; FVW2-NEXT:    [[TMP11:%.*]] = bitcast i32* [[TMP10]] to <2 x i32>*
-; FVW2-NEXT:    [[WIDE_LOAD_1:%.*]] = load <2 x i32>, <2 x i32>* [[TMP11]], align 4
-; FVW2-NEXT:    [[TMP12:%.*]] = icmp sgt <2 x i32> [[WIDE_LOAD_1]], zeroinitializer
-; FVW2-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i32, i32* [[INDEX]], i64 [[INDEX_NEXT]]
-; FVW2-NEXT:    [[TMP14:%.*]] = bitcast i32* [[TMP13]] to <2 x i32>*
-; FVW2-NEXT:    [[WIDE_MASKED_LOAD_1:%.*]] = call <2 x i32> @llvm.masked.load.v2i32.p0v2i32(<2 x i32>* nonnull [[TMP14]], i32 4, <2 x i1> [[TMP12]], <2 x i32> poison)
-; FVW2-NEXT:    [[TMP15:%.*]] = sext <2 x i32> [[WIDE_MASKED_LOAD_1]] to <2 x i64>
-; FVW2-NEXT:    [[TMP16:%.*]] = getelementptr inbounds float, float* [[IN]], <2 x i64> [[TMP15]]
-; FVW2-NEXT:    [[WIDE_MASKED_GATHER_1:%.*]] = call <2 x float> @llvm.masked.gather.v2f32.v2p0f32(<2 x float*> [[TMP16]], i32 4, <2 x i1> [[TMP12]], <2 x float> undef)
-; FVW2-NEXT:    [[TMP17:%.*]] = fadd <2 x float> [[WIDE_MASKED_GATHER_1]], <float 5.000000e-01, float 5.000000e-01>
-; FVW2-NEXT:    [[TMP18:%.*]] = getelementptr inbounds float, float* [[OUT]], i64 [[INDEX_NEXT]]
-; FVW2-NEXT:    [[TMP19:%.*]] = bitcast float* [[TMP18]] to <2 x float>*
-; FVW2-NEXT:    call void @llvm.masked.store.v2f32.p0v2f32(<2 x float> [[TMP17]], <2 x float>* [[TMP19]], i32 4, <2 x i1> [[TMP12]])
-; FVW2-NEXT:    [[INDEX_NEXT_1]] = add nuw nsw i64 [[INDEX6]], 4
-; FVW2-NEXT:    [[TMP20:%.*]] = icmp eq i64 [[INDEX_NEXT_1]], 4096
-; FVW2-NEXT:    br i1 [[TMP20]], label [[FOR_END:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; FVW2-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i32, i32* [[TMP0]], i64 2
+; FVW2-NEXT:    [[TMP3:%.*]] = bitcast i32* [[TMP2]] to <2 x i32>*
+; FVW2-NEXT:    [[WIDE_LOAD7:%.*]] = load <2 x i32>, <2 x i32>* [[TMP3]], align 4
+; FVW2-NEXT:    [[TMP4:%.*]] = getelementptr inbounds i32, i32* [[TMP0]], i64 4
+; FVW2-NEXT:    [[TMP5:%.*]] = bitcast i32* [[TMP4]] to <2 x i32>*
+; FVW2-NEXT:    [[WIDE_LOAD8:%.*]] = load <2 x i32>, <2 x i32>* [[TMP5]], align 4
+; FVW2-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i32, i32* [[TMP0]], i64 6
+; FVW2-NEXT:    [[TMP7:%.*]] = bitcast i32* [[TMP6]] to <2 x i32>*
+; FVW2-NEXT:    [[WIDE_LOAD9:%.*]] = load <2 x i32>, <2 x i32>* [[TMP7]], align 4
+; FVW2-NEXT:    [[TMP8:%.*]] = icmp sgt <2 x i32> [[WIDE_LOAD]], zeroinitializer
+; FVW2-NEXT:    [[TMP9:%.*]] = icmp sgt <2 x i32> [[WIDE_LOAD7]], zeroinitializer
