[llvm] 0776924 - [CostModel][X86] getCmpSelInstrCost - treat BAD_PREDICATEs the same as the worst case cost predicates for ICMP/FCMP instructions

Wed Oct 6 02:15:03 PDT 2021

Author: Simon Pilgrim
Date: 2021-10-06T10:14:56+01:00
New Revision: 0776924a17efaf5269f154f899ecb5474e7800f2

URL: https://github.com/llvm/llvm-project/commit/0776924a17efaf5269f154f899ecb5474e7800f2
DIFF: https://github.com/llvm/llvm-project/commit/0776924a17efaf5269f154f899ecb5474e7800f2.diff

LOG: [CostModel][X86] getCmpSelInstrCost - treat BAD_PREDICATEs the same as the worst case cost predicates for ICMP/FCMP instructions

As suggested on D111024, we should treat getCmpSelInstrCost calls without a specific predicate as matching the worst case predicate cost.

These regressions will be addressed with a mixture of D111024 and fixing other specific getCmpSelInstrCost calls to have realistic predicates.

Added: 
    

Modified: 
    llvm/lib/Target/X86/X86TargetTransformInfo.cpp
    llvm/test/Analysis/CostModel/X86/arith-overflow.ll
    llvm/test/Analysis/CostModel/X86/arith-sminmax.ll
    llvm/test/Analysis/CostModel/X86/arith-ssat.ll
    llvm/test/Analysis/CostModel/X86/arith-uminmax.ll
    llvm/test/Analysis/CostModel/X86/arith-usat.ll
    llvm/test/Analysis/CostModel/X86/fshl.ll
    llvm/test/Analysis/CostModel/X86/fshr.ll
    llvm/test/Analysis/CostModel/X86/intrinsic-cost-kinds.ll
    llvm/test/Analysis/CostModel/X86/reduce-smax.ll
    llvm/test/Analysis/CostModel/X86/reduce-smin.ll
    llvm/test/Analysis/CostModel/X86/reduce-umax.ll
    llvm/test/Analysis/CostModel/X86/reduce-umin.ll
    llvm/test/Transforms/SLPVectorizer/X86/horizontal-minmax.ll
    llvm/test/Transforms/SLPVectorizer/X86/horizontal-smax.ll
    llvm/test/Transforms/VectorCombine/X86/extract-cmp-binop.ll
    llvm/test/Transforms/VectorCombine/X86/extract-cmp.ll
    llvm/test/Transforms/VectorCombine/X86/scalarize-cmp-inseltpoison.ll
    llvm/test/Transforms/VectorCombine/X86/scalarize-cmp.ll

Removed: 
    


################################################################################
diff  --git a/llvm/lib/Target/X86/X86TargetTransformInfo.cpp b/llvm/lib/Target/X86/X86TargetTransformInfo.cpp
index d240333bde27a..844676df2ae54 100644

--- a/llvm/lib/Target/X86/X86TargetTransformInfo.cpp
+++ b/llvm/lib/Target/X86/X86TargetTransformInfo.cpp
@@ -2462,6 +2462,11 @@ InstructionCost X86TTIImpl::getCmpSelInstrCost(unsigned Opcode, Type *ValTy,
           ExtraCost = 3;
         }
         break;
+      case CmpInst::Predicate::BAD_ICMP_PREDICATE:
+      case CmpInst::Predicate::BAD_FCMP_PREDICATE:
+        // Assume worst case scenario and add the maximum extra cost.
+        ExtraCost = 3;
+        break;
       default:
         break;
       }

diff  --git a/llvm/test/Analysis/CostModel/X86/arith-overflow.ll b/llvm/test/Analysis/CostModel/X86/arith-overflow.ll
index 8418e8c9fde5c..4bc6b85625c75 100644
--- a/llvm/test/Analysis/CostModel/X86/arith-overflow.ll
+++ b/llvm/test/Analysis/CostModel/X86/arith-overflow.ll
@@ -276,78 +276,78 @@ declare {<64 x i8>, <64 x i1>}  @llvm.uadd.with.overflow.v64i8(<64 x i8>, <64 x
 define i32 @uadd(i32 %arg) {
 ; SSSE3-LABEL: 'uadd'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = call { i64, i1 } @llvm.uadd.with.overflow.i64(i64 undef, i64 undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.uadd.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.uadd.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.uadd.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.uadd.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.uadd.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.uadd.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = call { i32, i1 } @llvm.uadd.with.overflow.i32(i32 undef, i32 undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.uadd.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.uadd.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.uadd.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.uadd.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.uadd.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.uadd.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = call { i16, i1 } @llvm.uadd.with.overflow.i16(i16 undef, i16 undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.uadd.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.uadd.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.uadd.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.uadd.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.uadd.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.uadd.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = call { i8, i1 } @llvm.uadd.with.overflow.i8(i8 undef, i8 undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.uadd.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.uadd.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.uadd.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.uadd.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.uadd.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.uadd.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'uadd'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = call { i64, i1 } @llvm.uadd.with.overflow.i64(i64 undef, i64 undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.uadd.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.uadd.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.uadd.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.uadd.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.uadd.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.uadd.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = call { i32, i1 } @llvm.uadd.with.overflow.i32(i32 undef, i32 undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.uadd.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.uadd.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.uadd.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.uadd.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.uadd.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.uadd.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = call { i16, i1 } @llvm.uadd.with.overflow.i16(i16 undef, i16 undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.uadd.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.uadd.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.uadd.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.uadd.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.uadd.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.uadd.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = call { i8, i1 } @llvm.uadd.with.overflow.i8(i8 undef, i8 undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.uadd.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.uadd.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.uadd.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.uadd.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.uadd.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.uadd.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX1-LABEL: 'uadd'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = call { i64, i1 } @llvm.uadd.with.overflow.i64(i64 undef, i64 undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.uadd.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.uadd.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.uadd.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.uadd.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.uadd.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.uadd.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = call { i32, i1 } @llvm.uadd.with.overflow.i32(i32 undef, i32 undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.uadd.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.uadd.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.uadd.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.uadd.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.uadd.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.uadd.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = call { i16, i1 } @llvm.uadd.with.overflow.i16(i16 undef, i16 undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.uadd.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.uadd.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.uadd.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.uadd.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.uadd.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.uadd.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = call { i8, i1 } @llvm.uadd.with.overflow.i8(i8 undef, i8 undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.uadd.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.uadd.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.uadd.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.uadd.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.uadd.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.uadd.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'uadd'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = call { i64, i1 } @llvm.uadd.with.overflow.i64(i64 undef, i64 undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.uadd.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.uadd.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.uadd.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.uadd.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.uadd.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.uadd.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = call { i32, i1 } @llvm.uadd.with.overflow.i32(i32 undef, i32 undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.uadd.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.uadd.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.uadd.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.uadd.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.uadd.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.uadd.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = call { i16, i1 } @llvm.uadd.with.overflow.i16(i16 undef, i16 undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.uadd.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.uadd.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.uadd.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.uadd.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.uadd.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.uadd.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = call { i8, i1 } @llvm.uadd.with.overflow.i8(i8 undef, i8 undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.uadd.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.uadd.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.uadd.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.uadd.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.uadd.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.uadd.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512F-LABEL: 'uadd'
@@ -360,13 +360,13 @@ define i32 @uadd(i32 %arg) {
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.uadd.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.uadd.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = call { i16, i1 } @llvm.uadd.with.overflow.i16(i16 undef, i16 undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.uadd.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.uadd.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.uadd.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.uadd.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.uadd.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.uadd.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = call { i8, i1 } @llvm.uadd.with.overflow.i8(i8 undef, i8 undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.uadd.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.uadd.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.uadd.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.uadd.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.uadd.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.uadd.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512BW-LABEL: 'uadd'
@@ -398,70 +398,70 @@ define i32 @uadd(i32 %arg) {
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.uadd.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.uadd.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = call { i16, i1 } @llvm.uadd.with.overflow.i16(i16 undef, i16 undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.uadd.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.uadd.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.uadd.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.uadd.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.uadd.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.uadd.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = call { i8, i1 } @llvm.uadd.with.overflow.i8(i8 undef, i8 undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.uadd.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.uadd.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.uadd.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.uadd.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.uadd.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.uadd.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SLM-LABEL: 'uadd'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = call { i64, i1 } @llvm.uadd.with.overflow.i64(i64 undef, i64 undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.uadd.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.uadd.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.uadd.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.uadd.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.uadd.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.uadd.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = call { i32, i1 } @llvm.uadd.with.overflow.i32(i32 undef, i32 undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.uadd.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.uadd.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.uadd.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.uadd.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.uadd.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.uadd.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = call { i16, i1 } @llvm.uadd.with.overflow.i16(i16 undef, i16 undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.uadd.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.uadd.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.uadd.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.uadd.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.uadd.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.uadd.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = call { i8, i1 } @llvm.uadd.with.overflow.i8(i8 undef, i8 undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.uadd.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.uadd.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.uadd.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.uadd.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.uadd.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.uadd.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; GLM-LABEL: 'uadd'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = call { i64, i1 } @llvm.uadd.with.overflow.i64(i64 undef, i64 undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.uadd.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.uadd.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.uadd.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.uadd.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.uadd.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.uadd.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = call { i32, i1 } @llvm.uadd.with.overflow.i32(i32 undef, i32 undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.uadd.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.uadd.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.uadd.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.uadd.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.uadd.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.uadd.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = call { i16, i1 } @llvm.uadd.with.overflow.i16(i16 undef, i16 undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.uadd.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.uadd.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.uadd.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.uadd.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.uadd.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.uadd.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = call { i8, i1 } @llvm.uadd.with.overflow.i8(i8 undef, i8 undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.uadd.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.uadd.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.uadd.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.uadd.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.uadd.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.uadd.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; BTVER2-LABEL: 'uadd'
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = call { i64, i1 } @llvm.uadd.with.overflow.i64(i64 undef, i64 undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.uadd.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.uadd.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.uadd.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.uadd.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.uadd.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.uadd.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = call { i32, i1 } @llvm.uadd.with.overflow.i32(i32 undef, i32 undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.uadd.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.uadd.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.uadd.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.uadd.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.uadd.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.uadd.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = call { i16, i1 } @llvm.uadd.with.overflow.i16(i16 undef, i16 undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.uadd.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.uadd.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.uadd.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.uadd.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.uadd.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.uadd.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = call { i8, i1 } @llvm.uadd.with.overflow.i8(i8 undef, i8 undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.uadd.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.uadd.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.uadd.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.uadd.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.uadd.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.uadd.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
   %I64 = call {i64, i1} @llvm.uadd.with.overflow.i64(i64 undef, i64 undef)
@@ -752,78 +752,78 @@ declare {<64 x i8>, <64 x i1>}  @llvm.usub.with.overflow.v64i8(<64 x i8>, <64 x
 define i32 @usub(i32 %arg) {
 ; SSSE3-LABEL: 'usub'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = call { i64, i1 } @llvm.usub.with.overflow.i64(i64 undef, i64 undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.usub.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.usub.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.usub.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.usub.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.usub.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.usub.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = call { i32, i1 } @llvm.usub.with.overflow.i32(i32 undef, i32 undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.usub.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.usub.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.usub.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.usub.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.usub.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.usub.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = call { i16, i1 } @llvm.usub.with.overflow.i16(i16 undef, i16 undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.usub.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.usub.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.usub.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.usub.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.usub.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.usub.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = call { i8, i1 } @llvm.usub.with.overflow.i8(i8 undef, i8 undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.usub.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.usub.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.usub.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.usub.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.usub.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.usub.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'usub'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = call { i64, i1 } @llvm.usub.with.overflow.i64(i64 undef, i64 undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.usub.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.usub.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.usub.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.usub.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.usub.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.usub.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = call { i32, i1 } @llvm.usub.with.overflow.i32(i32 undef, i32 undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.usub.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.usub.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.usub.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.usub.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.usub.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.usub.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = call { i16, i1 } @llvm.usub.with.overflow.i16(i16 undef, i16 undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.usub.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.usub.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.usub.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.usub.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.usub.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.usub.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = call { i8, i1 } @llvm.usub.with.overflow.i8(i8 undef, i8 undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.usub.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.usub.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.usub.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.usub.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.usub.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.usub.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX1-LABEL: 'usub'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = call { i64, i1 } @llvm.usub.with.overflow.i64(i64 undef, i64 undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.usub.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.usub.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.usub.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.usub.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.usub.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.usub.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = call { i32, i1 } @llvm.usub.with.overflow.i32(i32 undef, i32 undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.usub.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.usub.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.usub.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.usub.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.usub.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.usub.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = call { i16, i1 } @llvm.usub.with.overflow.i16(i16 undef, i16 undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.usub.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.usub.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.usub.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.usub.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.usub.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.usub.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = call { i8, i1 } @llvm.usub.with.overflow.i8(i8 undef, i8 undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.usub.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.usub.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.usub.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.usub.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.usub.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.usub.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'usub'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = call { i64, i1 } @llvm.usub.with.overflow.i64(i64 undef, i64 undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.usub.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.usub.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.usub.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.usub.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.usub.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.usub.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = call { i32, i1 } @llvm.usub.with.overflow.i32(i32 undef, i32 undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.usub.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.usub.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.usub.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.usub.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.usub.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.usub.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = call { i16, i1 } @llvm.usub.with.overflow.i16(i16 undef, i16 undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.usub.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.usub.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.usub.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.usub.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.usub.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.usub.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = call { i8, i1 } @llvm.usub.with.overflow.i8(i8 undef, i8 undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.usub.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.usub.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.usub.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.usub.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.usub.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.usub.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512F-LABEL: 'usub'
@@ -836,13 +836,13 @@ define i32 @usub(i32 %arg) {
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.usub.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.usub.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = call { i16, i1 } @llvm.usub.with.overflow.i16(i16 undef, i16 undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.usub.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.usub.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.usub.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.usub.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.usub.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.usub.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = call { i8, i1 } @llvm.usub.with.overflow.i8(i8 undef, i8 undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.usub.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.usub.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.usub.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.usub.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.usub.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.usub.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512BW-LABEL: 'usub'
@@ -874,70 +874,70 @@ define i32 @usub(i32 %arg) {
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.usub.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.usub.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = call { i16, i1 } @llvm.usub.with.overflow.i16(i16 undef, i16 undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.usub.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.usub.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.usub.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.usub.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.usub.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.usub.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = call { i8, i1 } @llvm.usub.with.overflow.i8(i8 undef, i8 undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.usub.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.usub.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.usub.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.usub.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.usub.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.usub.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SLM-LABEL: 'usub'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = call { i64, i1 } @llvm.usub.with.overflow.i64(i64 undef, i64 undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.usub.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.usub.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.usub.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.usub.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.usub.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.usub.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = call { i32, i1 } @llvm.usub.with.overflow.i32(i32 undef, i32 undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.usub.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.usub.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.usub.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.usub.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.usub.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.usub.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = call { i16, i1 } @llvm.usub.with.overflow.i16(i16 undef, i16 undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.usub.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.usub.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.usub.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.usub.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.usub.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.usub.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = call { i8, i1 } @llvm.usub.with.overflow.i8(i8 undef, i8 undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.usub.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.usub.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.usub.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.usub.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.usub.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.usub.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; GLM-LABEL: 'usub'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = call { i64, i1 } @llvm.usub.with.overflow.i64(i64 undef, i64 undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.usub.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.usub.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.usub.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.usub.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.usub.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.usub.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = call { i32, i1 } @llvm.usub.with.overflow.i32(i32 undef, i32 undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.usub.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.usub.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.usub.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.usub.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.usub.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.usub.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = call { i16, i1 } @llvm.usub.with.overflow.i16(i16 undef, i16 undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.usub.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.usub.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.usub.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.usub.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.usub.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.usub.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = call { i8, i1 } @llvm.usub.with.overflow.i8(i8 undef, i8 undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.usub.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.usub.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.usub.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.usub.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.usub.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.usub.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; BTVER2-LABEL: 'usub'
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = call { i64, i1 } @llvm.usub.with.overflow.i64(i64 undef, i64 undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.usub.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.usub.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.usub.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.usub.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.usub.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.usub.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = call { i32, i1 } @llvm.usub.with.overflow.i32(i32 undef, i32 undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.usub.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.usub.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.usub.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.usub.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.usub.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.usub.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = call { i16, i1 } @llvm.usub.with.overflow.i16(i16 undef, i16 undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.usub.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.usub.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.usub.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.usub.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.usub.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.usub.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = call { i8, i1 } @llvm.usub.with.overflow.i8(i8 undef, i8 undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.usub.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.usub.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.usub.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.usub.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.usub.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.usub.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
   %I64 = call {i64, i1} @llvm.usub.with.overflow.i64(i64 undef, i64 undef)
@@ -990,78 +990,78 @@ declare {<64 x i8>, <64 x i1>}  @llvm.smul.with.overflow.v64i8(<64 x i8>, <64 x
 define i32 @smul(i32 %arg) {
 ; SSSE3-LABEL: 'smul'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call { i64, i1 } @llvm.smul.with.overflow.i64(i64 undef, i64 undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.smul.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.smul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.smul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.smul.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.smul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 144 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.smul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call { i32, i1 } @llvm.smul.with.overflow.i32(i32 undef, i32 undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.smul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.smul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 108 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.smul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.smul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.smul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 120 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.smul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I16 = call { i16, i1 } @llvm.smul.with.overflow.i16(i16 undef, i16 undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.smul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.smul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 124 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.smul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.smul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.smul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 136 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.smul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I8 = call { i8, i1 } @llvm.smul.with.overflow.i8(i8 undef, i8 undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.smul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.smul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.smul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.smul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.smul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.smul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'smul'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call { i64, i1 } @llvm.smul.with.overflow.i64(i64 undef, i64 undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.smul.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.smul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.smul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.smul.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.smul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.smul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call { i32, i1 } @llvm.smul.with.overflow.i32(i32 undef, i32 undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.smul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.smul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 92 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.smul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.smul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.smul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.smul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I16 = call { i16, i1 } @llvm.smul.with.overflow.i16(i16 undef, i16 undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.smul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.smul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.smul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.smul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.smul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.smul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I8 = call { i8, i1 } @llvm.smul.with.overflow.i8(i8 undef, i8 undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.smul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.smul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.smul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.smul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.smul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.smul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX1-LABEL: 'smul'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call { i64, i1 } @llvm.smul.with.overflow.i64(i64 undef, i64 undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.smul.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 62 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.smul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 124 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.smul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.smul.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.smul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 130 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.smul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call { i32, i1 } @llvm.smul.with.overflow.i32(i32 undef, i32 undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.smul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.smul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 120 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.smul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.smul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 63 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.smul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 126 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.smul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I16 = call { i16, i1 } @llvm.smul.with.overflow.i16(i16 undef, i16 undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.smul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.smul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.smul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.smul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.smul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 102 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.smul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I8 = call { i8, i1 } @llvm.smul.with.overflow.i8(i8 undef, i8 undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.smul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.smul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.smul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.smul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 55 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.smul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 110 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.smul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'smul'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call { i64, i1 } @llvm.smul.with.overflow.i64(i64 undef, i64 undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.smul.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.smul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.smul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.smul.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 53 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.smul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 106 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.smul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call { i32, i1 } @llvm.smul.with.overflow.i32(i32 undef, i32 undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.smul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.smul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 62 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.smul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.smul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.smul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.smul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I16 = call { i16, i1 } @llvm.smul.with.overflow.i16(i16 undef, i16 undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.smul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.smul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.smul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.smul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.smul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.smul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I8 = call { i8, i1 } @llvm.smul.with.overflow.i8(i8 undef, i8 undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.smul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.smul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.smul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.smul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.smul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.smul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512F-LABEL: 'smul'
@@ -1074,13 +1074,13 @@ define i32 @smul(i32 %arg) {
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.smul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.smul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I16 = call { i16, i1 } @llvm.smul.with.overflow.i16(i16 undef, i16 undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.smul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.smul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.smul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.smul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.smul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.smul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I8 = call { i8, i1 } @llvm.smul.with.overflow.i8(i8 undef, i8 undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.smul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.smul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.smul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.smul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.smul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 69 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.smul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512BW-LABEL: 'smul'
@@ -1112,70 +1112,70 @@ define i32 @smul(i32 %arg) {
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.smul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.smul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I16 = call { i16, i1 } @llvm.smul.with.overflow.i16(i16 undef, i16 undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.smul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.smul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.smul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.smul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.smul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.smul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I8 = call { i8, i1 } @llvm.smul.with.overflow.i8(i8 undef, i8 undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.smul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.smul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.smul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.smul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.smul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 69 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.smul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SLM-LABEL: 'smul'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call { i64, i1 } @llvm.smul.with.overflow.i64(i64 undef, i64 undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.smul.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.smul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 108 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.smul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.smul.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.smul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 120 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.smul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call { i32, i1 } @llvm.smul.with.overflow.i32(i32 undef, i32 undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.smul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 90 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.smul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 180 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.smul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.smul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.smul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.smul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I16 = call { i16, i1 } @llvm.smul.with.overflow.i16(i16 undef, i16 undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.smul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 78 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.smul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 148 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.smul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.smul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.smul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.smul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I8 = call { i8, i1 } @llvm.smul.with.overflow.i8(i8 undef, i8 undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.smul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.smul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 92 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.smul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.smul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.smul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.smul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; GLM-LABEL: 'smul'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call { i64, i1 } @llvm.smul.with.overflow.i64(i64 undef, i64 undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.smul.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.smul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.smul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.smul.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.smul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.smul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call { i32, i1 } @llvm.smul.with.overflow.i32(i32 undef, i32 undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.smul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.smul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 92 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.smul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.smul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.smul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.smul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I16 = call { i16, i1 } @llvm.smul.with.overflow.i16(i16 undef, i16 undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.smul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.smul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.smul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.smul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.smul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.smul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I8 = call { i8, i1 } @llvm.smul.with.overflow.i8(i8 undef, i8 undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.smul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.smul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.smul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.smul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.smul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.smul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; BTVER2-LABEL: 'smul'
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call { i64, i1 } @llvm.smul.with.overflow.i64(i64 undef, i64 undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.smul.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 62 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.smul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 124 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.smul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.smul.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.smul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 130 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.smul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call { i32, i1 } @llvm.smul.with.overflow.i32(i32 undef, i32 undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.smul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.smul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 120 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.smul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.smul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 63 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.smul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 126 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.smul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I16 = call { i16, i1 } @llvm.smul.with.overflow.i16(i16 undef, i16 undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.smul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.smul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.smul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.smul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.smul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 102 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.smul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I8 = call { i8, i1 } @llvm.smul.with.overflow.i8(i8 undef, i8 undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.smul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.smul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.smul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.smul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 55 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.smul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 110 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.smul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
   %I64 = call {i64, i1} @llvm.smul.with.overflow.i64(i64 undef, i64 undef)
@@ -1228,78 +1228,78 @@ declare {<64 x i8>, <64 x i1>}  @llvm.umul.with.overflow.v64i8(<64 x i8>, <64 x
 define i32 @umul(i32 %arg) {
 ; SSSE3-LABEL: 'umul'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call { i64, i1 } @llvm.umul.with.overflow.i64(i64 undef, i64 undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.umul.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.umul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.umul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.umul.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.umul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.umul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call { i32, i1 } @llvm.umul.with.overflow.i32(i32 undef, i32 undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.umul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.umul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.umul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.umul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.umul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.umul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I16 = call { i16, i1 } @llvm.umul.with.overflow.i16(i16 undef, i16 undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.umul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.umul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 120 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.umul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.umul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 62 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.umul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.umul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I8 = call { i8, i1 } @llvm.umul.with.overflow.i8(i8 undef, i8 undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.umul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.umul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.umul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.umul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.umul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.umul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'umul'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call { i64, i1 } @llvm.umul.with.overflow.i64(i64 undef, i64 undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.umul.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.umul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.umul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.umul.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.umul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.umul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call { i32, i1 } @llvm.umul.with.overflow.i32(i32 undef, i32 undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.umul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.umul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.umul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.umul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.umul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.umul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I16 = call { i16, i1 } @llvm.umul.with.overflow.i16(i16 undef, i16 undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.umul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.umul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.umul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.umul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.umul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.umul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I8 = call { i8, i1 } @llvm.umul.with.overflow.i8(i8 undef, i8 undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.umul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.umul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.umul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.umul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.umul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.umul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX1-LABEL: 'umul'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call { i64, i1 } @llvm.umul.with.overflow.i64(i64 undef, i64 undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.umul.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.umul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.umul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.umul.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 55 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.umul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 110 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.umul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call { i32, i1 } @llvm.umul.with.overflow.i32(i32 undef, i32 undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.umul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.umul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 112 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.umul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.umul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.umul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 118 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.umul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I16 = call { i16, i1 } @llvm.umul.with.overflow.i16(i16 undef, i16 undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.umul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.umul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.umul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.umul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.umul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 94 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.umul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I8 = call { i8, i1 } @llvm.umul.with.overflow.i8(i8 undef, i8 undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.umul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.umul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.umul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.umul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.umul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 90 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.umul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'umul'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call { i64, i1 } @llvm.umul.with.overflow.i64(i64 undef, i64 undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.umul.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.umul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 92 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.umul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.umul.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.umul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 98 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.umul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call { i32, i1 } @llvm.umul.with.overflow.i32(i32 undef, i32 undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.umul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.umul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.umul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.umul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.umul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.umul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I16 = call { i16, i1 } @llvm.umul.with.overflow.i16(i16 undef, i16 undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.umul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.umul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.umul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.umul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.umul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.umul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I8 = call { i8, i1 } @llvm.umul.with.overflow.i8(i8 undef, i8 undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.umul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.umul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.umul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.umul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.umul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.umul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512F-LABEL: 'umul'
@@ -1312,13 +1312,13 @@ define i32 @umul(i32 %arg) {
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.umul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.umul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I16 = call { i16, i1 } @llvm.umul.with.overflow.i16(i16 undef, i16 undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.umul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.umul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.umul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.umul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.umul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.umul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I8 = call { i8, i1 } @llvm.umul.with.overflow.i8(i8 undef, i8 undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.umul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.umul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.umul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.umul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.umul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.umul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512BW-LABEL: 'umul'
@@ -1350,70 +1350,70 @@ define i32 @umul(i32 %arg) {
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.umul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.umul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I16 = call { i16, i1 } @llvm.umul.with.overflow.i16(i16 undef, i16 undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.umul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.umul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.umul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.umul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.umul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.umul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I8 = call { i8, i1 } @llvm.umul.with.overflow.i8(i8 undef, i8 undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.umul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.umul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.umul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.umul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.umul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.umul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SLM-LABEL: 'umul'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call { i64, i1 } @llvm.umul.with.overflow.i64(i64 undef, i64 undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.umul.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.umul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 92 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.umul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.umul.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.umul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.umul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call { i32, i1 } @llvm.umul.with.overflow.i32(i32 undef, i32 undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.umul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.umul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 176 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.umul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.umul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 94 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.umul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 188 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.umul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I16 = call { i16, i1 } @llvm.umul.with.overflow.i16(i16 undef, i16 undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.umul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.umul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 144 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.umul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.umul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 82 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.umul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 156 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.umul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I8 = call { i8, i1 } @llvm.umul.with.overflow.i8(i8 undef, i8 undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.umul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.umul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.umul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.umul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.umul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.umul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; GLM-LABEL: 'umul'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call { i64, i1 } @llvm.umul.with.overflow.i64(i64 undef, i64 undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.umul.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.umul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.umul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.umul.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.umul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.umul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call { i32, i1 } @llvm.umul.with.overflow.i32(i32 undef, i32 undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.umul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.umul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.umul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.umul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.umul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.umul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I16 = call { i16, i1 } @llvm.umul.with.overflow.i16(i16 undef, i16 undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.umul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.umul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.umul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.umul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.umul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.umul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I8 = call { i8, i1 } @llvm.umul.with.overflow.i8(i8 undef, i8 undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.umul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.umul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.umul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.umul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.umul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.umul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; BTVER2-LABEL: 'umul'
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call { i64, i1 } @llvm.umul.with.overflow.i64(i64 undef, i64 undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.umul.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.umul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.umul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.umul.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 55 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.umul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 110 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.umul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call { i32, i1 } @llvm.umul.with.overflow.i32(i32 undef, i32 undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.umul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.umul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 112 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.umul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.umul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 59 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.umul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 118 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.umul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I16 = call { i16, i1 } @llvm.umul.with.overflow.i16(i16 undef, i16 undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.umul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.umul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.umul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.umul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.umul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 94 for instruction: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.umul.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I8 = call { i8, i1 } @llvm.umul.with.overflow.i8(i8 undef, i8 undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.umul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.umul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.umul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.umul.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.umul.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 90 for instruction: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.umul.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
   %I64 = call {i64, i1} @llvm.umul.with.overflow.i64(i64 undef, i64 undef)