+; FVW2-NEXT:    [[TMP10:%.*]] = icmp sgt <2 x i32> [[WIDE_LOAD8]], zeroinitializer
+; FVW2-NEXT:    [[TMP11:%.*]] = icmp sgt <2 x i32> [[WIDE_LOAD9]], zeroinitializer
+; FVW2-NEXT:    [[TMP12:%.*]] = getelementptr inbounds i32, i32* [[INDEX:%.*]], i64 [[INDEX6]]
+; FVW2-NEXT:    [[TMP13:%.*]] = bitcast i32* [[TMP12]] to <2 x i32>*
+; FVW2-NEXT:    [[WIDE_MASKED_LOAD:%.*]] = call <2 x i32> @llvm.masked.load.v2i32.p0v2i32(<2 x i32>* [[TMP13]], i32 4, <2 x i1> [[TMP8]], <2 x i32> poison)
+; FVW2-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i32, i32* [[TMP12]], i64 2
+; FVW2-NEXT:    [[TMP15:%.*]] = bitcast i32* [[TMP14]] to <2 x i32>*
+; FVW2-NEXT:    [[WIDE_MASKED_LOAD10:%.*]] = call <2 x i32> @llvm.masked.load.v2i32.p0v2i32(<2 x i32>* nonnull [[TMP15]], i32 4, <2 x i1> [[TMP9]], <2 x i32> poison)
+; FVW2-NEXT:    [[TMP16:%.*]] = getelementptr inbounds i32, i32* [[TMP12]], i64 4
+; FVW2-NEXT:    [[TMP17:%.*]] = bitcast i32* [[TMP16]] to <2 x i32>*
+; FVW2-NEXT:    [[WIDE_MASKED_LOAD11:%.*]] = call <2 x i32> @llvm.masked.load.v2i32.p0v2i32(<2 x i32>* nonnull [[TMP17]], i32 4, <2 x i1> [[TMP10]], <2 x i32> poison)
+; FVW2-NEXT:    [[TMP18:%.*]] = getelementptr inbounds i32, i32* [[TMP12]], i64 6
+; FVW2-NEXT:    [[TMP19:%.*]] = bitcast i32* [[TMP18]] to <2 x i32>*
+; FVW2-NEXT:    [[WIDE_MASKED_LOAD12:%.*]] = call <2 x i32> @llvm.masked.load.v2i32.p0v2i32(<2 x i32>* nonnull [[TMP19]], i32 4, <2 x i1> [[TMP11]], <2 x i32> poison)
+; FVW2-NEXT:    [[TMP20:%.*]] = sext <2 x i32> [[WIDE_MASKED_LOAD]] to <2 x i64>
+; FVW2-NEXT:    [[TMP21:%.*]] = sext <2 x i32> [[WIDE_MASKED_LOAD10]] to <2 x i64>
+; FVW2-NEXT:    [[TMP22:%.*]] = sext <2 x i32> [[WIDE_MASKED_LOAD11]] to <2 x i64>
+; FVW2-NEXT:    [[TMP23:%.*]] = sext <2 x i32> [[WIDE_MASKED_LOAD12]] to <2 x i64>
+; FVW2-NEXT:    [[TMP24:%.*]] = getelementptr inbounds float, float* [[IN:%.*]], <2 x i64> [[TMP20]]
+; FVW2-NEXT:    [[TMP25:%.*]] = getelementptr inbounds float, float* [[IN]], <2 x i64> [[TMP21]]
+; FVW2-NEXT:    [[TMP26:%.*]] = getelementptr inbounds float, float* [[IN]], <2 x i64> [[TMP22]]
+; FVW2-NEXT:    [[TMP27:%.*]] = getelementptr inbounds float, float* [[IN]], <2 x i64> [[TMP23]]
+; FVW2-NEXT:    [[WIDE_MASKED_GATHER:%.*]] = call <2 x float> @llvm.masked.gather.v2f32.v2p0f32(<2 x float*> [[TMP24]], i32 4, <2 x i1> [[TMP8]], <2 x float> undef)
+; FVW2-NEXT:    [[WIDE_MASKED_GATHER13:%.*]] = call <2 x float> @llvm.masked.gather.v2f32.v2p0f32(<2 x float*> [[TMP25]], i32 4, <2 x i1> [[TMP9]], <2 x float> undef)
+; FVW2-NEXT:    [[WIDE_MASKED_GATHER14:%.*]] = call <2 x float> @llvm.masked.gather.v2f32.v2p0f32(<2 x float*> [[TMP26]], i32 4, <2 x i1> [[TMP10]], <2 x float> undef)
+; FVW2-NEXT:    [[WIDE_MASKED_GATHER15:%.*]] = call <2 x float> @llvm.masked.gather.v2f32.v2p0f32(<2 x float*> [[TMP27]], i32 4, <2 x i1> [[TMP11]], <2 x float> undef)
+; FVW2-NEXT:    [[TMP28:%.*]] = fadd <2 x float> [[WIDE_MASKED_GATHER]], <float 5.000000e-01, float 5.000000e-01>
+; FVW2-NEXT:    [[TMP29:%.*]] = fadd <2 x float> [[WIDE_MASKED_GATHER13]], <float 5.000000e-01, float 5.000000e-01>
+; FVW2-NEXT:    [[TMP30:%.*]] = fadd <2 x float> [[WIDE_MASKED_GATHER14]], <float 5.000000e-01, float 5.000000e-01>
+; FVW2-NEXT:    [[TMP31:%.*]] = fadd <2 x float> [[WIDE_MASKED_GATHER15]], <float 5.000000e-01, float 5.000000e-01>
+; FVW2-NEXT:    [[TMP32:%.*]] = getelementptr inbounds float, float* [[OUT:%.*]], i64 [[INDEX6]]
+; FVW2-NEXT:    [[TMP33:%.*]] = bitcast float* [[TMP32]] to <2 x float>*
+; FVW2-NEXT:    call void @llvm.masked.store.v2f32.p0v2f32(<2 x float> [[TMP28]], <2 x float>* [[TMP33]], i32 4, <2 x i1> [[TMP8]])
+; FVW2-NEXT:    [[TMP34:%.*]] = getelementptr inbounds float, float* [[TMP32]], i64 2
+; FVW2-NEXT:    [[TMP35:%.*]] = bitcast float* [[TMP34]] to <2 x float>*
+; FVW2-NEXT:    call void @llvm.masked.store.v2f32.p0v2f32(<2 x float> [[TMP29]], <2 x float>* [[TMP35]], i32 4, <2 x i1> [[TMP9]])
+; FVW2-NEXT:    [[TMP36:%.*]] = getelementptr inbounds float, float* [[TMP32]], i64 4
+; FVW2-NEXT:    [[TMP37:%.*]] = bitcast float* [[TMP36]] to <2 x float>*
+; FVW2-NEXT:    call void @llvm.masked.store.v2f32.p0v2f32(<2 x float> [[TMP30]], <2 x float>* [[TMP37]], i32 4, <2 x i1> [[TMP10]])
+; FVW2-NEXT:    [[TMP38:%.*]] = getelementptr inbounds float, float* [[TMP32]], i64 6
+; FVW2-NEXT:    [[TMP39:%.*]] = bitcast float* [[TMP38]] to <2 x float>*
+; FVW2-NEXT:    call void @llvm.masked.store.v2f32.p0v2f32(<2 x float> [[TMP31]], <2 x float>* [[TMP39]], i32 4, <2 x i1> [[TMP11]])
+; FVW2-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX6]], 8
+; FVW2-NEXT:    [[TMP40:%.*]] = icmp eq i64 [[INDEX_NEXT]], 4096
+; FVW2-NEXT:    br i1 [[TMP40]], label [[FOR_END:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
 ; FVW2:       for.end:
 ; FVW2-NEXT:    ret void
 ;