diff  --git a/llvm/test/Analysis/CostModel/X86/arith-sminmax.ll b/llvm/test/Analysis/CostModel/X86/arith-sminmax.ll
index 8e9ef5bba0bb5..36bf15c7b820d 100644
--- a/llvm/test/Analysis/CostModel/X86/arith-sminmax.ll
+++ b/llvm/test/Analysis/CostModel/X86/arith-sminmax.ll
@@ -31,47 +31,47 @@ declare <64 x i8>  @llvm.smax.v64i8(<64 x i8>, <64 x i8>)
 define i32 @smax(i32 %arg) {
 ; SSE2-LABEL: 'smax'
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.smax.i64(i64 undef, i64 undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2I64 = call <2 x i64> @llvm.smax.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4I64 = call <4 x i64> @llvm.smax.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V8I64 = call <8 x i64> @llvm.smax.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2I64 = call <2 x i64> @llvm.smax.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V4I64 = call <4 x i64> @llvm.smax.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V8I64 = call <8 x i64> @llvm.smax.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.smax.i32(i32 undef, i32 undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4I32 = call <4 x i32> @llvm.smax.v4i32(<4 x i32> undef, <4 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I32 = call <8 x i32> @llvm.smax.v8i32(<8 x i32> undef, <8 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16I32 = call <16 x i32> @llvm.smax.v16i32(<16 x i32> undef, <16 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I32 = call <4 x i32> @llvm.smax.v4i32(<4 x i32> undef, <4 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8I32 = call <8 x i32> @llvm.smax.v8i32(<8 x i32> undef, <8 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16I32 = call <16 x i32> @llvm.smax.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I16 = call i16 @llvm.smax.i16(i16 undef, i16 undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.smax.v8i16(<8 x i16> undef, <8 x i16> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = call <16 x i16> @llvm.smax.v16i16(<16 x i16> undef, <16 x i16> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32I16 = call <32 x i16> @llvm.smax.v32i16(<32 x i16> undef, <32 x i16> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I8 = call i8 @llvm.smax.i8(i8 undef, i8 undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I8 = call <16 x i8> @llvm.smax.v16i8(<16 x i8> undef, <16 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32I8 = call <32 x i8> @llvm.smax.v32i8(<32 x i8> undef, <32 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V64I8 = call <64 x i8> @llvm.smax.v64i8(<64 x i8> undef, <64 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V16I8 = call <16 x i8> @llvm.smax.v16i8(<16 x i8> undef, <16 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V32I8 = call <32 x i8> @llvm.smax.v32i8(<32 x i8> undef, <32 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V64I8 = call <64 x i8> @llvm.smax.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'smax'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.smax.i64(i64 undef, i64 undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2I64 = call <2 x i64> @llvm.smax.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4I64 = call <4 x i64> @llvm.smax.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V8I64 = call <8 x i64> @llvm.smax.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2I64 = call <2 x i64> @llvm.smax.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V4I64 = call <4 x i64> @llvm.smax.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V8I64 = call <8 x i64> @llvm.smax.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.smax.i32(i32 undef, i32 undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4I32 = call <4 x i32> @llvm.smax.v4i32(<4 x i32> undef, <4 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I32 = call <8 x i32> @llvm.smax.v8i32(<8 x i32> undef, <8 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16I32 = call <16 x i32> @llvm.smax.v16i32(<16 x i32> undef, <16 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I32 = call <4 x i32> @llvm.smax.v4i32(<4 x i32> undef, <4 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8I32 = call <8 x i32> @llvm.smax.v8i32(<8 x i32> undef, <8 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16I32 = call <16 x i32> @llvm.smax.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I16 = call i16 @llvm.smax.i16(i16 undef, i16 undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.smax.v8i16(<8 x i16> undef, <8 x i16> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = call <16 x i16> @llvm.smax.v16i16(<16 x i16> undef, <16 x i16> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32I16 = call <32 x i16> @llvm.smax.v32i16(<32 x i16> undef, <32 x i16> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I8 = call i8 @llvm.smax.i8(i8 undef, i8 undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I8 = call <16 x i8> @llvm.smax.v16i8(<16 x i8> undef, <16 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32I8 = call <32 x i8> @llvm.smax.v32i8(<32 x i8> undef, <32 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V64I8 = call <64 x i8> @llvm.smax.v64i8(<64 x i8> undef, <64 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V16I8 = call <16 x i8> @llvm.smax.v16i8(<16 x i8> undef, <16 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V32I8 = call <32 x i8> @llvm.smax.v32i8(<32 x i8> undef, <32 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V64I8 = call <64 x i8> @llvm.smax.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'smax'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.smax.i64(i64 undef, i64 undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call <2 x i64> @llvm.smax.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4I64 = call <4 x i64> @llvm.smax.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I64 = call <8 x i64> @llvm.smax.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.smax.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call <4 x i64> @llvm.smax.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V8I64 = call <8 x i64> @llvm.smax.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.smax.i32(i32 undef, i32 undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.smax.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I32 = call <8 x i32> @llvm.smax.v8i32(<8 x i32> undef, <8 x i32> undef)
@@ -88,9 +88,9 @@ define i32 @smax(i32 %arg) {
 ;
 ; AVX1-LABEL: 'smax'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.smax.i64(i64 undef, i64 undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call <2 x i64> @llvm.smax.v2i64(<2 x i64> undef, <2 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4I64 = call <4 x i64> @llvm.smax.v4i64(<4 x i64> undef, <4 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I64 = call <8 x i64> @llvm.smax.v8i64(<8 x i64> undef, <8 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.smax.v2i64(<2 x i64> undef, <2 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I64 = call <4 x i64> @llvm.smax.v4i64(<4 x i64> undef, <4 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I64 = call <8 x i64> @llvm.smax.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.smax.i32(i32 undef, i32 undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.smax.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I32 = call <8 x i32> @llvm.smax.v8i32(<8 x i32> undef, <8 x i32> undef)
@@ -107,9 +107,9 @@ define i32 @smax(i32 %arg) {
 ;
 ; AVX2-LABEL: 'smax'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.smax.i64(i64 undef, i64 undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call <2 x i64> @llvm.smax.v2i64(<2 x i64> undef, <2 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I64 = call <4 x i64> @llvm.smax.v4i64(<4 x i64> undef, <4 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I64 = call <8 x i64> @llvm.smax.v8i64(<8 x i64> undef, <8 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.smax.v2i64(<2 x i64> undef, <2 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4I64 = call <4 x i64> @llvm.smax.v4i64(<4 x i64> undef, <4 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I64 = call <8 x i64> @llvm.smax.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.smax.i32(i32 undef, i32 undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.smax.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.smax.v8i32(<8 x i32> undef, <8 x i32> undef)
@@ -227,47 +227,47 @@ declare <64 x i8>  @llvm.smin.v64i8(<64 x i8>, <64 x i8>)
 define i32 @smin(i32 %arg) {
 ; SSE2-LABEL: 'smin'
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.smin.i64(i64 undef, i64 undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2I64 = call <2 x i64> @llvm.smin.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4I64 = call <4 x i64> @llvm.smin.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V8I64 = call <8 x i64> @llvm.smin.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2I64 = call <2 x i64> @llvm.smin.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V4I64 = call <4 x i64> @llvm.smin.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V8I64 = call <8 x i64> @llvm.smin.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.smin.i32(i32 undef, i32 undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4I32 = call <4 x i32> @llvm.smin.v4i32(<4 x i32> undef, <4 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I32 = call <8 x i32> @llvm.smin.v8i32(<8 x i32> undef, <8 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16I32 = call <16 x i32> @llvm.smin.v16i32(<16 x i32> undef, <16 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I32 = call <4 x i32> @llvm.smin.v4i32(<4 x i32> undef, <4 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8I32 = call <8 x i32> @llvm.smin.v8i32(<8 x i32> undef, <8 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16I32 = call <16 x i32> @llvm.smin.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I16 = call i16 @llvm.smin.i16(i16 undef, i16 undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.smin.v8i16(<8 x i16> undef, <8 x i16> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = call <16 x i16> @llvm.smin.v16i16(<16 x i16> undef, <16 x i16> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32I16 = call <32 x i16> @llvm.smin.v32i16(<32 x i16> undef, <32 x i16> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I8 = call i8 @llvm.smin.i8(i8 undef, i8 undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I8 = call <16 x i8> @llvm.smin.v16i8(<16 x i8> undef, <16 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32I8 = call <32 x i8> @llvm.smin.v32i8(<32 x i8> undef, <32 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V64I8 = call <64 x i8> @llvm.smin.v64i8(<64 x i8> undef, <64 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V16I8 = call <16 x i8> @llvm.smin.v16i8(<16 x i8> undef, <16 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V32I8 = call <32 x i8> @llvm.smin.v32i8(<32 x i8> undef, <32 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V64I8 = call <64 x i8> @llvm.smin.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'smin'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.smin.i64(i64 undef, i64 undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2I64 = call <2 x i64> @llvm.smin.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4I64 = call <4 x i64> @llvm.smin.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V8I64 = call <8 x i64> @llvm.smin.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2I64 = call <2 x i64> @llvm.smin.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V4I64 = call <4 x i64> @llvm.smin.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V8I64 = call <8 x i64> @llvm.smin.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.smin.i32(i32 undef, i32 undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4I32 = call <4 x i32> @llvm.smin.v4i32(<4 x i32> undef, <4 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I32 = call <8 x i32> @llvm.smin.v8i32(<8 x i32> undef, <8 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16I32 = call <16 x i32> @llvm.smin.v16i32(<16 x i32> undef, <16 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I32 = call <4 x i32> @llvm.smin.v4i32(<4 x i32> undef, <4 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8I32 = call <8 x i32> @llvm.smin.v8i32(<8 x i32> undef, <8 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16I32 = call <16 x i32> @llvm.smin.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I16 = call i16 @llvm.smin.i16(i16 undef, i16 undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.smin.v8i16(<8 x i16> undef, <8 x i16> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = call <16 x i16> @llvm.smin.v16i16(<16 x i16> undef, <16 x i16> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32I16 = call <32 x i16> @llvm.smin.v32i16(<32 x i16> undef, <32 x i16> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I8 = call i8 @llvm.smin.i8(i8 undef, i8 undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I8 = call <16 x i8> @llvm.smin.v16i8(<16 x i8> undef, <16 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32I8 = call <32 x i8> @llvm.smin.v32i8(<32 x i8> undef, <32 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V64I8 = call <64 x i8> @llvm.smin.v64i8(<64 x i8> undef, <64 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V16I8 = call <16 x i8> @llvm.smin.v16i8(<16 x i8> undef, <16 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V32I8 = call <32 x i8> @llvm.smin.v32i8(<32 x i8> undef, <32 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V64I8 = call <64 x i8> @llvm.smin.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'smin'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.smin.i64(i64 undef, i64 undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call <2 x i64> @llvm.smin.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4I64 = call <4 x i64> @llvm.smin.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I64 = call <8 x i64> @llvm.smin.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.smin.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call <4 x i64> @llvm.smin.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V8I64 = call <8 x i64> @llvm.smin.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.smin.i32(i32 undef, i32 undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.smin.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I32 = call <8 x i32> @llvm.smin.v8i32(<8 x i32> undef, <8 x i32> undef)
@@ -284,9 +284,9 @@ define i32 @smin(i32 %arg) {
 ;
 ; AVX1-LABEL: 'smin'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.smin.i64(i64 undef, i64 undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call <2 x i64> @llvm.smin.v2i64(<2 x i64> undef, <2 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4I64 = call <4 x i64> @llvm.smin.v4i64(<4 x i64> undef, <4 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I64 = call <8 x i64> @llvm.smin.v8i64(<8 x i64> undef, <8 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.smin.v2i64(<2 x i64> undef, <2 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I64 = call <4 x i64> @llvm.smin.v4i64(<4 x i64> undef, <4 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I64 = call <8 x i64> @llvm.smin.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.smin.i32(i32 undef, i32 undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.smin.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I32 = call <8 x i32> @llvm.smin.v8i32(<8 x i32> undef, <8 x i32> undef)
@@ -303,9 +303,9 @@ define i32 @smin(i32 %arg) {
 ;
 ; AVX2-LABEL: 'smin'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.smin.i64(i64 undef, i64 undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call <2 x i64> @llvm.smin.v2i64(<2 x i64> undef, <2 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I64 = call <4 x i64> @llvm.smin.v4i64(<4 x i64> undef, <4 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I64 = call <8 x i64> @llvm.smin.v8i64(<8 x i64> undef, <8 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.smin.v2i64(<2 x i64> undef, <2 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4I64 = call <4 x i64> @llvm.smin.v4i64(<4 x i64> undef, <4 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I64 = call <8 x i64> @llvm.smin.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.smin.i32(i32 undef, i32 undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.smin.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.smin.v8i32(<8 x i32> undef, <8 x i32> undef)

diff  --git a/llvm/test/Analysis/CostModel/X86/arith-ssat.ll b/llvm/test/Analysis/CostModel/X86/arith-ssat.ll
index 6018fc4987d11..ad4200b59b0fe 100644
--- a/llvm/test/Analysis/CostModel/X86/arith-ssat.ll
+++ b/llvm/test/Analysis/CostModel/X86/arith-ssat.ll
@@ -37,13 +37,13 @@ declare <64 x i8>  @llvm.sadd.sat.v64i8(<64 x i8>, <64 x i8>)
 define i32 @add(i32 %arg) {
 ; SSSE3-LABEL: 'add'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.sadd.sat.i64(i64 undef, i64 undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V2I64 = call <2 x i64> @llvm.sadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 63 for instruction: %V4I64 = call <4 x i64> @llvm.sadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 125 for instruction: %V8I64 = call <8 x i64> @llvm.sadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V2I64 = call <2 x i64> @llvm.sadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 69 for instruction: %V4I64 = call <4 x i64> @llvm.sadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 137 for instruction: %V8I64 = call <8 x i64> @llvm.sadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.sadd.sat.i32(i32 undef, i32 undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4I32 = call <4 x i32> @llvm.sadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V8I32 = call <8 x i32> @llvm.sadd.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: %V16I32 = call <16 x i32> @llvm.sadd.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I32 = call <4 x i32> @llvm.sadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V8I32 = call <8 x i32> @llvm.sadd.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 53 for instruction: %V16I32 = call <16 x i32> @llvm.sadd.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.sadd.sat.i16(i16 undef, i16 undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.sadd.sat.v8i16(<8 x i16> undef, <8 x i16> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = call <16 x i16> @llvm.sadd.sat.v16i16(<16 x i16> undef, <16 x i16> undef)
@@ -56,13 +56,13 @@ define i32 @add(i32 %arg) {
 ;
 ; SSE42-LABEL: 'add'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.sadd.sat.i64(i64 undef, i64 undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.sadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I64 = call <4 x i64> @llvm.sadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8I64 = call <8 x i64> @llvm.sadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.sadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4I64 = call <4 x i64> @llvm.sadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V8I64 = call <8 x i64> @llvm.sadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.sadd.sat.i32(i32 undef, i32 undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I32 = call <4 x i32> @llvm.sadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8I32 = call <8 x i32> @llvm.sadd.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V16I32 = call <16 x i32> @llvm.sadd.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I32 = call <4 x i32> @llvm.sadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8I32 = call <8 x i32> @llvm.sadd.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V16I32 = call <16 x i32> @llvm.sadd.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.sadd.sat.i16(i16 undef, i16 undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.sadd.sat.v8i16(<8 x i16> undef, <8 x i16> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = call <16 x i16> @llvm.sadd.sat.v16i16(<16 x i16> undef, <16 x i16> undef)
@@ -75,13 +75,13 @@ define i32 @add(i32 %arg) {
 ;
 ; AVX1-LABEL: 'add'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.sadd.sat.i64(i64 undef, i64 undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.sadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4I64 = call <4 x i64> @llvm.sadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V8I64 = call <8 x i64> @llvm.sadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.sadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4I64 = call <4 x i64> @llvm.sadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V8I64 = call <8 x i64> @llvm.sadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.sadd.sat.i32(i32 undef, i32 undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I32 = call <4 x i32> @llvm.sadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8I32 = call <8 x i32> @llvm.sadd.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V16I32 = call <16 x i32> @llvm.sadd.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I32 = call <4 x i32> @llvm.sadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V8I32 = call <8 x i32> @llvm.sadd.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V16I32 = call <16 x i32> @llvm.sadd.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.sadd.sat.i16(i16 undef, i16 undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.sadd.sat.v8i16(<8 x i16> undef, <8 x i16> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I16 = call <16 x i16> @llvm.sadd.sat.v16i16(<16 x i16> undef, <16 x i16> undef)
@@ -94,13 +94,13 @@ define i32 @add(i32 %arg) {
 ;
 ; AVX2-LABEL: 'add'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.sadd.sat.i64(i64 undef, i64 undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.sadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I64 = call <4 x i64> @llvm.sadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8I64 = call <8 x i64> @llvm.sadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.sadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call <4 x i64> @llvm.sadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8I64 = call <8 x i64> @llvm.sadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.sadd.sat.i32(i32 undef, i32 undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I32 = call <4 x i32> @llvm.sadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I32 = call <8 x i32> @llvm.sadd.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V16I32 = call <16 x i32> @llvm.sadd.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I32 = call <4 x i32> @llvm.sadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I32 = call <8 x i32> @llvm.sadd.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V16I32 = call <16 x i32> @llvm.sadd.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.sadd.sat.i16(i16 undef, i16 undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.sadd.sat.v8i16(<8 x i16> undef, <8 x i16> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = call <16 x i16> @llvm.sadd.sat.v16i16(<16 x i16> undef, <16 x i16> undef)
@@ -170,13 +170,13 @@ define i32 @add(i32 %arg) {
 ;
 ; SLM-LABEL: 'add'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.sadd.sat.i64(i64 undef, i64 undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V2I64 = call <2 x i64> @llvm.sadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V4I64 = call <4 x i64> @llvm.sadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: %V8I64 = call <8 x i64> @llvm.sadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V2I64 = call <2 x i64> @llvm.sadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V4I64 = call <4 x i64> @llvm.sadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V8I64 = call <8 x i64> @llvm.sadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.sadd.sat.i32(i32 undef, i32 undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I32 = call <4 x i32> @llvm.sadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8I32 = call <8 x i32> @llvm.sadd.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V16I32 = call <16 x i32> @llvm.sadd.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I32 = call <4 x i32> @llvm.sadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8I32 = call <8 x i32> @llvm.sadd.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V16I32 = call <16 x i32> @llvm.sadd.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.sadd.sat.i16(i16 undef, i16 undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.sadd.sat.v8i16(<8 x i16> undef, <8 x i16> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = call <16 x i16> @llvm.sadd.sat.v16i16(<16 x i16> undef, <16 x i16> undef)
@@ -189,13 +189,13 @@ define i32 @add(i32 %arg) {
 ;
 ; GLM-LABEL: 'add'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.sadd.sat.i64(i64 undef, i64 undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.sadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I64 = call <4 x i64> @llvm.sadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8I64 = call <8 x i64> @llvm.sadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.sadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4I64 = call <4 x i64> @llvm.sadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V8I64 = call <8 x i64> @llvm.sadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.sadd.sat.i32(i32 undef, i32 undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I32 = call <4 x i32> @llvm.sadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8I32 = call <8 x i32> @llvm.sadd.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V16I32 = call <16 x i32> @llvm.sadd.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I32 = call <4 x i32> @llvm.sadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8I32 = call <8 x i32> @llvm.sadd.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V16I32 = call <16 x i32> @llvm.sadd.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.sadd.sat.i16(i16 undef, i16 undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.sadd.sat.v8i16(<8 x i16> undef, <8 x i16> undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = call <16 x i16> @llvm.sadd.sat.v16i16(<16 x i16> undef, <16 x i16> undef)
@@ -208,13 +208,13 @@ define i32 @add(i32 %arg) {
 ;
 ; BTVER2-LABEL: 'add'
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.sadd.sat.i64(i64 undef, i64 undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.sadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4I64 = call <4 x i64> @llvm.sadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V8I64 = call <8 x i64> @llvm.sadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.sadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4I64 = call <4 x i64> @llvm.sadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V8I64 = call <8 x i64> @llvm.sadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.sadd.sat.i32(i32 undef, i32 undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I32 = call <4 x i32> @llvm.sadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8I32 = call <8 x i32> @llvm.sadd.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V16I32 = call <16 x i32> @llvm.sadd.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I32 = call <4 x i32> @llvm.sadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V8I32 = call <8 x i32> @llvm.sadd.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V16I32 = call <16 x i32> @llvm.sadd.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.sadd.sat.i16(i16 undef, i16 undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.sadd.sat.v8i16(<8 x i16> undef, <8 x i16> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I16 = call <16 x i16> @llvm.sadd.sat.v16i16(<16 x i16> undef, <16 x i16> undef)
@@ -271,13 +271,13 @@ declare <64 x i8>  @llvm.ssub.sat.v64i8(<64 x i8>, <64 x i8>)
 define i32 @sub(i32 %arg) {
 ; SSSE3-LABEL: 'sub'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.ssub.sat.i64(i64 undef, i64 undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V2I64 = call <2 x i64> @llvm.ssub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 63 for instruction: %V4I64 = call <4 x i64> @llvm.ssub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 125 for instruction: %V8I64 = call <8 x i64> @llvm.ssub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V2I64 = call <2 x i64> @llvm.ssub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 69 for instruction: %V4I64 = call <4 x i64> @llvm.ssub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 137 for instruction: %V8I64 = call <8 x i64> @llvm.ssub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.ssub.sat.i32(i32 undef, i32 undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4I32 = call <4 x i32> @llvm.ssub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V8I32 = call <8 x i32> @llvm.ssub.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: %V16I32 = call <16 x i32> @llvm.ssub.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I32 = call <4 x i32> @llvm.ssub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V8I32 = call <8 x i32> @llvm.ssub.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 53 for instruction: %V16I32 = call <16 x i32> @llvm.ssub.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.ssub.sat.i16(i16 undef, i16 undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.ssub.sat.v8i16(<8 x i16> undef, <8 x i16> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = call <16 x i16> @llvm.ssub.sat.v16i16(<16 x i16> undef, <16 x i16> undef)
@@ -290,13 +290,13 @@ define i32 @sub(i32 %arg) {
 ;
 ; SSE42-LABEL: 'sub'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.ssub.sat.i64(i64 undef, i64 undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.ssub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I64 = call <4 x i64> @llvm.ssub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8I64 = call <8 x i64> @llvm.ssub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.ssub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4I64 = call <4 x i64> @llvm.ssub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V8I64 = call <8 x i64> @llvm.ssub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.ssub.sat.i32(i32 undef, i32 undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I32 = call <4 x i32> @llvm.ssub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8I32 = call <8 x i32> @llvm.ssub.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V16I32 = call <16 x i32> @llvm.ssub.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I32 = call <4 x i32> @llvm.ssub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8I32 = call <8 x i32> @llvm.ssub.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V16I32 = call <16 x i32> @llvm.ssub.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.ssub.sat.i16(i16 undef, i16 undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.ssub.sat.v8i16(<8 x i16> undef, <8 x i16> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = call <16 x i16> @llvm.ssub.sat.v16i16(<16 x i16> undef, <16 x i16> undef)
@@ -309,13 +309,13 @@ define i32 @sub(i32 %arg) {
 ;
 ; AVX1-LABEL: 'sub'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.ssub.sat.i64(i64 undef, i64 undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.ssub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4I64 = call <4 x i64> @llvm.ssub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V8I64 = call <8 x i64> @llvm.ssub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.ssub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4I64 = call <4 x i64> @llvm.ssub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V8I64 = call <8 x i64> @llvm.ssub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.ssub.sat.i32(i32 undef, i32 undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I32 = call <4 x i32> @llvm.ssub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8I32 = call <8 x i32> @llvm.ssub.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V16I32 = call <16 x i32> @llvm.ssub.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I32 = call <4 x i32> @llvm.ssub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V8I32 = call <8 x i32> @llvm.ssub.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V16I32 = call <16 x i32> @llvm.ssub.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.ssub.sat.i16(i16 undef, i16 undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.ssub.sat.v8i16(<8 x i16> undef, <8 x i16> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I16 = call <16 x i16> @llvm.ssub.sat.v16i16(<16 x i16> undef, <16 x i16> undef)
@@ -328,13 +328,13 @@ define i32 @sub(i32 %arg) {
 ;
 ; AVX2-LABEL: 'sub'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.ssub.sat.i64(i64 undef, i64 undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.ssub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I64 = call <4 x i64> @llvm.ssub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8I64 = call <8 x i64> @llvm.ssub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.ssub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call <4 x i64> @llvm.ssub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8I64 = call <8 x i64> @llvm.ssub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.ssub.sat.i32(i32 undef, i32 undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I32 = call <4 x i32> @llvm.ssub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I32 = call <8 x i32> @llvm.ssub.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V16I32 = call <16 x i32> @llvm.ssub.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I32 = call <4 x i32> @llvm.ssub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I32 = call <8 x i32> @llvm.ssub.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V16I32 = call <16 x i32> @llvm.ssub.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.ssub.sat.i16(i16 undef, i16 undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.ssub.sat.v8i16(<8 x i16> undef, <8 x i16> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I16 = call <16 x i16> @llvm.ssub.sat.v16i16(<16 x i16> undef, <16 x i16> undef)
@@ -404,13 +404,13 @@ define i32 @sub(i32 %arg) {
 ;
 ; SLM-LABEL: 'sub'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.ssub.sat.i64(i64 undef, i64 undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V2I64 = call <2 x i64> @llvm.ssub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V4I64 = call <4 x i64> @llvm.ssub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: %V8I64 = call <8 x i64> @llvm.ssub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V2I64 = call <2 x i64> @llvm.ssub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V4I64 = call <4 x i64> @llvm.ssub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V8I64 = call <8 x i64> @llvm.ssub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.ssub.sat.i32(i32 undef, i32 undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I32 = call <4 x i32> @llvm.ssub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8I32 = call <8 x i32> @llvm.ssub.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V16I32 = call <16 x i32> @llvm.ssub.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I32 = call <4 x i32> @llvm.ssub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8I32 = call <8 x i32> @llvm.ssub.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V16I32 = call <16 x i32> @llvm.ssub.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.ssub.sat.i16(i16 undef, i16 undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.ssub.sat.v8i16(<8 x i16> undef, <8 x i16> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = call <16 x i16> @llvm.ssub.sat.v16i16(<16 x i16> undef, <16 x i16> undef)
@@ -423,13 +423,13 @@ define i32 @sub(i32 %arg) {
 ;
 ; GLM-LABEL: 'sub'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.ssub.sat.i64(i64 undef, i64 undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.ssub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I64 = call <4 x i64> @llvm.ssub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8I64 = call <8 x i64> @llvm.ssub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.ssub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4I64 = call <4 x i64> @llvm.ssub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V8I64 = call <8 x i64> @llvm.ssub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.ssub.sat.i32(i32 undef, i32 undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I32 = call <4 x i32> @llvm.ssub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8I32 = call <8 x i32> @llvm.ssub.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V16I32 = call <16 x i32> @llvm.ssub.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I32 = call <4 x i32> @llvm.ssub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8I32 = call <8 x i32> @llvm.ssub.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V16I32 = call <16 x i32> @llvm.ssub.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.ssub.sat.i16(i16 undef, i16 undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.ssub.sat.v8i16(<8 x i16> undef, <8 x i16> undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = call <16 x i16> @llvm.ssub.sat.v16i16(<16 x i16> undef, <16 x i16> undef)
@@ -442,13 +442,13 @@ define i32 @sub(i32 %arg) {
 ;
 ; BTVER2-LABEL: 'sub'
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.ssub.sat.i64(i64 undef, i64 undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.ssub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4I64 = call <4 x i64> @llvm.ssub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V8I64 = call <8 x i64> @llvm.ssub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.ssub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4I64 = call <4 x i64> @llvm.ssub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V8I64 = call <8 x i64> @llvm.ssub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.ssub.sat.i32(i32 undef, i32 undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I32 = call <4 x i32> @llvm.ssub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8I32 = call <8 x i32> @llvm.ssub.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V16I32 = call <16 x i32> @llvm.ssub.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I32 = call <4 x i32> @llvm.ssub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V8I32 = call <8 x i32> @llvm.ssub.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V16I32 = call <16 x i32> @llvm.ssub.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.ssub.sat.i16(i16 undef, i16 undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.ssub.sat.v8i16(<8 x i16> undef, <8 x i16> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I16 = call <16 x i16> @llvm.ssub.sat.v16i16(<16 x i16> undef, <16 x i16> undef)

diff  --git a/llvm/test/Analysis/CostModel/X86/arith-uminmax.ll b/llvm/test/Analysis/CostModel/X86/arith-uminmax.ll
index b8e748d3fda01..43c3833ad481d 100644
--- a/llvm/test/Analysis/CostModel/X86/arith-uminmax.ll
+++ b/llvm/test/Analysis/CostModel/X86/arith-uminmax.ll
@@ -31,13 +31,13 @@ declare <64 x i8>  @llvm.umax.v64i8(<64 x i8>, <64 x i8>)
 define i32 @umax(i32 %arg) {
 ; SSE2-LABEL: 'umax'
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.umax.i64(i64 undef, i64 undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2I64 = call <2 x i64> @llvm.umax.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4I64 = call <4 x i64> @llvm.umax.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V8I64 = call <8 x i64> @llvm.umax.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2I64 = call <2 x i64> @llvm.umax.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V4I64 = call <4 x i64> @llvm.umax.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V8I64 = call <8 x i64> @llvm.umax.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.umax.i32(i32 undef, i32 undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4I32 = call <4 x i32> @llvm.umax.v4i32(<4 x i32> undef, <4 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I32 = call <8 x i32> @llvm.umax.v8i32(<8 x i32> undef, <8 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16I32 = call <16 x i32> @llvm.umax.v16i32(<16 x i32> undef, <16 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I32 = call <4 x i32> @llvm.umax.v4i32(<4 x i32> undef, <4 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8I32 = call <8 x i32> @llvm.umax.v8i32(<8 x i32> undef, <8 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16I32 = call <16 x i32> @llvm.umax.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I16 = call i16 @llvm.umax.i16(i16 undef, i16 undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I16 = call <8 x i16> @llvm.umax.v8i16(<8 x i16> undef, <8 x i16> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I16 = call <16 x i16> @llvm.umax.v16i16(<16 x i16> undef, <16 x i16> undef)
@@ -50,13 +50,13 @@ define i32 @umax(i32 %arg) {
 ;
 ; SSSE3-LABEL: 'umax'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.umax.i64(i64 undef, i64 undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2I64 = call <2 x i64> @llvm.umax.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4I64 = call <4 x i64> @llvm.umax.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V8I64 = call <8 x i64> @llvm.umax.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2I64 = call <2 x i64> @llvm.umax.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V4I64 = call <4 x i64> @llvm.umax.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V8I64 = call <8 x i64> @llvm.umax.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.umax.i32(i32 undef, i32 undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4I32 = call <4 x i32> @llvm.umax.v4i32(<4 x i32> undef, <4 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I32 = call <8 x i32> @llvm.umax.v8i32(<8 x i32> undef, <8 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16I32 = call <16 x i32> @llvm.umax.v16i32(<16 x i32> undef, <16 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I32 = call <4 x i32> @llvm.umax.v4i32(<4 x i32> undef, <4 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8I32 = call <8 x i32> @llvm.umax.v8i32(<8 x i32> undef, <8 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16I32 = call <16 x i32> @llvm.umax.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I16 = call i16 @llvm.umax.i16(i16 undef, i16 undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I16 = call <8 x i16> @llvm.umax.v8i16(<8 x i16> undef, <8 x i16> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I16 = call <16 x i16> @llvm.umax.v16i16(<16 x i16> undef, <16 x i16> undef)
@@ -69,9 +69,9 @@ define i32 @umax(i32 %arg) {
 ;
 ; SSE42-LABEL: 'umax'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.umax.i64(i64 undef, i64 undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call <2 x i64> @llvm.umax.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4I64 = call <4 x i64> @llvm.umax.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I64 = call <8 x i64> @llvm.umax.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.umax.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call <4 x i64> @llvm.umax.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V8I64 = call <8 x i64> @llvm.umax.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.umax.i32(i32 undef, i32 undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.umax.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I32 = call <8 x i32> @llvm.umax.v8i32(<8 x i32> undef, <8 x i32> undef)
@@ -88,9 +88,9 @@ define i32 @umax(i32 %arg) {
 ;
 ; AVX1-LABEL: 'umax'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.umax.i64(i64 undef, i64 undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call <2 x i64> @llvm.umax.v2i64(<2 x i64> undef, <2 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4I64 = call <4 x i64> @llvm.umax.v4i64(<4 x i64> undef, <4 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I64 = call <8 x i64> @llvm.umax.v8i64(<8 x i64> undef, <8 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.umax.v2i64(<2 x i64> undef, <2 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I64 = call <4 x i64> @llvm.umax.v4i64(<4 x i64> undef, <4 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I64 = call <8 x i64> @llvm.umax.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.umax.i32(i32 undef, i32 undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.umax.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I32 = call <8 x i32> @llvm.umax.v8i32(<8 x i32> undef, <8 x i32> undef)
@@ -107,9 +107,9 @@ define i32 @umax(i32 %arg) {
 ;
 ; AVX2-LABEL: 'umax'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.umax.i64(i64 undef, i64 undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call <2 x i64> @llvm.umax.v2i64(<2 x i64> undef, <2 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I64 = call <4 x i64> @llvm.umax.v4i64(<4 x i64> undef, <4 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I64 = call <8 x i64> @llvm.umax.v8i64(<8 x i64> undef, <8 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.umax.v2i64(<2 x i64> undef, <2 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4I64 = call <4 x i64> @llvm.umax.v4i64(<4 x i64> undef, <4 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I64 = call <8 x i64> @llvm.umax.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.umax.i32(i32 undef, i32 undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.umax.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.umax.v8i32(<8 x i32> undef, <8 x i32> undef)
@@ -227,13 +227,13 @@ declare <64 x i8>  @llvm.umin.v64i8(<64 x i8>, <64 x i8>)
 define i32 @umin(i32 %arg) {
 ; SSE2-LABEL: 'umin'
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.umin.i64(i64 undef, i64 undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2I64 = call <2 x i64> @llvm.umin.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4I64 = call <4 x i64> @llvm.umin.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V8I64 = call <8 x i64> @llvm.umin.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2I64 = call <2 x i64> @llvm.umin.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V4I64 = call <4 x i64> @llvm.umin.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V8I64 = call <8 x i64> @llvm.umin.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.umin.i32(i32 undef, i32 undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4I32 = call <4 x i32> @llvm.umin.v4i32(<4 x i32> undef, <4 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I32 = call <8 x i32> @llvm.umin.v8i32(<8 x i32> undef, <8 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16I32 = call <16 x i32> @llvm.umin.v16i32(<16 x i32> undef, <16 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I32 = call <4 x i32> @llvm.umin.v4i32(<4 x i32> undef, <4 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8I32 = call <8 x i32> @llvm.umin.v8i32(<8 x i32> undef, <8 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16I32 = call <16 x i32> @llvm.umin.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I16 = call i16 @llvm.umin.i16(i16 undef, i16 undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I16 = call <8 x i16> @llvm.umin.v8i16(<8 x i16> undef, <8 x i16> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I16 = call <16 x i16> @llvm.umin.v16i16(<16 x i16> undef, <16 x i16> undef)
@@ -246,13 +246,13 @@ define i32 @umin(i32 %arg) {
 ;
 ; SSSE3-LABEL: 'umin'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.umin.i64(i64 undef, i64 undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2I64 = call <2 x i64> @llvm.umin.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4I64 = call <4 x i64> @llvm.umin.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V8I64 = call <8 x i64> @llvm.umin.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2I64 = call <2 x i64> @llvm.umin.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V4I64 = call <4 x i64> @llvm.umin.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V8I64 = call <8 x i64> @llvm.umin.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.umin.i32(i32 undef, i32 undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4I32 = call <4 x i32> @llvm.umin.v4i32(<4 x i32> undef, <4 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I32 = call <8 x i32> @llvm.umin.v8i32(<8 x i32> undef, <8 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16I32 = call <16 x i32> @llvm.umin.v16i32(<16 x i32> undef, <16 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I32 = call <4 x i32> @llvm.umin.v4i32(<4 x i32> undef, <4 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8I32 = call <8 x i32> @llvm.umin.v8i32(<8 x i32> undef, <8 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16I32 = call <16 x i32> @llvm.umin.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I16 = call i16 @llvm.umin.i16(i16 undef, i16 undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I16 = call <8 x i16> @llvm.umin.v8i16(<8 x i16> undef, <8 x i16> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I16 = call <16 x i16> @llvm.umin.v16i16(<16 x i16> undef, <16 x i16> undef)
@@ -265,9 +265,9 @@ define i32 @umin(i32 %arg) {
 ;
 ; SSE42-LABEL: 'umin'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.umin.i64(i64 undef, i64 undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call <2 x i64> @llvm.umin.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4I64 = call <4 x i64> @llvm.umin.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I64 = call <8 x i64> @llvm.umin.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.umin.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call <4 x i64> @llvm.umin.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V8I64 = call <8 x i64> @llvm.umin.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.umin.i32(i32 undef, i32 undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.umin.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I32 = call <8 x i32> @llvm.umin.v8i32(<8 x i32> undef, <8 x i32> undef)
@@ -284,9 +284,9 @@ define i32 @umin(i32 %arg) {
 ;
 ; AVX1-LABEL: 'umin'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.umin.i64(i64 undef, i64 undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call <2 x i64> @llvm.umin.v2i64(<2 x i64> undef, <2 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4I64 = call <4 x i64> @llvm.umin.v4i64(<4 x i64> undef, <4 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I64 = call <8 x i64> @llvm.umin.v8i64(<8 x i64> undef, <8 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.umin.v2i64(<2 x i64> undef, <2 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I64 = call <4 x i64> @llvm.umin.v4i64(<4 x i64> undef, <4 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I64 = call <8 x i64> @llvm.umin.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.umin.i32(i32 undef, i32 undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.umin.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I32 = call <8 x i32> @llvm.umin.v8i32(<8 x i32> undef, <8 x i32> undef)
@@ -303,9 +303,9 @@ define i32 @umin(i32 %arg) {
 ;
 ; AVX2-LABEL: 'umin'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.umin.i64(i64 undef, i64 undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call <2 x i64> @llvm.umin.v2i64(<2 x i64> undef, <2 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I64 = call <4 x i64> @llvm.umin.v4i64(<4 x i64> undef, <4 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I64 = call <8 x i64> @llvm.umin.v8i64(<8 x i64> undef, <8 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.umin.v2i64(<2 x i64> undef, <2 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4I64 = call <4 x i64> @llvm.umin.v4i64(<4 x i64> undef, <4 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I64 = call <8 x i64> @llvm.umin.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.umin.i32(i32 undef, i32 undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.umin.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = call <8 x i32> @llvm.umin.v8i32(<8 x i32> undef, <8 x i32> undef)

diff  --git a/llvm/test/Analysis/CostModel/X86/arith-usat.ll b/llvm/test/Analysis/CostModel/X86/arith-usat.ll
index 8c0e818c20c23..dc9f40caea8f0 100644
--- a/llvm/test/Analysis/CostModel/X86/arith-usat.ll
+++ b/llvm/test/Analysis/CostModel/X86/arith-usat.ll
@@ -38,13 +38,13 @@ declare <64 x i8>  @llvm.uadd.sat.v64i8(<64 x i8>, <64 x i8>)
 define i32 @add(i32 %arg) {
 ; SSSE3-LABEL: 'add'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.uadd.sat.i64(i64 undef, i64 undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I64 = call <2 x i64> @llvm.uadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V4I64 = call <4 x i64> @llvm.uadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V8I64 = call <8 x i64> @llvm.uadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V2I64 = call <2 x i64> @llvm.uadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V4I64 = call <4 x i64> @llvm.uadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V8I64 = call <8 x i64> @llvm.uadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.uadd.sat.i32(i32 undef, i32 undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4I32 = call <4 x i32> @llvm.uadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I32 = call <8 x i32> @llvm.uadd.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I32 = call <16 x i32> @llvm.uadd.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I32 = call <4 x i32> @llvm.uadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I32 = call <8 x i32> @llvm.uadd.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V16I32 = call <16 x i32> @llvm.uadd.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I16 = call i16 @llvm.uadd.sat.i16(i16 undef, i16 undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.uadd.sat.v8i16(<8 x i16> undef, <8 x i16> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = call <16 x i16> @llvm.uadd.sat.v16i16(<16 x i16> undef, <16 x i16> undef)
@@ -57,9 +57,9 @@ define i32 @add(i32 %arg) {
 ;
 ; SSE42-LABEL: 'add'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.uadd.sat.i64(i64 undef, i64 undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2I64 = call <2 x i64> @llvm.uadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4I64 = call <4 x i64> @llvm.uadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8I64 = call <8 x i64> @llvm.uadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.uadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I64 = call <4 x i64> @llvm.uadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I64 = call <8 x i64> @llvm.uadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.uadd.sat.i32(i32 undef, i32 undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4I32 = call <4 x i32> @llvm.uadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I32 = call <8 x i32> @llvm.uadd.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
@@ -76,9 +76,9 @@ define i32 @add(i32 %arg) {
 ;
 ; AVX1-LABEL: 'add'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.uadd.sat.i64(i64 undef, i64 undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2I64 = call <2 x i64> @llvm.uadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I64 = call <4 x i64> @llvm.uadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I64 = call <8 x i64> @llvm.uadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.uadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I64 = call <4 x i64> @llvm.uadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I64 = call <8 x i64> @llvm.uadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.uadd.sat.i32(i32 undef, i32 undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4I32 = call <4 x i32> @llvm.uadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I32 = call <8 x i32> @llvm.uadd.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
@@ -95,9 +95,9 @@ define i32 @add(i32 %arg) {
 ;
 ; AVX2-LABEL: 'add'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.uadd.sat.i64(i64 undef, i64 undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2I64 = call <2 x i64> @llvm.uadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4I64 = call <4 x i64> @llvm.uadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I64 = call <8 x i64> @llvm.uadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.uadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4I64 = call <4 x i64> @llvm.uadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8I64 = call <8 x i64> @llvm.uadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.uadd.sat.i32(i32 undef, i32 undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4I32 = call <4 x i32> @llvm.uadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8I32 = call <8 x i32> @llvm.uadd.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
@@ -171,9 +171,9 @@ define i32 @add(i32 %arg) {
 ;
 ; SLM-LABEL: 'add'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.uadd.sat.i64(i64 undef, i64 undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.uadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I64 = call <4 x i64> @llvm.uadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I64 = call <8 x i64> @llvm.uadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.uadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I64 = call <4 x i64> @llvm.uadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I64 = call <8 x i64> @llvm.uadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.uadd.sat.i32(i32 undef, i32 undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4I32 = call <4 x i32> @llvm.uadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I32 = call <8 x i32> @llvm.uadd.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
@@ -190,9 +190,9 @@ define i32 @add(i32 %arg) {
 ;
 ; GLM-LABEL: 'add'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.uadd.sat.i64(i64 undef, i64 undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2I64 = call <2 x i64> @llvm.uadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4I64 = call <4 x i64> @llvm.uadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8I64 = call <8 x i64> @llvm.uadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.uadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I64 = call <4 x i64> @llvm.uadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I64 = call <8 x i64> @llvm.uadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.uadd.sat.i32(i32 undef, i32 undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4I32 = call <4 x i32> @llvm.uadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I32 = call <8 x i32> @llvm.uadd.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
@@ -209,9 +209,9 @@ define i32 @add(i32 %arg) {
 ;
 ; BTVER2-LABEL: 'add'
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.uadd.sat.i64(i64 undef, i64 undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2I64 = call <2 x i64> @llvm.uadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I64 = call <4 x i64> @llvm.uadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I64 = call <8 x i64> @llvm.uadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.uadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I64 = call <4 x i64> @llvm.uadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I64 = call <8 x i64> @llvm.uadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.uadd.sat.i32(i32 undef, i32 undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4I32 = call <4 x i32> @llvm.uadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I32 = call <8 x i32> @llvm.uadd.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
@@ -272,13 +272,13 @@ declare <64 x i8>  @llvm.usub.sat.v64i8(<64 x i8>, <64 x i8>)
 define i32 @sub(i32 %arg) {
 ; SSSE3-LABEL: 'sub'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.usub.sat.i64(i64 undef, i64 undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.usub.sat.i32(i32 undef, i32 undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4I32 = call <4 x i32> @llvm.usub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I32 = call <8 x i32> @llvm.usub.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I32 = call <16 x i32> @llvm.usub.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I32 = call <4 x i32> @llvm.usub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I32 = call <8 x i32> @llvm.usub.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V16I32 = call <16 x i32> @llvm.usub.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I16 = call i16 @llvm.usub.sat.i16(i16 undef, i16 undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.usub.sat.v8i16(<8 x i16> undef, <8 x i16> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = call <16 x i16> @llvm.usub.sat.v16i16(<16 x i16> undef, <16 x i16> undef)
@@ -291,9 +291,9 @@ define i32 @sub(i32 %arg) {
 ;
 ; SSE42-LABEL: 'sub'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.usub.sat.i64(i64 undef, i64 undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.usub.sat.i32(i32 undef, i32 undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call <4 x i32> @llvm.usub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I32 = call <8 x i32> @llvm.usub.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
@@ -310,9 +310,9 @@ define i32 @sub(i32 %arg) {
 ;
 ; AVX1-LABEL: 'sub'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.usub.sat.i64(i64 undef, i64 undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.usub.sat.i32(i32 undef, i32 undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call <4 x i32> @llvm.usub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I32 = call <8 x i32> @llvm.usub.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
@@ -329,9 +329,9 @@ define i32 @sub(i32 %arg) {
 ;
 ; AVX2-LABEL: 'sub'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.usub.sat.i64(i64 undef, i64 undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.usub.sat.i32(i32 undef, i32 undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call <4 x i32> @llvm.usub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I32 = call <8 x i32> @llvm.usub.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
@@ -405,9 +405,9 @@ define i32 @sub(i32 %arg) {
 ;
 ; SLM-LABEL: 'sub'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.usub.sat.i64(i64 undef, i64 undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.usub.sat.i32(i32 undef, i32 undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call <4 x i32> @llvm.usub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I32 = call <8 x i32> @llvm.usub.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
@@ -424,9 +424,9 @@ define i32 @sub(i32 %arg) {
 ;
 ; GLM-LABEL: 'sub'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.usub.sat.i64(i64 undef, i64 undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.usub.sat.i32(i32 undef, i32 undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call <4 x i32> @llvm.usub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I32 = call <8 x i32> @llvm.usub.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
@@ -443,9 +443,9 @@ define i32 @sub(i32 %arg) {
 ;
 ; BTVER2-LABEL: 'sub'
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.usub.sat.i64(i64 undef, i64 undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.usub.sat.i32(i32 undef, i32 undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call <4 x i32> @llvm.usub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I32 = call <8 x i32> @llvm.usub.sat.v8i32(<8 x i32> undef, <8 x i32> undef)

diff  --git a/llvm/test/Analysis/CostModel/X86/fshl.ll b/llvm/test/Analysis/CostModel/X86/fshl.ll
index 8ea89ff69d0a3..42f5dae616983 100644
--- a/llvm/test/Analysis/CostModel/X86/fshl.ll
+++ b/llvm/test/Analysis/CostModel/X86/fshl.ll
@@ -22,30 +22,30 @@ target triple = "x86_64-apple-macosx10.8.0"
 define void @var_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256, <8 x i64> %a512, i64 %b64, <2 x i64> %b128, <4 x i64> %b256, <8 x i64> %b512, i64 %c64, <2 x i64> %c128, <4 x i64> %c256, <8 x i64> %c512) {
 ; SSSE3-LABEL: 'var_funnel_i64'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshl.i64(i64 %a64, i64 %b64, i64 %c64)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSE42-LABEL: 'var_funnel_i64'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshl.i64(i64 %a64, i64 %b64, i64 %c64)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'var_funnel_i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshl.i64(i64 %a64, i64 %b64, i64 %c64)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'var_funnel_i64'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshl.i64(i64 %a64, i64 %b64, i64 %c64)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512-LABEL: 'var_funnel_i64'
@@ -57,16 +57,16 @@ define void @var_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256, <8 x i64
 ;
 ; SLM-LABEL: 'var_funnel_i64'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshl.i64(i64 %a64, i64 %b64, i64 %c64)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'var_funnel_i64'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshl.i64(i64 %a64, i64 %b64, i64 %c64)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'var_funnel_i64'
@@ -86,30 +86,30 @@ define void @var_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256, <8 x i64
 define void @var_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16 x i32> %a512, i32 %b32, <4 x i32> %b128, <8 x i32> %b256, <16 x i32> %b512, i32 %c32, <4 x i32> %c128, <8 x i32> %c256, <16 x i32> %c512) {
 ; SSSE3-LABEL: 'var_funnel_i32'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %b32, i32 %c32)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 70 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 140 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 152 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSE42-LABEL: 'var_funnel_i32'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %b32, i32 %c32)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 112 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'var_funnel_i32'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %b32, i32 %c32)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 74 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'var_funnel_i32'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %b32, i32 %c32)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512-LABEL: 'var_funnel_i32'
@@ -121,16 +121,16 @@ define void @var_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16 x i3
 ;
 ; SLM-LABEL: 'var_funnel_i32'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %b32, i32 %c32)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 112 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'var_funnel_i32'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %b32, i32 %c32)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 112 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'var_funnel_i32'
@@ -150,37 +150,37 @@ define void @var_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16 x i3
 define void @var_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <32 x i16> %a512, i16 %b16, <8 x i16> %b128, <16 x i16> %b256, <32 x i16> %b512, i16 %c16, <8 x i16> %c128, <16 x i16> %c256, <32 x i16> %c512) {
 ; SSSE3-LABEL: 'var_funnel_i16'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 %c16)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 102 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 204 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSE42-LABEL: 'var_funnel_i16'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 %c16)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'var_funnel_i16'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 %c16)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 108 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 114 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'var_funnel_i16'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 %c16)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512F-LABEL: 'var_funnel_i16'
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 %c16)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512BW-LABEL: 'var_funnel_i16'
@@ -192,23 +192,23 @@ define void @var_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <32 x i
 ;
 ; AVX512DQ-LABEL: 'var_funnel_i16'
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 %c16)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SLM-LABEL: 'var_funnel_i16'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 %c16)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'var_funnel_i16'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 %c16)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'var_funnel_i16'
@@ -228,37 +228,37 @@ define void @var_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <32 x i
 define void @var_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x i8> %a512, i8 %b8, <16 x i8> %b128, <32 x i8> %b256, <64 x i8> %b512, i8 %c8, <16 x i8> %c128, <32 x i8> %c256, <64 x i8> %c512) {
 ; SSSE3-LABEL: 'var_funnel_i8'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 %c8)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 136 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 74 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 148 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSE42-LABEL: 'var_funnel_i8'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 %c8)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'var_funnel_i8'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 %c8)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 122 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'var_funnel_i8'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 %c8)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512F-LABEL: 'var_funnel_i8'
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 %c8)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512BW-LABEL: 'var_funnel_i8'
@@ -270,23 +270,23 @@ define void @var_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x i8> %
 ;
 ; AVX512DQ-LABEL: 'var_funnel_i8'
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 %c8)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SLM-LABEL: 'var_funnel_i8'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 %c8)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'var_funnel_i8'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 %c8)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'var_funnel_i8'
@@ -312,36 +312,36 @@ define void @splatvar_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256, <8
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <2 x i64> %c128, <2 x i64> undef, <2 x i32> zeroinitializer
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <4 x i64> %c256, <4 x i64> undef, <4 x i32> zeroinitializer
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <8 x i64> %c512, <8 x i64> undef, <8 x i32> zeroinitializer
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSE42-LABEL: 'splatvar_funnel_i64'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <2 x i64> %c128, <2 x i64> undef, <2 x i32> zeroinitializer
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <4 x i64> %c256, <4 x i64> undef, <4 x i32> zeroinitializer
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <8 x i64> %c512, <8 x i64> undef, <8 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'splatvar_funnel_i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <2 x i64> %c128, <2 x i64> undef, <2 x i32> zeroinitializer
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %u256 = shufflevector <4 x i64> %c256, <4 x i64> undef, <4 x i32> zeroinitializer
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %u512 = shufflevector <8 x i64> %c512, <8 x i64> undef, <8 x i32> zeroinitializer
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'splatvar_funnel_i64'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <2 x i64> %c128, <2 x i64> undef, <2 x i32> zeroinitializer
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <4 x i64> %c256, <4 x i64> undef, <4 x i32> zeroinitializer
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <8 x i64> %c512, <8 x i64> undef, <8 x i32> zeroinitializer
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512-LABEL: 'splatvar_funnel_i64'
@@ -357,18 +357,18 @@ define void @splatvar_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256, <8
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <2 x i64> %c128, <2 x i64> undef, <2 x i32> zeroinitializer
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <4 x i64> %c256, <4 x i64> undef, <4 x i32> zeroinitializer
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <8 x i64> %c512, <8 x i64> undef, <8 x i32> zeroinitializer
-; SLM-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'splatvar_funnel_i64'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <2 x i64> %c128, <2 x i64> undef, <2 x i32> zeroinitializer
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <4 x i64> %c256, <4 x i64> undef, <4 x i32> zeroinitializer
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <8 x i64> %c512, <8 x i64> undef, <8 x i32> zeroinitializer
-; GLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'splatvar_funnel_i64'
@@ -394,36 +394,36 @@ define void @splatvar_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <4 x i32> %c128, <4 x i32> undef, <4 x i32> zeroinitializer
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <8 x i32> %c256, <8 x i32> undef, <8 x i32> zeroinitializer
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <16 x i32> %c512, <16 x i32> undef, <16 x i32> zeroinitializer
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSE42-LABEL: 'splatvar_funnel_i32'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <4 x i32> %c128, <4 x i32> undef, <4 x i32> zeroinitializer
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <8 x i32> %c256, <8 x i32> undef, <8 x i32> zeroinitializer
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <16 x i32> %c512, <16 x i32> undef, <16 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'splatvar_funnel_i32'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <4 x i32> %c128, <4 x i32> undef, <4 x i32> zeroinitializer
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %u256 = shufflevector <8 x i32> %c256, <8 x i32> undef, <8 x i32> zeroinitializer
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %u512 = shufflevector <16 x i32> %c512, <16 x i32> undef, <16 x i32> zeroinitializer
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'splatvar_funnel_i32'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <4 x i32> %c128, <4 x i32> undef, <4 x i32> zeroinitializer
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <8 x i32> %c256, <8 x i32> undef, <8 x i32> zeroinitializer
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <16 x i32> %c512, <16 x i32> undef, <16 x i32> zeroinitializer
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512-LABEL: 'splatvar_funnel_i32'
@@ -439,18 +439,18 @@ define void @splatvar_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <4 x i32> %c128, <4 x i32> undef, <4 x i32> zeroinitializer
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <8 x i32> %c256, <8 x i32> undef, <8 x i32> zeroinitializer
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <16 x i32> %c512, <16 x i32> undef, <16 x i32> zeroinitializer
-; SLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'splatvar_funnel_i32'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <4 x i32> %c128, <4 x i32> undef, <4 x i32> zeroinitializer
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <8 x i32> %c256, <8 x i32> undef, <8 x i32> zeroinitializer
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <16 x i32> %c512, <16 x i32> undef, <16 x i32> zeroinitializer
-; GLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'splatvar_funnel_i32'
@@ -476,45 +476,45 @@ define void @splatvar_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <3
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <8 x i16> %c128, <8 x i16> undef, <8 x i32> zeroinitializer
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <16 x i16> %c256, <16 x i16> undef, <16 x i32> zeroinitializer
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <32 x i16> %c512, <32 x i16> undef, <32 x i32> zeroinitializer
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSE42-LABEL: 'splatvar_funnel_i16'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <8 x i16> %c128, <8 x i16> undef, <8 x i32> zeroinitializer
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <16 x i16> %c256, <16 x i16> undef, <16 x i32> zeroinitializer
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <32 x i16> %c512, <32 x i16> undef, <32 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'splatvar_funnel_i16'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <8 x i16> %c128, <8 x i16> undef, <8 x i32> zeroinitializer
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %u256 = shufflevector <16 x i16> %c256, <16 x i16> undef, <16 x i32> zeroinitializer
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %u512 = shufflevector <32 x i16> %c512, <32 x i16> undef, <32 x i32> zeroinitializer
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'splatvar_funnel_i16'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <8 x i16> %c128, <8 x i16> undef, <8 x i32> zeroinitializer
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <16 x i16> %c256, <16 x i16> undef, <16 x i32> zeroinitializer
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <32 x i16> %c512, <32 x i16> undef, <32 x i32> zeroinitializer
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512F-LABEL: 'splatvar_funnel_i16'
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <8 x i16> %c128, <8 x i16> undef, <8 x i32> zeroinitializer
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <16 x i16> %c256, <16 x i16> undef, <16 x i32> zeroinitializer
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <32 x i16> %c512, <32 x i16> undef, <32 x i32> zeroinitializer
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512BW-LABEL: 'splatvar_funnel_i16'
@@ -530,27 +530,27 @@ define void @splatvar_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <3
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <8 x i16> %c128, <8 x i16> undef, <8 x i32> zeroinitializer
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <16 x i16> %c256, <16 x i16> undef, <16 x i32> zeroinitializer
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <32 x i16> %c512, <32 x i16> undef, <32 x i32> zeroinitializer
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SLM-LABEL: 'splatvar_funnel_i16'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <8 x i16> %c128, <8 x i16> undef, <8 x i32> zeroinitializer
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <16 x i16> %c256, <16 x i16> undef, <16 x i32> zeroinitializer
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <32 x i16> %c512, <32 x i16> undef, <32 x i32> zeroinitializer
-; SLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'splatvar_funnel_i16'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <8 x i16> %c128, <8 x i16> undef, <8 x i32> zeroinitializer
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <16 x i16> %c256, <16 x i16> undef, <16 x i32> zeroinitializer
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <32 x i16> %c512, <32 x i16> undef, <32 x i32> zeroinitializer
-; GLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'splatvar_funnel_i16'
@@ -576,45 +576,45 @@ define void @splatvar_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <16 x i8> %c128, <16 x i8> undef, <16 x i32> zeroinitializer
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <32 x i8> %c256, <32 x i8> undef, <32 x i32> zeroinitializer
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <64 x i8> %c512, <64 x i8> undef, <64 x i32> zeroinitializer
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 136 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 74 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 148 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSE42-LABEL: 'splatvar_funnel_i8'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <16 x i8> %c128, <16 x i8> undef, <16 x i32> zeroinitializer
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <32 x i8> %c256, <32 x i8> undef, <32 x i32> zeroinitializer
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <64 x i8> %c512, <64 x i8> undef, <64 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'splatvar_funnel_i8'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <16 x i8> %c128, <16 x i8> undef, <16 x i32> zeroinitializer
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %u256 = shufflevector <32 x i8> %c256, <32 x i8> undef, <32 x i32> zeroinitializer
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %u512 = shufflevector <64 x i8> %c512, <64 x i8> undef, <64 x i32> zeroinitializer
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 122 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'splatvar_funnel_i8'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <16 x i8> %c128, <16 x i8> undef, <16 x i32> zeroinitializer
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <32 x i8> %c256, <32 x i8> undef, <32 x i32> zeroinitializer
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <64 x i8> %c512, <64 x i8> undef, <64 x i32> zeroinitializer
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512F-LABEL: 'splatvar_funnel_i8'
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <16 x i8> %c128, <16 x i8> undef, <16 x i32> zeroinitializer
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <32 x i8> %c256, <32 x i8> undef, <32 x i32> zeroinitializer
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <64 x i8> %c512, <64 x i8> undef, <64 x i32> zeroinitializer
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512BW-LABEL: 'splatvar_funnel_i8'
@@ -630,27 +630,27 @@ define void @splatvar_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <16 x i8> %c128, <16 x i8> undef, <16 x i32> zeroinitializer
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <32 x i8> %c256, <32 x i8> undef, <32 x i32> zeroinitializer
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <64 x i8> %c512, <64 x i8> undef, <64 x i32> zeroinitializer
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SLM-LABEL: 'splatvar_funnel_i8'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <16 x i8> %c128, <16 x i8> undef, <16 x i32> zeroinitializer
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <32 x i8> %c256, <32 x i8> undef, <32 x i32> zeroinitializer
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <64 x i8> %c512, <64 x i8> undef, <64 x i32> zeroinitializer
-; SLM-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'splatvar_funnel_i8'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <16 x i8> %c128, <16 x i8> undef, <16 x i32> zeroinitializer
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <32 x i8> %c256, <32 x i8> undef, <32 x i32> zeroinitializer
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <64 x i8> %c512, <64 x i8> undef, <64 x i32> zeroinitializer
-; GLM-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'splatvar_funnel_i8'
@@ -678,30 +678,30 @@ define void @splatvar_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x
 define void @constant_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256, <8 x i64> %a512, i64 %b64, <2 x i64> %b128, <4 x i64> %b256, <8 x i64> %b512) {
 ; SSSE3-LABEL: 'constant_funnel_i64'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshl.i64(i64 %a64, i64 %b64, i64 7)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSE42-LABEL: 'constant_funnel_i64'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshl.i64(i64 %a64, i64 %b64, i64 7)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'constant_funnel_i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshl.i64(i64 %a64, i64 %b64, i64 7)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'constant_funnel_i64'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshl.i64(i64 %a64, i64 %b64, i64 7)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512-LABEL: 'constant_funnel_i64'
@@ -713,16 +713,16 @@ define void @constant_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256, <8
 ;
 ; SLM-LABEL: 'constant_funnel_i64'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshl.i64(i64 %a64, i64 %b64, i64 7)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'constant_funnel_i64'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshl.i64(i64 %a64, i64 %b64, i64 7)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'constant_funnel_i64'
@@ -742,30 +742,30 @@ define void @constant_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256, <8
 define void @constant_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16 x i32> %a512, i32 %b32, <4 x i32> %b128, <8 x i32> %b256, <16 x i32> %b512) {
 ; SSSE3-LABEL: 'constant_funnel_i32'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %b32, i32 7)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 108 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSE42-LABEL: 'constant_funnel_i32'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %b32, i32 7)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'constant_funnel_i32'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %b32, i32 7)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'constant_funnel_i32'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %b32, i32 7)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512-LABEL: 'constant_funnel_i32'
@@ -777,16 +777,16 @@ define void @constant_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16
 ;
 ; SLM-LABEL: 'constant_funnel_i32'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %b32, i32 7)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'constant_funnel_i32'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %b32, i32 7)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'constant_funnel_i32'
@@ -806,37 +806,37 @@ define void @constant_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16
 define void @constant_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <32 x i16> %a512, i16 %b16, <8 x i16> %b128, <16 x i16> %b256, <32 x i16> %b512) {
 ; SSSE3-LABEL: 'constant_funnel_i16'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 7)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 92 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSE42-LABEL: 'constant_funnel_i16'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 7)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'constant_funnel_i16'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 7)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 94 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'constant_funnel_i16'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 7)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512F-LABEL: 'constant_funnel_i16'
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 7)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512BW-LABEL: 'constant_funnel_i16'
@@ -848,23 +848,23 @@ define void @constant_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <3
 ;
 ; AVX512DQ-LABEL: 'constant_funnel_i16'
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 7)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SLM-LABEL: 'constant_funnel_i16'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 7)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'constant_funnel_i16'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 7)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'constant_funnel_i16'
@@ -884,37 +884,37 @@ define void @constant_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <3
 define void @constant_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x i8> %a512, i8 %b8, <16 x i8> %b128, <32 x i8> %b256, <64 x i8> %b512) {
 ; SSSE3-LABEL: 'constant_funnel_i8'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 7)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 144 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSE42-LABEL: 'constant_funnel_i8'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 7)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 112 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'constant_funnel_i8'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 7)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 114 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 120 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'constant_funnel_i8'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 7)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512F-LABEL: 'constant_funnel_i8'
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 7)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512BW-LABEL: 'constant_funnel_i8'
@@ -926,23 +926,23 @@ define void @constant_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x
 ;
 ; AVX512DQ-LABEL: 'constant_funnel_i8'
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 7)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SLM-LABEL: 'constant_funnel_i8'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 7)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 112 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'constant_funnel_i8'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 7)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 112 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'constant_funnel_i8'
@@ -966,30 +966,30 @@ define void @constant_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x
 define void @splatconstant_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256, <8 x i64> %a512, i64 %b64, <2 x i64> %b128, <4 x i64> %b256, <8 x i64> %b512) {
 ; SSSE3-LABEL: 'splatconstant_funnel_i64'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshl.i64(i64 %a64, i64 %b64, i64 7)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSE42-LABEL: 'splatconstant_funnel_i64'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshl.i64(i64 %a64, i64 %b64, i64 7)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'splatconstant_funnel_i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshl.i64(i64 %a64, i64 %b64, i64 7)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'splatconstant_funnel_i64'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshl.i64(i64 %a64, i64 %b64, i64 7)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512-LABEL: 'splatconstant_funnel_i64'
@@ -1001,16 +1001,16 @@ define void @splatconstant_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256
 ;
 ; SLM-LABEL: 'splatconstant_funnel_i64'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshl.i64(i64 %a64, i64 %b64, i64 7)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'splatconstant_funnel_i64'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshl.i64(i64 %a64, i64 %b64, i64 7)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'splatconstant_funnel_i64'
@@ -1030,30 +1030,30 @@ define void @splatconstant_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256
 define void @splatconstant_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16 x i32> %a512, i32 %b32, <4 x i32> %b128, <8 x i32> %b256, <16 x i32> %b512) {
 ; SSSE3-LABEL: 'splatconstant_funnel_i32'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %b32, i32 5)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSE42-LABEL: 'splatconstant_funnel_i32'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %b32, i32 5)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'splatconstant_funnel_i32'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %b32, i32 5)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'splatconstant_funnel_i32'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %b32, i32 5)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512-LABEL: 'splatconstant_funnel_i32'
@@ -1065,16 +1065,16 @@ define void @splatconstant_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256
 ;
 ; SLM-LABEL: 'splatconstant_funnel_i32'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %b32, i32 5)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'splatconstant_funnel_i32'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %b32, i32 5)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'splatconstant_funnel_i32'
@@ -1094,37 +1094,37 @@ define void @splatconstant_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256
 define void @splatconstant_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <32 x i16> %a512, i16 %b16, <8 x i16> %b128, <16 x i16> %b256, <32 x i16> %b512) {
 ; SSSE3-LABEL: 'splatconstant_funnel_i16'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 3)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSE42-LABEL: 'splatconstant_funnel_i16'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 3)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'splatconstant_funnel_i16'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 3)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'splatconstant_funnel_i16'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 3)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512F-LABEL: 'splatconstant_funnel_i16'
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 3)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512BW-LABEL: 'splatconstant_funnel_i16'
@@ -1136,23 +1136,23 @@ define void @splatconstant_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a25
 ;
 ; AVX512DQ-LABEL: 'splatconstant_funnel_i16'
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 3)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SLM-LABEL: 'splatconstant_funnel_i16'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 3)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'splatconstant_funnel_i16'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 3)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'splatconstant_funnel_i16'
@@ -1172,37 +1172,37 @@ define void @splatconstant_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a25
 define void @splatconstant_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x i8> %a512, i8 %b8, <16 x i8> %b128, <32 x i8> %b256, <64 x i8> %b512) {
 ; SSSE3-LABEL: 'splatconstant_funnel_i8'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 3)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSE42-LABEL: 'splatconstant_funnel_i8'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 3)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'splatconstant_funnel_i8'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 3)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'splatconstant_funnel_i8'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 3)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512F-LABEL: 'splatconstant_funnel_i8'
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 3)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512BW-LABEL: 'splatconstant_funnel_i8'
@@ -1214,23 +1214,23 @@ define void @splatconstant_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <
 ;
 ; AVX512DQ-LABEL: 'splatconstant_funnel_i8'
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 3)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SLM-LABEL: 'splatconstant_funnel_i8'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 3)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'splatconstant_funnel_i8'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 3)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'splatconstant_funnel_i8'

diff  --git a/llvm/test/Analysis/CostModel/X86/fshr.ll b/llvm/test/Analysis/CostModel/X86/fshr.ll
index 9d4f529be0621..5407fbc45628b 100644
--- a/llvm/test/Analysis/CostModel/X86/fshr.ll
+++ b/llvm/test/Analysis/CostModel/X86/fshr.ll
@@ -22,30 +22,30 @@ target triple = "x86_64-apple-macosx10.8.0"
 define void @var_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256, <8 x i64> %a512, i64 %b64, <2 x i64> %b128, <4 x i64> %b256, <8 x i64> %b512, i64 %c64, <2 x i64> %c128, <4 x i64> %c256, <8 x i64> %c512) {
 ; SSSE3-LABEL: 'var_funnel_i64'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshr.i64(i64 %a64, i64 %b64, i64 %c64)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSE42-LABEL: 'var_funnel_i64'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshr.i64(i64 %a64, i64 %b64, i64 %c64)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'var_funnel_i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshr.i64(i64 %a64, i64 %b64, i64 %c64)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'var_funnel_i64'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshr.i64(i64 %a64, i64 %b64, i64 %c64)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512-LABEL: 'var_funnel_i64'
@@ -57,16 +57,16 @@ define void @var_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256, <8 x i64
 ;
 ; SLM-LABEL: 'var_funnel_i64'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshr.i64(i64 %a64, i64 %b64, i64 %c64)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'var_funnel_i64'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshr.i64(i64 %a64, i64 %b64, i64 %c64)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'var_funnel_i64'
@@ -86,30 +86,30 @@ define void @var_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256, <8 x i64
 define void @var_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16 x i32> %a512, i32 %b32, <4 x i32> %b128, <8 x i32> %b256, <16 x i32> %b512, i32 %c32, <4 x i32> %c128, <8 x i32> %c256, <16 x i32> %c512) {
 ; SSSE3-LABEL: 'var_funnel_i32'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %b32, i32 %c32)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 70 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 140 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 152 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSE42-LABEL: 'var_funnel_i32'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %b32, i32 %c32)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 112 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'var_funnel_i32'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %b32, i32 %c32)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 74 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'var_funnel_i32'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %b32, i32 %c32)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512-LABEL: 'var_funnel_i32'
@@ -121,16 +121,16 @@ define void @var_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16 x i3
 ;
 ; SLM-LABEL: 'var_funnel_i32'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %b32, i32 %c32)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 112 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'var_funnel_i32'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %b32, i32 %c32)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 112 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'var_funnel_i32'
@@ -150,37 +150,37 @@ define void @var_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16 x i3
 define void @var_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <32 x i16> %a512, i16 %b16, <8 x i16> %b128, <16 x i16> %b256, <32 x i16> %b512, i16 %c16, <8 x i16> %c128, <16 x i16> %c256, <32 x i16> %c512) {
 ; SSSE3-LABEL: 'var_funnel_i16'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 %c16)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 51 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 102 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 204 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSE42-LABEL: 'var_funnel_i16'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 %c16)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'var_funnel_i16'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 %c16)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 108 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 114 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'var_funnel_i16'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 %c16)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512F-LABEL: 'var_funnel_i16'
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 %c16)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512BW-LABEL: 'var_funnel_i16'
@@ -192,23 +192,23 @@ define void @var_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <32 x i
 ;
 ; AVX512DQ-LABEL: 'var_funnel_i16'
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 %c16)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SLM-LABEL: 'var_funnel_i16'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 %c16)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'var_funnel_i16'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 %c16)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'var_funnel_i16'
@@ -228,37 +228,37 @@ define void @var_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <32 x i
 define void @var_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x i8> %a512, i8 %b8, <16 x i8> %b128, <32 x i8> %b256, <64 x i8> %b512, i8 %c8, <16 x i8> %c128, <32 x i8> %c256, <64 x i8> %c512) {
 ; SSSE3-LABEL: 'var_funnel_i8'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 %c8)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 136 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 74 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 148 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSE42-LABEL: 'var_funnel_i8'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 %c8)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'var_funnel_i8'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 %c8)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 122 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'var_funnel_i8'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 %c8)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512F-LABEL: 'var_funnel_i8'
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 %c8)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512BW-LABEL: 'var_funnel_i8'
@@ -270,23 +270,23 @@ define void @var_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x i8> %
 ;
 ; AVX512DQ-LABEL: 'var_funnel_i8'
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 %c8)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SLM-LABEL: 'var_funnel_i8'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 %c8)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'var_funnel_i8'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 %c8)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'var_funnel_i8'
@@ -312,36 +312,36 @@ define void @splatvar_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256, <8
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <2 x i64> %c128, <2 x i64> undef, <2 x i32> zeroinitializer
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <4 x i64> %c256, <4 x i64> undef, <4 x i32> zeroinitializer
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <8 x i64> %c512, <8 x i64> undef, <8 x i32> zeroinitializer
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSE42-LABEL: 'splatvar_funnel_i64'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <2 x i64> %c128, <2 x i64> undef, <2 x i32> zeroinitializer
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <4 x i64> %c256, <4 x i64> undef, <4 x i32> zeroinitializer
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <8 x i64> %c512, <8 x i64> undef, <8 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'splatvar_funnel_i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <2 x i64> %c128, <2 x i64> undef, <2 x i32> zeroinitializer
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %u256 = shufflevector <4 x i64> %c256, <4 x i64> undef, <4 x i32> zeroinitializer
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %u512 = shufflevector <8 x i64> %c512, <8 x i64> undef, <8 x i32> zeroinitializer
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'splatvar_funnel_i64'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <2 x i64> %c128, <2 x i64> undef, <2 x i32> zeroinitializer
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <4 x i64> %c256, <4 x i64> undef, <4 x i32> zeroinitializer
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <8 x i64> %c512, <8 x i64> undef, <8 x i32> zeroinitializer
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512-LABEL: 'splatvar_funnel_i64'
@@ -357,18 +357,18 @@ define void @splatvar_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256, <8
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <2 x i64> %c128, <2 x i64> undef, <2 x i32> zeroinitializer
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <4 x i64> %c256, <4 x i64> undef, <4 x i32> zeroinitializer
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <8 x i64> %c512, <8 x i64> undef, <8 x i32> zeroinitializer
-; SLM-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'splatvar_funnel_i64'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <2 x i64> %c128, <2 x i64> undef, <2 x i32> zeroinitializer
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <4 x i64> %c256, <4 x i64> undef, <4 x i32> zeroinitializer
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <8 x i64> %c512, <8 x i64> undef, <8 x i32> zeroinitializer
-; GLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'splatvar_funnel_i64'
@@ -394,36 +394,36 @@ define void @splatvar_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <4 x i32> %c128, <4 x i32> undef, <4 x i32> zeroinitializer
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <8 x i32> %c256, <8 x i32> undef, <8 x i32> zeroinitializer
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <16 x i32> %c512, <16 x i32> undef, <16 x i32> zeroinitializer
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSE42-LABEL: 'splatvar_funnel_i32'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <4 x i32> %c128, <4 x i32> undef, <4 x i32> zeroinitializer
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <8 x i32> %c256, <8 x i32> undef, <8 x i32> zeroinitializer
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <16 x i32> %c512, <16 x i32> undef, <16 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'splatvar_funnel_i32'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <4 x i32> %c128, <4 x i32> undef, <4 x i32> zeroinitializer
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %u256 = shufflevector <8 x i32> %c256, <8 x i32> undef, <8 x i32> zeroinitializer
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %u512 = shufflevector <16 x i32> %c512, <16 x i32> undef, <16 x i32> zeroinitializer
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'splatvar_funnel_i32'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <4 x i32> %c128, <4 x i32> undef, <4 x i32> zeroinitializer
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <8 x i32> %c256, <8 x i32> undef, <8 x i32> zeroinitializer
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <16 x i32> %c512, <16 x i32> undef, <16 x i32> zeroinitializer
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512-LABEL: 'splatvar_funnel_i32'
@@ -439,18 +439,18 @@ define void @splatvar_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <4 x i32> %c128, <4 x i32> undef, <4 x i32> zeroinitializer
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <8 x i32> %c256, <8 x i32> undef, <8 x i32> zeroinitializer
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <16 x i32> %c512, <16 x i32> undef, <16 x i32> zeroinitializer
-; SLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'splatvar_funnel_i32'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <4 x i32> %c128, <4 x i32> undef, <4 x i32> zeroinitializer
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <8 x i32> %c256, <8 x i32> undef, <8 x i32> zeroinitializer
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <16 x i32> %c512, <16 x i32> undef, <16 x i32> zeroinitializer
-; GLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'splatvar_funnel_i32'
@@ -476,45 +476,45 @@ define void @splatvar_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <3
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <8 x i16> %c128, <8 x i16> undef, <8 x i32> zeroinitializer
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <16 x i16> %c256, <16 x i16> undef, <16 x i32> zeroinitializer
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <32 x i16> %c512, <32 x i16> undef, <32 x i32> zeroinitializer
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSE42-LABEL: 'splatvar_funnel_i16'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <8 x i16> %c128, <8 x i16> undef, <8 x i32> zeroinitializer
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <16 x i16> %c256, <16 x i16> undef, <16 x i32> zeroinitializer
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <32 x i16> %c512, <32 x i16> undef, <32 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'splatvar_funnel_i16'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <8 x i16> %c128, <8 x i16> undef, <8 x i32> zeroinitializer
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %u256 = shufflevector <16 x i16> %c256, <16 x i16> undef, <16 x i32> zeroinitializer
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %u512 = shufflevector <32 x i16> %c512, <32 x i16> undef, <32 x i32> zeroinitializer
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'splatvar_funnel_i16'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <8 x i16> %c128, <8 x i16> undef, <8 x i32> zeroinitializer
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <16 x i16> %c256, <16 x i16> undef, <16 x i32> zeroinitializer
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <32 x i16> %c512, <32 x i16> undef, <32 x i32> zeroinitializer
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512F-LABEL: 'splatvar_funnel_i16'
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <8 x i16> %c128, <8 x i16> undef, <8 x i32> zeroinitializer
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <16 x i16> %c256, <16 x i16> undef, <16 x i32> zeroinitializer
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <32 x i16> %c512, <32 x i16> undef, <32 x i32> zeroinitializer
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512BW-LABEL: 'splatvar_funnel_i16'
@@ -530,27 +530,27 @@ define void @splatvar_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <3
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <8 x i16> %c128, <8 x i16> undef, <8 x i32> zeroinitializer
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <16 x i16> %c256, <16 x i16> undef, <16 x i32> zeroinitializer
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <32 x i16> %c512, <32 x i16> undef, <32 x i32> zeroinitializer
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SLM-LABEL: 'splatvar_funnel_i16'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <8 x i16> %c128, <8 x i16> undef, <8 x i32> zeroinitializer
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <16 x i16> %c256, <16 x i16> undef, <16 x i32> zeroinitializer
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <32 x i16> %c512, <32 x i16> undef, <32 x i32> zeroinitializer
-; SLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'splatvar_funnel_i16'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <8 x i16> %c128, <8 x i16> undef, <8 x i32> zeroinitializer
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <16 x i16> %c256, <16 x i16> undef, <16 x i32> zeroinitializer
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <32 x i16> %c512, <32 x i16> undef, <32 x i32> zeroinitializer
-; GLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'splatvar_funnel_i16'
@@ -576,45 +576,45 @@ define void @splatvar_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <16 x i8> %c128, <16 x i8> undef, <16 x i32> zeroinitializer
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <32 x i8> %c256, <32 x i8> undef, <32 x i32> zeroinitializer
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <64 x i8> %c512, <64 x i8> undef, <64 x i32> zeroinitializer
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 136 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 74 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 148 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSE42-LABEL: 'splatvar_funnel_i8'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <16 x i8> %c128, <16 x i8> undef, <16 x i32> zeroinitializer
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <32 x i8> %c256, <32 x i8> undef, <32 x i32> zeroinitializer
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <64 x i8> %c512, <64 x i8> undef, <64 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'splatvar_funnel_i8'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <16 x i8> %c128, <16 x i8> undef, <16 x i32> zeroinitializer
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %u256 = shufflevector <32 x i8> %c256, <32 x i8> undef, <32 x i32> zeroinitializer
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %u512 = shufflevector <64 x i8> %c512, <64 x i8> undef, <64 x i32> zeroinitializer
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 61 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 122 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'splatvar_funnel_i8'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <16 x i8> %c128, <16 x i8> undef, <16 x i32> zeroinitializer
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <32 x i8> %c256, <32 x i8> undef, <32 x i32> zeroinitializer
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <64 x i8> %c512, <64 x i8> undef, <64 x i32> zeroinitializer
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512F-LABEL: 'splatvar_funnel_i8'
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <16 x i8> %c128, <16 x i8> undef, <16 x i32> zeroinitializer
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <32 x i8> %c256, <32 x i8> undef, <32 x i32> zeroinitializer
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <64 x i8> %c512, <64 x i8> undef, <64 x i32> zeroinitializer
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512BW-LABEL: 'splatvar_funnel_i8'
@@ -630,27 +630,27 @@ define void @splatvar_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <16 x i8> %c128, <16 x i8> undef, <16 x i32> zeroinitializer
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <32 x i8> %c256, <32 x i8> undef, <32 x i32> zeroinitializer
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <64 x i8> %c512, <64 x i8> undef, <64 x i32> zeroinitializer
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SLM-LABEL: 'splatvar_funnel_i8'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <16 x i8> %c128, <16 x i8> undef, <16 x i32> zeroinitializer
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <32 x i8> %c256, <32 x i8> undef, <32 x i32> zeroinitializer
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <64 x i8> %c512, <64 x i8> undef, <64 x i32> zeroinitializer
-; SLM-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'splatvar_funnel_i8'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <16 x i8> %c128, <16 x i8> undef, <16 x i32> zeroinitializer
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <32 x i8> %c256, <32 x i8> undef, <32 x i32> zeroinitializer
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <64 x i8> %c512, <64 x i8> undef, <64 x i32> zeroinitializer
-; GLM-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'splatvar_funnel_i8'
@@ -678,30 +678,30 @@ define void @splatvar_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x
 define void @constant_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256, <8 x i64> %a512, i64 %b64, <2 x i64> %b128, <4 x i64> %b256, <8 x i64> %b512) {
 ; SSSE3-LABEL: 'constant_funnel_i64'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshr.i64(i64 %a64, i64 %b64, i64 7)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSE42-LABEL: 'constant_funnel_i64'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshr.i64(i64 %a64, i64 %b64, i64 7)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'constant_funnel_i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshr.i64(i64 %a64, i64 %b64, i64 7)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'constant_funnel_i64'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshr.i64(i64 %a64, i64 %b64, i64 7)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512-LABEL: 'constant_funnel_i64'
@@ -713,16 +713,16 @@ define void @constant_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256, <8
 ;
 ; SLM-LABEL: 'constant_funnel_i64'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshr.i64(i64 %a64, i64 %b64, i64 7)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'constant_funnel_i64'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshr.i64(i64 %a64, i64 %b64, i64 7)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'constant_funnel_i64'
@@ -742,30 +742,30 @@ define void @constant_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256, <8
 define void @constant_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16 x i32> %a512, i32 %b32, <4 x i32> %b128, <8 x i32> %b256, <16 x i32> %b512) {
 ; SSSE3-LABEL: 'constant_funnel_i32'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %b32, i32 7)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 108 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSE42-LABEL: 'constant_funnel_i32'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %b32, i32 7)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'constant_funnel_i32'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %b32, i32 7)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'constant_funnel_i32'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %b32, i32 7)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512-LABEL: 'constant_funnel_i32'
@@ -777,16 +777,16 @@ define void @constant_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16
 ;
 ; SLM-LABEL: 'constant_funnel_i32'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %b32, i32 7)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'constant_funnel_i32'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %b32, i32 7)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'constant_funnel_i32'
@@ -806,37 +806,37 @@ define void @constant_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16
 define void @constant_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <32 x i16> %a512, i16 %b16, <8 x i16> %b128, <16 x i16> %b256, <32 x i16> %b512) {
 ; SSSE3-LABEL: 'constant_funnel_i16'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 7)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 92 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSE42-LABEL: 'constant_funnel_i16'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 7)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'constant_funnel_i16'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 7)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 94 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'constant_funnel_i16'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 7)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512F-LABEL: 'constant_funnel_i16'
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 7)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512BW-LABEL: 'constant_funnel_i16'
@@ -848,23 +848,23 @@ define void @constant_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <3
 ;
 ; AVX512DQ-LABEL: 'constant_funnel_i16'
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 7)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SLM-LABEL: 'constant_funnel_i16'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 7)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'constant_funnel_i16'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 7)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'constant_funnel_i16'
@@ -884,37 +884,37 @@ define void @constant_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <3
 define void @constant_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x i8> %a512, i8 %b8, <16 x i8> %b128, <32 x i8> %b256, <64 x i8> %b512) {
 ; SSSE3-LABEL: 'constant_funnel_i8'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 7)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 144 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSE42-LABEL: 'constant_funnel_i8'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 7)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 112 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'constant_funnel_i8'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 7)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 114 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 120 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'constant_funnel_i8'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 7)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512F-LABEL: 'constant_funnel_i8'
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 7)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512BW-LABEL: 'constant_funnel_i8'
@@ -926,23 +926,23 @@ define void @constant_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x
 ;
 ; AVX512DQ-LABEL: 'constant_funnel_i8'
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 7)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SLM-LABEL: 'constant_funnel_i8'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 7)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 112 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'constant_funnel_i8'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 7)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 112 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'constant_funnel_i8'
@@ -966,30 +966,30 @@ define void @constant_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x
 define void @splatconstant_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256, <8 x i64> %a512, i64 %b64, <2 x i64> %b128, <4 x i64> %b256, <8 x i64> %b512) {
 ; SSSE3-LABEL: 'splatconstant_funnel_i64'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshr.i64(i64 %a64, i64 %b64, i64 7)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSE42-LABEL: 'splatconstant_funnel_i64'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshr.i64(i64 %a64, i64 %b64, i64 7)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'splatconstant_funnel_i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshr.i64(i64 %a64, i64 %b64, i64 7)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'splatconstant_funnel_i64'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshr.i64(i64 %a64, i64 %b64, i64 7)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512-LABEL: 'splatconstant_funnel_i64'
@@ -1001,16 +1001,16 @@ define void @splatconstant_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256
 ;
 ; SLM-LABEL: 'splatconstant_funnel_i64'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshr.i64(i64 %a64, i64 %b64, i64 7)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'splatconstant_funnel_i64'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshr.i64(i64 %a64, i64 %b64, i64 7)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'splatconstant_funnel_i64'
@@ -1030,30 +1030,30 @@ define void @splatconstant_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256
 define void @splatconstant_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16 x i32> %a512, i32 %b32, <4 x i32> %b128, <8 x i32> %b256, <16 x i32> %b512) {
 ; SSSE3-LABEL: 'splatconstant_funnel_i32'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %b32, i32 5)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSE42-LABEL: 'splatconstant_funnel_i32'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %b32, i32 5)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'splatconstant_funnel_i32'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %b32, i32 5)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'splatconstant_funnel_i32'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %b32, i32 5)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512-LABEL: 'splatconstant_funnel_i32'
@@ -1065,16 +1065,16 @@ define void @splatconstant_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256
 ;
 ; SLM-LABEL: 'splatconstant_funnel_i32'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %b32, i32 5)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'splatconstant_funnel_i32'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %b32, i32 5)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'splatconstant_funnel_i32'
@@ -1094,37 +1094,37 @@ define void @splatconstant_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256
 define void @splatconstant_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <32 x i16> %a512, i16 %b16, <8 x i16> %b128, <16 x i16> %b256, <32 x i16> %b512) {
 ; SSSE3-LABEL: 'splatconstant_funnel_i16'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 3)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSE42-LABEL: 'splatconstant_funnel_i16'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 3)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'splatconstant_funnel_i16'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 3)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'splatconstant_funnel_i16'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 3)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512F-LABEL: 'splatconstant_funnel_i16'
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 3)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512BW-LABEL: 'splatconstant_funnel_i16'
@@ -1136,23 +1136,23 @@ define void @splatconstant_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a25
 ;
 ; AVX512DQ-LABEL: 'splatconstant_funnel_i16'
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 3)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SLM-LABEL: 'splatconstant_funnel_i16'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 3)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'splatconstant_funnel_i16'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 3)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'splatconstant_funnel_i16'
@@ -1172,37 +1172,37 @@ define void @splatconstant_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a25
 define void @splatconstant_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x i8> %a512, i8 %b8, <16 x i8> %b128, <32 x i8> %b256, <64 x i8> %b512) {
 ; SSSE3-LABEL: 'splatconstant_funnel_i8'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 3)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SSE42-LABEL: 'splatconstant_funnel_i8'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 3)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'splatconstant_funnel_i8'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 3)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'splatconstant_funnel_i8'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 3)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512F-LABEL: 'splatconstant_funnel_i8'
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 3)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX512BW-LABEL: 'splatconstant_funnel_i8'
@@ -1214,23 +1214,23 @@ define void @splatconstant_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <
 ;
 ; AVX512DQ-LABEL: 'splatconstant_funnel_i8'
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 3)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; SLM-LABEL: 'splatconstant_funnel_i8'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 3)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'splatconstant_funnel_i8'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 3)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'splatconstant_funnel_i8'

diff  --git a/llvm/test/Analysis/CostModel/X86/intrinsic-cost-kinds.ll b/llvm/test/Analysis/CostModel/X86/intrinsic-cost-kinds.ll
index 3d173e36697f3..46ee566eb7736 100644
--- a/llvm/test/Analysis/CostModel/X86/intrinsic-cost-kinds.ll
+++ b/llvm/test/Analysis/CostModel/X86/intrinsic-cost-kinds.ll
@@ -48,7 +48,7 @@ declare void @llvm.memcpy.p0i8.p0i8.i32(i8*, i8*, i32, i1)
 define void @umul(i32 %a, i32 %b, <16 x i32> %va, <16 x i32> %vb) {
 ; THRU-LABEL: 'umul'
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %s = call { i32, i1 } @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
-; THRU-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %v = call { <16 x i32>, <16 x i1> } @llvm.umul.with.overflow.v16i32(<16 x i32> %va, <16 x i32> %vb)
+; THRU-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %v = call { <16 x i32>, <16 x i1> } @llvm.umul.with.overflow.v16i32(<16 x i32> %va, <16 x i32> %vb)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; LATE-LABEL: 'umul'
@@ -74,7 +74,7 @@ define void @umul(i32 %a, i32 %b, <16 x i32> %va, <16 x i32> %vb) {
 define void @smax(i32 %a, i32 %b, <16 x i32> %va, <16 x i32> %vb) {
 ; THRU-LABEL: 'smax'
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %s = call i32 @llvm.smax.i32(i32 %a, i32 %b)
-; THRU-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %v = call <16 x i32> @llvm.smax.v16i32(<16 x i32> %va, <16 x i32> %vb)
+; THRU-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %v = call <16 x i32> @llvm.smax.v16i32(<16 x i32> %va, <16 x i32> %vb)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; LATE-LABEL: 'smax'
@@ -256,7 +256,7 @@ define void @ctlz(i32 %a, <16 x i32> %va) {
 define void @fshl(i32 %a, i32 %b, i32 %c, <16 x i32> %va, <16 x i32> %vb, <16 x i32> %vc) {
 ; THRU-LABEL: 'fshl'
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %s = call i32 @llvm.fshl.i32(i32 %a, i32 %b, i32 %c)
-; THRU-NEXT:  Cost Model: Found an estimated cost of 140 for instruction: %v = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %va, <16 x i32> %vb, <16 x i32> %vc)
+; THRU-NEXT:  Cost Model: Found an estimated cost of 152 for instruction: %v = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %va, <16 x i32> %vb, <16 x i32> %vc)
 ; THRU-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; LATE-LABEL: 'fshl'

diff  --git a/llvm/test/Analysis/CostModel/X86/reduce-smax.ll b/llvm/test/Analysis/CostModel/X86/reduce-smax.ll
index ee3cffa69896e..1e5096c45032b 100644
--- a/llvm/test/Analysis/CostModel/X86/reduce-smax.ll
+++ b/llvm/test/Analysis/CostModel/X86/reduce-smax.ll
@@ -12,50 +12,50 @@
 define i32 @reduce_i64(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i64'
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.vector.reduce.smax.v1i64(<1 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V2 = call i64 @llvm.vector.reduce.smax.v2i64(<2 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V4 = call i64 @llvm.vector.reduce.smax.v4i64(<4 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V8 = call i64 @llvm.vector.reduce.smax.v8i64(<8 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 90 for instruction: %V16 = call i64 @llvm.vector.reduce.smax.v16i64(<16 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V2 = call i64 @llvm.vector.reduce.smax.v2i64(<2 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V4 = call i64 @llvm.vector.reduce.smax.v4i64(<4 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V8 = call i64 @llvm.vector.reduce.smax.v8i64(<8 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 114 for instruction: %V16 = call i64 @llvm.vector.reduce.smax.v16i64(<16 x i64> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i64'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.vector.reduce.smax.v1i64(<1 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V2 = call i64 @llvm.vector.reduce.smax.v2i64(<2 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V4 = call i64 @llvm.vector.reduce.smax.v4i64(<4 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V8 = call i64 @llvm.vector.reduce.smax.v8i64(<8 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 90 for instruction: %V16 = call i64 @llvm.vector.reduce.smax.v16i64(<16 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V2 = call i64 @llvm.vector.reduce.smax.v2i64(<2 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V4 = call i64 @llvm.vector.reduce.smax.v4i64(<4 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V8 = call i64 @llvm.vector.reduce.smax.v8i64(<8 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 114 for instruction: %V16 = call i64 @llvm.vector.reduce.smax.v16i64(<16 x i64> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE41-LABEL: 'reduce_i64'
 ; SSE41-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.vector.reduce.smax.v1i64(<1 x i64> undef)
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2 = call i64 @llvm.vector.reduce.smax.v2i64(<2 x i64> undef)
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4 = call i64 @llvm.vector.reduce.smax.v4i64(<4 x i64> undef)
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V8 = call i64 @llvm.vector.reduce.smax.v8i64(<8 x i64> undef)
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 74 for instruction: %V16 = call i64 @llvm.vector.reduce.smax.v16i64(<16 x i64> undef)
+; SSE41-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2 = call i64 @llvm.vector.reduce.smax.v2i64(<2 x i64> undef)
+; SSE41-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V4 = call i64 @llvm.vector.reduce.smax.v4i64(<4 x i64> undef)
+; SSE41-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V8 = call i64 @llvm.vector.reduce.smax.v8i64(<8 x i64> undef)
+; SSE41-NEXT:  Cost Model: Found an estimated cost of 98 for instruction: %V16 = call i64 @llvm.vector.reduce.smax.v16i64(<16 x i64> undef)
 ; SSE41-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'reduce_i64'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.vector.reduce.smax.v1i64(<1 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i64 @llvm.vector.reduce.smax.v2i64(<2 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4 = call i64 @llvm.vector.reduce.smax.v4i64(<4 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8 = call i64 @llvm.vector.reduce.smax.v8i64(<8 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V16 = call i64 @llvm.vector.reduce.smax.v16i64(<16 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2 = call i64 @llvm.vector.reduce.smax.v2i64(<2 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4 = call i64 @llvm.vector.reduce.smax.v4i64(<4 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V8 = call i64 @llvm.vector.reduce.smax.v8i64(<8 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V16 = call i64 @llvm.vector.reduce.smax.v16i64(<16 x i64> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX1-LABEL: 'reduce_i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.vector.reduce.smax.v1i64(<1 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i64 @llvm.vector.reduce.smax.v2i64(<2 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4 = call i64 @llvm.vector.reduce.smax.v4i64(<4 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8 = call i64 @llvm.vector.reduce.smax.v8i64(<8 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V16 = call i64 @llvm.vector.reduce.smax.v16i64(<16 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2 = call i64 @llvm.vector.reduce.smax.v2i64(<2 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4 = call i64 @llvm.vector.reduce.smax.v4i64(<4 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V8 = call i64 @llvm.vector.reduce.smax.v8i64(<8 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V16 = call i64 @llvm.vector.reduce.smax.v16i64(<16 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'reduce_i64'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.vector.reduce.smax.v1i64(<1 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i64 @llvm.vector.reduce.smax.v2i64(<2 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4 = call i64 @llvm.vector.reduce.smax.v4i64(<4 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8 = call i64 @llvm.vector.reduce.smax.v8i64(<8 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V16 = call i64 @llvm.vector.reduce.smax.v16i64(<16 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2 = call i64 @llvm.vector.reduce.smax.v2i64(<2 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4 = call i64 @llvm.vector.reduce.smax.v4i64(<4 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8 = call i64 @llvm.vector.reduce.smax.v8i64(<8 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16 = call i64 @llvm.vector.reduce.smax.v16i64(<16 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512-LABEL: 'reduce_i64'
@@ -76,19 +76,19 @@ define i32 @reduce_i64(i32 %arg) {
 
 define i32 @reduce_i32(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2 = call i32 @llvm.vector.reduce.smax.v2i32(<2 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4 = call i32 @llvm.vector.reduce.smax.v4i32(<4 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8 = call i32 @llvm.vector.reduce.smax.v8i32(<8 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V16 = call i32 @llvm.vector.reduce.smax.v16i32(<16 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %V32 = call i32 @llvm.vector.reduce.smax.v32i32(<32 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2 = call i32 @llvm.vector.reduce.smax.v2i32(<2 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V4 = call i32 @llvm.vector.reduce.smax.v4i32(<4 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8 = call i32 @llvm.vector.reduce.smax.v8i32(<8 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V16 = call i32 @llvm.vector.reduce.smax.v16i32(<16 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32 = call i32 @llvm.vector.reduce.smax.v32i32(<32 x i32> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i32'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2 = call i32 @llvm.vector.reduce.smax.v2i32(<2 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4 = call i32 @llvm.vector.reduce.smax.v4i32(<4 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8 = call i32 @llvm.vector.reduce.smax.v8i32(<8 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V16 = call i32 @llvm.vector.reduce.smax.v16i32(<16 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %V32 = call i32 @llvm.vector.reduce.smax.v32i32(<32 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2 = call i32 @llvm.vector.reduce.smax.v2i32(<2 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V4 = call i32 @llvm.vector.reduce.smax.v4i32(<4 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8 = call i32 @llvm.vector.reduce.smax.v8i32(<8 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V16 = call i32 @llvm.vector.reduce.smax.v16i32(<16 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32 = call i32 @llvm.vector.reduce.smax.v32i32(<32 x i32> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE4-LABEL: 'reduce_i32'
@@ -183,7 +183,7 @@ define i32 @reduce_i16(i32 %arg) {
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8 = call i16 @llvm.vector.reduce.smax.v8i16(<8 x i16> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V16 = call i16 @llvm.vector.reduce.smax.v16i16(<16 x i16> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32 = call i16 @llvm.vector.reduce.smax.v32i16(<32 x i16> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V64 = call i16 @llvm.vector.reduce.smax.v64i16(<64 x i16> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V64 = call i16 @llvm.vector.reduce.smax.v64i16(<64 x i16> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512BW-LABEL: 'reduce_i16'
@@ -201,7 +201,7 @@ define i32 @reduce_i16(i32 %arg) {
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8 = call i16 @llvm.vector.reduce.smax.v8i16(<8 x i16> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V16 = call i16 @llvm.vector.reduce.smax.v16i16(<16 x i16> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32 = call i16 @llvm.vector.reduce.smax.v32i16(<32 x i16> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V64 = call i16 @llvm.vector.reduce.smax.v64i16(<64 x i16> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V64 = call i16 @llvm.vector.reduce.smax.v64i16(<64 x i16> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
   %V2  = call i16 @llvm.vector.reduce.smax.v2i16(<2 x i16> undef)
@@ -215,23 +215,23 @@ define i32 @reduce_i16(i32 %arg) {
 
 define i32 @reduce_i8(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2 = call i8 @llvm.vector.reduce.smax.v2i8(<2 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4 = call i8 @llvm.vector.reduce.smax.v4i8(<4 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8 = call i8 @llvm.vector.reduce.smax.v8i8(<8 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V16 = call i8 @llvm.vector.reduce.smax.v16i8(<16 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V32 = call i8 @llvm.vector.reduce.smax.v32i8(<32 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V64 = call i8 @llvm.vector.reduce.smax.v64i8(<64 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: %V128 = call i8 @llvm.vector.reduce.smax.v128i8(<128 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2 = call i8 @llvm.vector.reduce.smax.v2i8(<2 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V4 = call i8 @llvm.vector.reduce.smax.v4i8(<4 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8 = call i8 @llvm.vector.reduce.smax.v8i8(<8 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16 = call i8 @llvm.vector.reduce.smax.v16i8(<16 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V32 = call i8 @llvm.vector.reduce.smax.v32i8(<32 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V64 = call i8 @llvm.vector.reduce.smax.v64i8(<64 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 82 for instruction: %V128 = call i8 @llvm.vector.reduce.smax.v128i8(<128 x i8> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i8'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2 = call i8 @llvm.vector.reduce.smax.v2i8(<2 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4 = call i8 @llvm.vector.reduce.smax.v4i8(<4 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8 = call i8 @llvm.vector.reduce.smax.v8i8(<8 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V16 = call i8 @llvm.vector.reduce.smax.v16i8(<16 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V32 = call i8 @llvm.vector.reduce.smax.v32i8(<32 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V64 = call i8 @llvm.vector.reduce.smax.v64i8(<64 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: %V128 = call i8 @llvm.vector.reduce.smax.v128i8(<128 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2 = call i8 @llvm.vector.reduce.smax.v2i8(<2 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V4 = call i8 @llvm.vector.reduce.smax.v4i8(<4 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8 = call i8 @llvm.vector.reduce.smax.v8i8(<8 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16 = call i8 @llvm.vector.reduce.smax.v16i8(<16 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V32 = call i8 @llvm.vector.reduce.smax.v32i8(<32 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V64 = call i8 @llvm.vector.reduce.smax.v64i8(<64 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 82 for instruction: %V128 = call i8 @llvm.vector.reduce.smax.v128i8(<128 x i8> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE4-LABEL: 'reduce_i8'
@@ -271,7 +271,7 @@ define i32 @reduce_i8(i32 %arg) {
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V16 = call i8 @llvm.vector.reduce.smax.v16i8(<16 x i8> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32 = call i8 @llvm.vector.reduce.smax.v32i8(<32 x i8> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V64 = call i8 @llvm.vector.reduce.smax.v64i8(<64 x i8> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V128 = call i8 @llvm.vector.reduce.smax.v128i8(<128 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V128 = call i8 @llvm.vector.reduce.smax.v128i8(<128 x i8> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512BW-LABEL: 'reduce_i8'
@@ -291,7 +291,7 @@ define i32 @reduce_i8(i32 %arg) {
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V16 = call i8 @llvm.vector.reduce.smax.v16i8(<16 x i8> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32 = call i8 @llvm.vector.reduce.smax.v32i8(<32 x i8> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V64 = call i8 @llvm.vector.reduce.smax.v64i8(<64 x i8> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V128 = call i8 @llvm.vector.reduce.smax.v128i8(<128 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V128 = call i8 @llvm.vector.reduce.smax.v128i8(<128 x i8> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
   %V2   = call i8 @llvm.vector.reduce.smax.v2i8(<2 x i8> undef)

diff  --git a/llvm/test/Analysis/CostModel/X86/reduce-smin.ll b/llvm/test/Analysis/CostModel/X86/reduce-smin.ll
index 2a48aaf97c24c..d366ee5bdfac2 100644
--- a/llvm/test/Analysis/CostModel/X86/reduce-smin.ll
+++ b/llvm/test/Analysis/CostModel/X86/reduce-smin.ll
@@ -12,50 +12,50 @@
 define i32 @reduce_i64(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i64'
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.vector.reduce.smin.v1i64(<1 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V2 = call i64 @llvm.vector.reduce.smin.v2i64(<2 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V4 = call i64 @llvm.vector.reduce.smin.v4i64(<4 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V8 = call i64 @llvm.vector.reduce.smin.v8i64(<8 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 90 for instruction: %V16 = call i64 @llvm.vector.reduce.smin.v16i64(<16 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V2 = call i64 @llvm.vector.reduce.smin.v2i64(<2 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V4 = call i64 @llvm.vector.reduce.smin.v4i64(<4 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V8 = call i64 @llvm.vector.reduce.smin.v8i64(<8 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 114 for instruction: %V16 = call i64 @llvm.vector.reduce.smin.v16i64(<16 x i64> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i64'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.vector.reduce.smin.v1i64(<1 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V2 = call i64 @llvm.vector.reduce.smin.v2i64(<2 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V4 = call i64 @llvm.vector.reduce.smin.v4i64(<4 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V8 = call i64 @llvm.vector.reduce.smin.v8i64(<8 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 90 for instruction: %V16 = call i64 @llvm.vector.reduce.smin.v16i64(<16 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V2 = call i64 @llvm.vector.reduce.smin.v2i64(<2 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V4 = call i64 @llvm.vector.reduce.smin.v4i64(<4 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V8 = call i64 @llvm.vector.reduce.smin.v8i64(<8 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 114 for instruction: %V16 = call i64 @llvm.vector.reduce.smin.v16i64(<16 x i64> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE41-LABEL: 'reduce_i64'
 ; SSE41-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.vector.reduce.smin.v1i64(<1 x i64> undef)
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2 = call i64 @llvm.vector.reduce.smin.v2i64(<2 x i64> undef)
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4 = call i64 @llvm.vector.reduce.smin.v4i64(<4 x i64> undef)
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V8 = call i64 @llvm.vector.reduce.smin.v8i64(<8 x i64> undef)
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 74 for instruction: %V16 = call i64 @llvm.vector.reduce.smin.v16i64(<16 x i64> undef)
+; SSE41-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2 = call i64 @llvm.vector.reduce.smin.v2i64(<2 x i64> undef)
+; SSE41-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V4 = call i64 @llvm.vector.reduce.smin.v4i64(<4 x i64> undef)
+; SSE41-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V8 = call i64 @llvm.vector.reduce.smin.v8i64(<8 x i64> undef)
+; SSE41-NEXT:  Cost Model: Found an estimated cost of 98 for instruction: %V16 = call i64 @llvm.vector.reduce.smin.v16i64(<16 x i64> undef)
 ; SSE41-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'reduce_i64'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.vector.reduce.smin.v1i64(<1 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i64 @llvm.vector.reduce.smin.v2i64(<2 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4 = call i64 @llvm.vector.reduce.smin.v4i64(<4 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8 = call i64 @llvm.vector.reduce.smin.v8i64(<8 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V16 = call i64 @llvm.vector.reduce.smin.v16i64(<16 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2 = call i64 @llvm.vector.reduce.smin.v2i64(<2 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4 = call i64 @llvm.vector.reduce.smin.v4i64(<4 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V8 = call i64 @llvm.vector.reduce.smin.v8i64(<8 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V16 = call i64 @llvm.vector.reduce.smin.v16i64(<16 x i64> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX1-LABEL: 'reduce_i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.vector.reduce.smin.v1i64(<1 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i64 @llvm.vector.reduce.smin.v2i64(<2 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4 = call i64 @llvm.vector.reduce.smin.v4i64(<4 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8 = call i64 @llvm.vector.reduce.smin.v8i64(<8 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V16 = call i64 @llvm.vector.reduce.smin.v16i64(<16 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2 = call i64 @llvm.vector.reduce.smin.v2i64(<2 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4 = call i64 @llvm.vector.reduce.smin.v4i64(<4 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V8 = call i64 @llvm.vector.reduce.smin.v8i64(<8 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V16 = call i64 @llvm.vector.reduce.smin.v16i64(<16 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'reduce_i64'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.vector.reduce.smin.v1i64(<1 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i64 @llvm.vector.reduce.smin.v2i64(<2 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4 = call i64 @llvm.vector.reduce.smin.v4i64(<4 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8 = call i64 @llvm.vector.reduce.smin.v8i64(<8 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V16 = call i64 @llvm.vector.reduce.smin.v16i64(<16 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2 = call i64 @llvm.vector.reduce.smin.v2i64(<2 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4 = call i64 @llvm.vector.reduce.smin.v4i64(<4 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8 = call i64 @llvm.vector.reduce.smin.v8i64(<8 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V16 = call i64 @llvm.vector.reduce.smin.v16i64(<16 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512-LABEL: 'reduce_i64'
@@ -76,19 +76,19 @@ define i32 @reduce_i64(i32 %arg) {
 
 define i32 @reduce_i32(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2 = call i32 @llvm.vector.reduce.smin.v2i32(<2 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4 = call i32 @llvm.vector.reduce.smin.v4i32(<4 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8 = call i32 @llvm.vector.reduce.smin.v8i32(<8 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V16 = call i32 @llvm.vector.reduce.smin.v16i32(<16 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %V32 = call i32 @llvm.vector.reduce.smin.v32i32(<32 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2 = call i32 @llvm.vector.reduce.smin.v2i32(<2 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V4 = call i32 @llvm.vector.reduce.smin.v4i32(<4 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8 = call i32 @llvm.vector.reduce.smin.v8i32(<8 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V16 = call i32 @llvm.vector.reduce.smin.v16i32(<16 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32 = call i32 @llvm.vector.reduce.smin.v32i32(<32 x i32> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i32'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2 = call i32 @llvm.vector.reduce.smin.v2i32(<2 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4 = call i32 @llvm.vector.reduce.smin.v4i32(<4 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8 = call i32 @llvm.vector.reduce.smin.v8i32(<8 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V16 = call i32 @llvm.vector.reduce.smin.v16i32(<16 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %V32 = call i32 @llvm.vector.reduce.smin.v32i32(<32 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2 = call i32 @llvm.vector.reduce.smin.v2i32(<2 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V4 = call i32 @llvm.vector.reduce.smin.v4i32(<4 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8 = call i32 @llvm.vector.reduce.smin.v8i32(<8 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V16 = call i32 @llvm.vector.reduce.smin.v16i32(<16 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32 = call i32 @llvm.vector.reduce.smin.v32i32(<32 x i32> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE4-LABEL: 'reduce_i32'
@@ -183,7 +183,7 @@ define i32 @reduce_i16(i32 %arg) {
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8 = call i16 @llvm.vector.reduce.smin.v8i16(<8 x i16> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V16 = call i16 @llvm.vector.reduce.smin.v16i16(<16 x i16> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32 = call i16 @llvm.vector.reduce.smin.v32i16(<32 x i16> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V64 = call i16 @llvm.vector.reduce.smin.v64i16(<64 x i16> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V64 = call i16 @llvm.vector.reduce.smin.v64i16(<64 x i16> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512BW-LABEL: 'reduce_i16'
@@ -201,7 +201,7 @@ define i32 @reduce_i16(i32 %arg) {
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8 = call i16 @llvm.vector.reduce.smin.v8i16(<8 x i16> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V16 = call i16 @llvm.vector.reduce.smin.v16i16(<16 x i16> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32 = call i16 @llvm.vector.reduce.smin.v32i16(<32 x i16> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V64 = call i16 @llvm.vector.reduce.smin.v64i16(<64 x i16> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V64 = call i16 @llvm.vector.reduce.smin.v64i16(<64 x i16> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
   %V2  = call i16 @llvm.vector.reduce.smin.v2i16(<2 x i16> undef)
@@ -215,23 +215,23 @@ define i32 @reduce_i16(i32 %arg) {
 
 define i32 @reduce_i8(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2 = call i8 @llvm.vector.reduce.smin.v2i8(<2 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4 = call i8 @llvm.vector.reduce.smin.v4i8(<4 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8 = call i8 @llvm.vector.reduce.smin.v8i8(<8 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V16 = call i8 @llvm.vector.reduce.smin.v16i8(<16 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V32 = call i8 @llvm.vector.reduce.smin.v32i8(<32 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V64 = call i8 @llvm.vector.reduce.smin.v64i8(<64 x i8> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: %V128 = call i8 @llvm.vector.reduce.smin.v128i8(<128 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2 = call i8 @llvm.vector.reduce.smin.v2i8(<2 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V4 = call i8 @llvm.vector.reduce.smin.v4i8(<4 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8 = call i8 @llvm.vector.reduce.smin.v8i8(<8 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16 = call i8 @llvm.vector.reduce.smin.v16i8(<16 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V32 = call i8 @llvm.vector.reduce.smin.v32i8(<32 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V64 = call i8 @llvm.vector.reduce.smin.v64i8(<64 x i8> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 82 for instruction: %V128 = call i8 @llvm.vector.reduce.smin.v128i8(<128 x i8> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i8'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2 = call i8 @llvm.vector.reduce.smin.v2i8(<2 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4 = call i8 @llvm.vector.reduce.smin.v4i8(<4 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8 = call i8 @llvm.vector.reduce.smin.v8i8(<8 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V16 = call i8 @llvm.vector.reduce.smin.v16i8(<16 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V32 = call i8 @llvm.vector.reduce.smin.v32i8(<32 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V64 = call i8 @llvm.vector.reduce.smin.v64i8(<64 x i8> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: %V128 = call i8 @llvm.vector.reduce.smin.v128i8(<128 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2 = call i8 @llvm.vector.reduce.smin.v2i8(<2 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V4 = call i8 @llvm.vector.reduce.smin.v4i8(<4 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8 = call i8 @llvm.vector.reduce.smin.v8i8(<8 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16 = call i8 @llvm.vector.reduce.smin.v16i8(<16 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V32 = call i8 @llvm.vector.reduce.smin.v32i8(<32 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V64 = call i8 @llvm.vector.reduce.smin.v64i8(<64 x i8> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 82 for instruction: %V128 = call i8 @llvm.vector.reduce.smin.v128i8(<128 x i8> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE4-LABEL: 'reduce_i8'
@@ -271,7 +271,7 @@ define i32 @reduce_i8(i32 %arg) {
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V16 = call i8 @llvm.vector.reduce.smin.v16i8(<16 x i8> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32 = call i8 @llvm.vector.reduce.smin.v32i8(<32 x i8> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V64 = call i8 @llvm.vector.reduce.smin.v64i8(<64 x i8> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V128 = call i8 @llvm.vector.reduce.smin.v128i8(<128 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V128 = call i8 @llvm.vector.reduce.smin.v128i8(<128 x i8> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512BW-LABEL: 'reduce_i8'
@@ -291,7 +291,7 @@ define i32 @reduce_i8(i32 %arg) {
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V16 = call i8 @llvm.vector.reduce.smin.v16i8(<16 x i8> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32 = call i8 @llvm.vector.reduce.smin.v32i8(<32 x i8> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V64 = call i8 @llvm.vector.reduce.smin.v64i8(<64 x i8> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V128 = call i8 @llvm.vector.reduce.smin.v128i8(<128 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V128 = call i8 @llvm.vector.reduce.smin.v128i8(<128 x i8> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
   %V2   = call i8 @llvm.vector.reduce.smin.v2i8(<2 x i8> undef)

diff  --git a/llvm/test/Analysis/CostModel/X86/reduce-umax.ll b/llvm/test/Analysis/CostModel/X86/reduce-umax.ll
index 7be97c184cd86..ec09d227e9489 100644
--- a/llvm/test/Analysis/CostModel/X86/reduce-umax.ll
+++ b/llvm/test/Analysis/CostModel/X86/reduce-umax.ll
@@ -12,26 +12,26 @@
 define i32 @reduce_i64(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i64'
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.vector.reduce.umax.v1i64(<1 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V2 = call i64 @llvm.vector.reduce.umax.v2i64(<2 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V4 = call i64 @llvm.vector.reduce.umax.v4i64(<4 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V8 = call i64 @llvm.vector.reduce.umax.v8i64(<8 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 90 for instruction: %V16 = call i64 @llvm.vector.reduce.umax.v16i64(<16 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V2 = call i64 @llvm.vector.reduce.umax.v2i64(<2 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V4 = call i64 @llvm.vector.reduce.umax.v4i64(<4 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V8 = call i64 @llvm.vector.reduce.umax.v8i64(<8 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 114 for instruction: %V16 = call i64 @llvm.vector.reduce.umax.v16i64(<16 x i64> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i64'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.vector.reduce.umax.v1i64(<1 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V2 = call i64 @llvm.vector.reduce.umax.v2i64(<2 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V4 = call i64 @llvm.vector.reduce.umax.v4i64(<4 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V8 = call i64 @llvm.vector.reduce.umax.v8i64(<8 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 90 for instruction: %V16 = call i64 @llvm.vector.reduce.umax.v16i64(<16 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V2 = call i64 @llvm.vector.reduce.umax.v2i64(<2 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V4 = call i64 @llvm.vector.reduce.umax.v4i64(<4 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V8 = call i64 @llvm.vector.reduce.umax.v8i64(<8 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 114 for instruction: %V16 = call i64 @llvm.vector.reduce.umax.v16i64(<16 x i64> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE41-LABEL: 'reduce_i64'
 ; SSE41-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.vector.reduce.umax.v1i64(<1 x i64> undef)
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2 = call i64 @llvm.vector.reduce.umax.v2i64(<2 x i64> undef)
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4 = call i64 @llvm.vector.reduce.umax.v4i64(<4 x i64> undef)
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V8 = call i64 @llvm.vector.reduce.umax.v8i64(<8 x i64> undef)
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 74 for instruction: %V16 = call i64 @llvm.vector.reduce.umax.v16i64(<16 x i64> undef)
+; SSE41-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2 = call i64 @llvm.vector.reduce.umax.v2i64(<2 x i64> undef)
+; SSE41-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V4 = call i64 @llvm.vector.reduce.umax.v4i64(<4 x i64> undef)
+; SSE41-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V8 = call i64 @llvm.vector.reduce.umax.v8i64(<8 x i64> undef)
+; SSE41-NEXT:  Cost Model: Found an estimated cost of 98 for instruction: %V16 = call i64 @llvm.vector.reduce.umax.v16i64(<16 x i64> undef)
 ; SSE41-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'reduce_i64'
@@ -46,16 +46,16 @@ define i32 @reduce_i64(i32 %arg) {
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.vector.reduce.umax.v1i64(<1 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2 = call i64 @llvm.vector.reduce.umax.v2i64(<2 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4 = call i64 @llvm.vector.reduce.umax.v4i64(<4 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i64 @llvm.vector.reduce.umax.v8i64(<8 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i64 @llvm.vector.reduce.umax.v16i64(<16 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8 = call i64 @llvm.vector.reduce.umax.v8i64(<8 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16 = call i64 @llvm.vector.reduce.umax.v16i64(<16 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'reduce_i64'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.vector.reduce.umax.v1i64(<1 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2 = call i64 @llvm.vector.reduce.umax.v2i64(<2 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4 = call i64 @llvm.vector.reduce.umax.v4i64(<4 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V8 = call i64 @llvm.vector.reduce.umax.v8i64(<8 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V16 = call i64 @llvm.vector.reduce.umax.v16i64(<16 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i64 @llvm.vector.reduce.umax.v8i64(<8 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i64 @llvm.vector.reduce.umax.v16i64(<16 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512-LABEL: 'reduce_i64'
@@ -76,19 +76,19 @@ define i32 @reduce_i64(i32 %arg) {
 
 define i32 @reduce_i32(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2 = call i32 @llvm.vector.reduce.umax.v2i32(<2 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4 = call i32 @llvm.vector.reduce.umax.v4i32(<4 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8 = call i32 @llvm.vector.reduce.umax.v8i32(<8 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V16 = call i32 @llvm.vector.reduce.umax.v16i32(<16 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %V32 = call i32 @llvm.vector.reduce.umax.v32i32(<32 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2 = call i32 @llvm.vector.reduce.umax.v2i32(<2 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V4 = call i32 @llvm.vector.reduce.umax.v4i32(<4 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8 = call i32 @llvm.vector.reduce.umax.v8i32(<8 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V16 = call i32 @llvm.vector.reduce.umax.v16i32(<16 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32 = call i32 @llvm.vector.reduce.umax.v32i32(<32 x i32> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i32'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2 = call i32 @llvm.vector.reduce.umax.v2i32(<2 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4 = call i32 @llvm.vector.reduce.umax.v4i32(<4 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8 = call i32 @llvm.vector.reduce.umax.v8i32(<8 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V16 = call i32 @llvm.vector.reduce.umax.v16i32(<16 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %V32 = call i32 @llvm.vector.reduce.umax.v32i32(<32 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2 = call i32 @llvm.vector.reduce.umax.v2i32(<2 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V4 = call i32 @llvm.vector.reduce.umax.v4i32(<4 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8 = call i32 @llvm.vector.reduce.umax.v8i32(<8 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V16 = call i32 @llvm.vector.reduce.umax.v16i32(<16 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32 = call i32 @llvm.vector.reduce.umax.v32i32(<32 x i32> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE4-LABEL: 'reduce_i32'
@@ -136,18 +136,18 @@ define i32 @reduce_i16(i32 %arg) {
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2 = call i16 @llvm.vector.reduce.umax.v2i16(<2 x i16> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4 = call i16 @llvm.vector.reduce.umax.v4i16(<4 x i16> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8 = call i16 @llvm.vector.reduce.umax.v8i16(<8 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V16 = call i16 @llvm.vector.reduce.umax.v16i16(<16 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V32 = call i16 @llvm.vector.reduce.umax.v32i16(<32 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V64 = call i16 @llvm.vector.reduce.umax.v64i16(<64 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i16 @llvm.vector.reduce.umax.v16i16(<16 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V32 = call i16 @llvm.vector.reduce.umax.v32i16(<32 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V64 = call i16 @llvm.vector.reduce.umax.v64i16(<64 x i16> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i16'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2 = call i16 @llvm.vector.reduce.umax.v2i16(<2 x i16> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4 = call i16 @llvm.vector.reduce.umax.v4i16(<4 x i16> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8 = call i16 @llvm.vector.reduce.umax.v8i16(<8 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V16 = call i16 @llvm.vector.reduce.umax.v16i16(<16 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V32 = call i16 @llvm.vector.reduce.umax.v32i16(<32 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V64 = call i16 @llvm.vector.reduce.umax.v64i16(<64 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i16 @llvm.vector.reduce.umax.v16i16(<16 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V32 = call i16 @llvm.vector.reduce.umax.v32i16(<32 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V64 = call i16 @llvm.vector.reduce.umax.v64i16(<64 x i16> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE4-LABEL: 'reduce_i16'
@@ -183,7 +183,7 @@ define i32 @reduce_i16(i32 %arg) {
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8 = call i16 @llvm.vector.reduce.umax.v8i16(<8 x i16> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V16 = call i16 @llvm.vector.reduce.umax.v16i16(<16 x i16> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32 = call i16 @llvm.vector.reduce.umax.v32i16(<32 x i16> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V64 = call i16 @llvm.vector.reduce.umax.v64i16(<64 x i16> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V64 = call i16 @llvm.vector.reduce.umax.v64i16(<64 x i16> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512BW-LABEL: 'reduce_i16'
@@ -201,7 +201,7 @@ define i32 @reduce_i16(i32 %arg) {
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8 = call i16 @llvm.vector.reduce.umax.v8i16(<8 x i16> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V16 = call i16 @llvm.vector.reduce.umax.v16i16(<16 x i16> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32 = call i16 @llvm.vector.reduce.umax.v32i16(<32 x i16> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V64 = call i16 @llvm.vector.reduce.umax.v64i16(<64 x i16> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V64 = call i16 @llvm.vector.reduce.umax.v64i16(<64 x i16> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
   %V2  = call i16 @llvm.vector.reduce.umax.v2i16(<2 x i16> undef)
@@ -271,7 +271,7 @@ define i32 @reduce_i8(i32 %arg) {
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V16 = call i8 @llvm.vector.reduce.umax.v16i8(<16 x i8> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32 = call i8 @llvm.vector.reduce.umax.v32i8(<32 x i8> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V64 = call i8 @llvm.vector.reduce.umax.v64i8(<64 x i8> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V128 = call i8 @llvm.vector.reduce.umax.v128i8(<128 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V128 = call i8 @llvm.vector.reduce.umax.v128i8(<128 x i8> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512BW-LABEL: 'reduce_i8'
@@ -291,7 +291,7 @@ define i32 @reduce_i8(i32 %arg) {
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V16 = call i8 @llvm.vector.reduce.umax.v16i8(<16 x i8> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32 = call i8 @llvm.vector.reduce.umax.v32i8(<32 x i8> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V64 = call i8 @llvm.vector.reduce.umax.v64i8(<64 x i8> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V128 = call i8 @llvm.vector.reduce.umax.v128i8(<128 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V128 = call i8 @llvm.vector.reduce.umax.v128i8(<128 x i8> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
   %V2   = call i8 @llvm.vector.reduce.umax.v2i8(<2 x i8> undef)

diff  --git a/llvm/test/Analysis/CostModel/X86/reduce-umin.ll b/llvm/test/Analysis/CostModel/X86/reduce-umin.ll
index 4916a98ad962d..82ed5b74fbfcc 100644
--- a/llvm/test/Analysis/CostModel/X86/reduce-umin.ll
+++ b/llvm/test/Analysis/CostModel/X86/reduce-umin.ll
@@ -12,26 +12,26 @@
 define i32 @reduce_i64(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i64'
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.vector.reduce.umin.v1i64(<1 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V2 = call i64 @llvm.vector.reduce.umin.v2i64(<2 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V4 = call i64 @llvm.vector.reduce.umin.v4i64(<4 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V8 = call i64 @llvm.vector.reduce.umin.v8i64(<8 x i64> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 90 for instruction: %V16 = call i64 @llvm.vector.reduce.umin.v16i64(<16 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V2 = call i64 @llvm.vector.reduce.umin.v2i64(<2 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V4 = call i64 @llvm.vector.reduce.umin.v4i64(<4 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V8 = call i64 @llvm.vector.reduce.umin.v8i64(<8 x i64> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 114 for instruction: %V16 = call i64 @llvm.vector.reduce.umin.v16i64(<16 x i64> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i64'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.vector.reduce.umin.v1i64(<1 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V2 = call i64 @llvm.vector.reduce.umin.v2i64(<2 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V4 = call i64 @llvm.vector.reduce.umin.v4i64(<4 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V8 = call i64 @llvm.vector.reduce.umin.v8i64(<8 x i64> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 90 for instruction: %V16 = call i64 @llvm.vector.reduce.umin.v16i64(<16 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V2 = call i64 @llvm.vector.reduce.umin.v2i64(<2 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V4 = call i64 @llvm.vector.reduce.umin.v4i64(<4 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V8 = call i64 @llvm.vector.reduce.umin.v8i64(<8 x i64> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 114 for instruction: %V16 = call i64 @llvm.vector.reduce.umin.v16i64(<16 x i64> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE41-LABEL: 'reduce_i64'
 ; SSE41-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.vector.reduce.umin.v1i64(<1 x i64> undef)
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2 = call i64 @llvm.vector.reduce.umin.v2i64(<2 x i64> undef)
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4 = call i64 @llvm.vector.reduce.umin.v4i64(<4 x i64> undef)
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V8 = call i64 @llvm.vector.reduce.umin.v8i64(<8 x i64> undef)
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 74 for instruction: %V16 = call i64 @llvm.vector.reduce.umin.v16i64(<16 x i64> undef)
+; SSE41-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2 = call i64 @llvm.vector.reduce.umin.v2i64(<2 x i64> undef)
+; SSE41-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V4 = call i64 @llvm.vector.reduce.umin.v4i64(<4 x i64> undef)
+; SSE41-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V8 = call i64 @llvm.vector.reduce.umin.v8i64(<8 x i64> undef)
+; SSE41-NEXT:  Cost Model: Found an estimated cost of 98 for instruction: %V16 = call i64 @llvm.vector.reduce.umin.v16i64(<16 x i64> undef)
 ; SSE41-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'reduce_i64'
@@ -46,16 +46,16 @@ define i32 @reduce_i64(i32 %arg) {
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.vector.reduce.umin.v1i64(<1 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2 = call i64 @llvm.vector.reduce.umin.v2i64(<2 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4 = call i64 @llvm.vector.reduce.umin.v4i64(<4 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i64 @llvm.vector.reduce.umin.v8i64(<8 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i64 @llvm.vector.reduce.umin.v16i64(<16 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8 = call i64 @llvm.vector.reduce.umin.v8i64(<8 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16 = call i64 @llvm.vector.reduce.umin.v16i64(<16 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'reduce_i64'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.vector.reduce.umin.v1i64(<1 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2 = call i64 @llvm.vector.reduce.umin.v2i64(<2 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4 = call i64 @llvm.vector.reduce.umin.v4i64(<4 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V8 = call i64 @llvm.vector.reduce.umin.v8i64(<8 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V16 = call i64 @llvm.vector.reduce.umin.v16i64(<16 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8 = call i64 @llvm.vector.reduce.umin.v8i64(<8 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16 = call i64 @llvm.vector.reduce.umin.v16i64(<16 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512-LABEL: 'reduce_i64'
@@ -76,19 +76,19 @@ define i32 @reduce_i64(i32 %arg) {
 
 define i32 @reduce_i32(i32 %arg) {
 ; SSE2-LABEL: 'reduce_i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2 = call i32 @llvm.vector.reduce.umin.v2i32(<2 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4 = call i32 @llvm.vector.reduce.umin.v4i32(<4 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8 = call i32 @llvm.vector.reduce.umin.v8i32(<8 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V16 = call i32 @llvm.vector.reduce.umin.v16i32(<16 x i32> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %V32 = call i32 @llvm.vector.reduce.umin.v32i32(<32 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2 = call i32 @llvm.vector.reduce.umin.v2i32(<2 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V4 = call i32 @llvm.vector.reduce.umin.v4i32(<4 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8 = call i32 @llvm.vector.reduce.umin.v8i32(<8 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V16 = call i32 @llvm.vector.reduce.umin.v16i32(<16 x i32> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32 = call i32 @llvm.vector.reduce.umin.v32i32(<32 x i32> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i32'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2 = call i32 @llvm.vector.reduce.umin.v2i32(<2 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V4 = call i32 @llvm.vector.reduce.umin.v4i32(<4 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8 = call i32 @llvm.vector.reduce.umin.v8i32(<8 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V16 = call i32 @llvm.vector.reduce.umin.v16i32(<16 x i32> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %V32 = call i32 @llvm.vector.reduce.umin.v32i32(<32 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2 = call i32 @llvm.vector.reduce.umin.v2i32(<2 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V4 = call i32 @llvm.vector.reduce.umin.v4i32(<4 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8 = call i32 @llvm.vector.reduce.umin.v8i32(<8 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V16 = call i32 @llvm.vector.reduce.umin.v16i32(<16 x i32> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %V32 = call i32 @llvm.vector.reduce.umin.v32i32(<32 x i32> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE4-LABEL: 'reduce_i32'
@@ -136,18 +136,18 @@ define i32 @reduce_i16(i32 %arg) {
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2 = call i16 @llvm.vector.reduce.umin.v2i16(<2 x i16> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4 = call i16 @llvm.vector.reduce.umin.v4i16(<4 x i16> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8 = call i16 @llvm.vector.reduce.umin.v8i16(<8 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V16 = call i16 @llvm.vector.reduce.umin.v16i16(<16 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V32 = call i16 @llvm.vector.reduce.umin.v32i16(<32 x i16> undef)
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V64 = call i16 @llvm.vector.reduce.umin.v64i16(<64 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i16 @llvm.vector.reduce.umin.v16i16(<16 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V32 = call i16 @llvm.vector.reduce.umin.v32i16(<32 x i16> undef)
+; SSE2-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V64 = call i16 @llvm.vector.reduce.umin.v64i16(<64 x i16> undef)
 ; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSSE3-LABEL: 'reduce_i16'
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2 = call i16 @llvm.vector.reduce.umin.v2i16(<2 x i16> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4 = call i16 @llvm.vector.reduce.umin.v4i16(<4 x i16> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8 = call i16 @llvm.vector.reduce.umin.v8i16(<8 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V16 = call i16 @llvm.vector.reduce.umin.v16i16(<16 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V32 = call i16 @llvm.vector.reduce.umin.v32i16(<32 x i16> undef)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %V64 = call i16 @llvm.vector.reduce.umin.v64i16(<64 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16 = call i16 @llvm.vector.reduce.umin.v16i16(<16 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V32 = call i16 @llvm.vector.reduce.umin.v32i16(<32 x i16> undef)
+; SSSE3-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V64 = call i16 @llvm.vector.reduce.umin.v64i16(<64 x i16> undef)
 ; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE4-LABEL: 'reduce_i16'
@@ -183,7 +183,7 @@ define i32 @reduce_i16(i32 %arg) {
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8 = call i16 @llvm.vector.reduce.umin.v8i16(<8 x i16> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V16 = call i16 @llvm.vector.reduce.umin.v16i16(<16 x i16> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32 = call i16 @llvm.vector.reduce.umin.v32i16(<32 x i16> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V64 = call i16 @llvm.vector.reduce.umin.v64i16(<64 x i16> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V64 = call i16 @llvm.vector.reduce.umin.v64i16(<64 x i16> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512BW-LABEL: 'reduce_i16'
@@ -201,7 +201,7 @@ define i32 @reduce_i16(i32 %arg) {
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8 = call i16 @llvm.vector.reduce.umin.v8i16(<8 x i16> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V16 = call i16 @llvm.vector.reduce.umin.v16i16(<16 x i16> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V32 = call i16 @llvm.vector.reduce.umin.v32i16(<32 x i16> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V64 = call i16 @llvm.vector.reduce.umin.v64i16(<64 x i16> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V64 = call i16 @llvm.vector.reduce.umin.v64i16(<64 x i16> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
   %V2  = call i16 @llvm.vector.reduce.umin.v2i16(<2 x i16> undef)
@@ -271,7 +271,7 @@ define i32 @reduce_i8(i32 %arg) {
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V16 = call i8 @llvm.vector.reduce.umin.v16i8(<16 x i8> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32 = call i8 @llvm.vector.reduce.umin.v32i8(<32 x i8> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V64 = call i8 @llvm.vector.reduce.umin.v64i8(<64 x i8> undef)
-; AVX512F-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V128 = call i8 @llvm.vector.reduce.umin.v128i8(<128 x i8> undef)
+; AVX512F-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V128 = call i8 @llvm.vector.reduce.umin.v128i8(<128 x i8> undef)
 ; AVX512F-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512BW-LABEL: 'reduce_i8'
@@ -291,7 +291,7 @@ define i32 @reduce_i8(i32 %arg) {
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V16 = call i8 @llvm.vector.reduce.umin.v16i8(<16 x i8> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32 = call i8 @llvm.vector.reduce.umin.v32i8(<32 x i8> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V64 = call i8 @llvm.vector.reduce.umin.v64i8(<64 x i8> undef)
-; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V128 = call i8 @llvm.vector.reduce.umin.v128i8(<128 x i8> undef)
+; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V128 = call i8 @llvm.vector.reduce.umin.v128i8(<128 x i8> undef)
 ; AVX512DQ-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
   %V2   = call i8 @llvm.vector.reduce.umin.v2i8(<2 x i8> undef)

diff  --git a/llvm/test/Transforms/SLPVectorizer/X86/horizontal-minmax.ll b/llvm/test/Transforms/SLPVectorizer/X86/horizontal-minmax.ll
index 4f0a8256255d6..eb317de222f2a 100644
--- a/llvm/test/Transforms/SLPVectorizer/X86/horizontal-minmax.ll
+++ b/llvm/test/Transforms/SLPVectorizer/X86/horizontal-minmax.ll
@@ -17,10 +17,45 @@ declare i32 @llvm.smin.i32(i32, i32)
 declare i32 @llvm.umin.i32(i32, i32)
 
 define i32 @maxi8(i32) {
-; CHECK-LABEL: @maxi8(
-; CHECK-NEXT:    [[TMP2:%.*]] = load <8 x i32>, <8 x i32>* bitcast ([32 x i32]* @arr to <8 x i32>*), align 16
-; CHECK-NEXT:    [[TMP3:%.*]] = call i32 @llvm.vector.reduce.smax.v8i32(<8 x i32> [[TMP2]])
-; CHECK-NEXT:    ret i32 [[TMP3]]
+; SSE-LABEL: @maxi8(
+; SSE-NEXT:    [[TMP2:%.*]] = load i32, i32* getelementptr inbounds ([32 x i32], [32 x i32]* @arr, i64 0, i64 0), align 16
+; SSE-NEXT:    [[TMP3:%.*]] = load i32, i32* getelementptr inbounds ([32 x i32], [32 x i32]* @arr, i64 0, i64 1), align 4
+; SSE-NEXT:    [[TMP4:%.*]] = icmp sgt i32 [[TMP2]], [[TMP3]]
+; SSE-NEXT:    [[TMP5:%.*]] = select i1 [[TMP4]], i32 [[TMP2]], i32 [[TMP3]]
+; SSE-NEXT:    [[TMP6:%.*]] = load i32, i32* getelementptr inbounds ([32 x i32], [32 x i32]* @arr, i64 0, i64 2), align 8
+; SSE-NEXT:    [[TMP7:%.*]] = icmp sgt i32 [[TMP5]], [[TMP6]]
+; SSE-NEXT:    [[TMP8:%.*]] = select i1 [[TMP7]], i32 [[TMP5]], i32 [[TMP6]]
+; SSE-NEXT:    [[TMP9:%.*]] = load i32, i32* getelementptr inbounds ([32 x i32], [32 x i32]* @arr, i64 0, i64 3), align 4
+; SSE-NEXT:    [[TMP10:%.*]] = icmp sgt i32 [[TMP8]], [[TMP9]]
+; SSE-NEXT:    [[TMP11:%.*]] = select i1 [[TMP10]], i32 [[TMP8]], i32 [[TMP9]]
+; SSE-NEXT:    [[TMP12:%.*]] = load i32, i32* getelementptr inbounds ([32 x i32], [32 x i32]* @arr, i64 0, i64 4), align 16
+; SSE-NEXT:    [[TMP13:%.*]] = icmp sgt i32 [[TMP11]], [[TMP12]]
+; SSE-NEXT:    [[TMP14:%.*]] = select i1 [[TMP13]], i32 [[TMP11]], i32 [[TMP12]]
+; SSE-NEXT:    [[TMP15:%.*]] = load i32, i32* getelementptr inbounds ([32 x i32], [32 x i32]* @arr, i64 0, i64 5), align 4
+; SSE-NEXT:    [[TMP16:%.*]] = icmp sgt i32 [[TMP14]], [[TMP15]]
+; SSE-NEXT:    [[TMP17:%.*]] = select i1 [[TMP16]], i32 [[TMP14]], i32 [[TMP15]]
+; SSE-NEXT:    [[TMP18:%.*]] = load i32, i32* getelementptr inbounds ([32 x i32], [32 x i32]* @arr, i64 0, i64 6), align 8
+; SSE-NEXT:    [[TMP19:%.*]] = icmp sgt i32 [[TMP17]], [[TMP18]]
+; SSE-NEXT:    [[TMP20:%.*]] = select i1 [[TMP19]], i32 [[TMP17]], i32 [[TMP18]]
+; SSE-NEXT:    [[TMP21:%.*]] = load i32, i32* getelementptr inbounds ([32 x i32], [32 x i32]* @arr, i64 0, i64 7), align 4
+; SSE-NEXT:    [[TMP22:%.*]] = icmp sgt i32 [[TMP20]], [[TMP21]]
+; SSE-NEXT:    [[TMP23:%.*]] = select i1 [[TMP22]], i32 [[TMP20]], i32 [[TMP21]]
+; SSE-NEXT:    ret i32 [[TMP23]]
+;
+; AVX-LABEL: @maxi8(
+; AVX-NEXT:    [[TMP2:%.*]] = load <8 x i32>, <8 x i32>* bitcast ([32 x i32]* @arr to <8 x i32>*), align 16
+; AVX-NEXT:    [[TMP3:%.*]] = call i32 @llvm.vector.reduce.smax.v8i32(<8 x i32> [[TMP2]])
+; AVX-NEXT:    ret i32 [[TMP3]]
+;
+; AVX2-LABEL: @maxi8(
+; AVX2-NEXT:    [[TMP2:%.*]] = load <8 x i32>, <8 x i32>* bitcast ([32 x i32]* @arr to <8 x i32>*), align 16
+; AVX2-NEXT:    [[TMP3:%.*]] = call i32 @llvm.vector.reduce.smax.v8i32(<8 x i32> [[TMP2]])
+; AVX2-NEXT:    ret i32 [[TMP3]]
+;
+; THRESH-LABEL: @maxi8(
+; THRESH-NEXT:    [[TMP2:%.*]] = load <8 x i32>, <8 x i32>* bitcast ([32 x i32]* @arr to <8 x i32>*), align 16
+; THRESH-NEXT:    [[TMP3:%.*]] = call i32 @llvm.vector.reduce.smax.v8i32(<8 x i32> [[TMP2]])
+; THRESH-NEXT:    ret i32 [[TMP3]]
 ;
   %2 = load i32, i32* getelementptr inbounds ([32 x i32], [32 x i32]* @arr, i64 0, i64 0), align 16
   %3 = load i32, i32* getelementptr inbounds ([32 x i32], [32 x i32]* @arr, i64 0, i64 1), align 4
@@ -48,11 +83,49 @@ define i32 @maxi8(i32) {
 }
 
 define i32 @maxi8_store_in(i32) {
-; CHECK-LABEL: @maxi8_store_in(
-; CHECK-NEXT:    [[TMP2:%.*]] = load <8 x i32>, <8 x i32>* bitcast ([32 x i32]* @arr to <8 x i32>*), align 16
-; CHECK-NEXT:    store i32 0, i32* @var, align 8
-; CHECK-NEXT:    [[TMP3:%.*]] = call i32 @llvm.vector.reduce.smax.v8i32(<8 x i32> [[TMP2]])
-; CHECK-NEXT:    ret i32 [[TMP3]]
+; SSE-LABEL: @maxi8_store_in(
+; SSE-NEXT:    [[TMP2:%.*]] = load i32, i32* getelementptr inbounds ([32 x i32], [32 x i32]* @arr, i64 0, i64 0), align 16
+; SSE-NEXT:    [[TMP3:%.*]] = load i32, i32* getelementptr inbounds ([32 x i32], [32 x i32]* @arr, i64 0, i64 1), align 4
+; SSE-NEXT:    [[TMP4:%.*]] = icmp sgt i32 [[TMP2]], [[TMP3]]
+; SSE-NEXT:    [[TMP5:%.*]] = select i1 [[TMP4]], i32 [[TMP2]], i32 [[TMP3]]
+; SSE-NEXT:    [[TMP6:%.*]] = load i32, i32* getelementptr inbounds ([32 x i32], [32 x i32]* @arr, i64 0, i64 2), align 8
+; SSE-NEXT:    [[TMP7:%.*]] = icmp sgt i32 [[TMP5]], [[TMP6]]
+; SSE-NEXT:    [[TMP8:%.*]] = select i1 [[TMP7]], i32 [[TMP5]], i32 [[TMP6]]
+; SSE-NEXT:    [[TMP9:%.*]] = load i32, i32* getelementptr inbounds ([32 x i32], [32 x i32]* @arr, i64 0, i64 3), align 4
+; SSE-NEXT:    [[TMP10:%.*]] = icmp sgt i32 [[TMP8]], [[TMP9]]
+; SSE-NEXT:    [[TMP11:%.*]] = select i1 [[TMP10]], i32 [[TMP8]], i32 [[TMP9]]
+; SSE-NEXT:    [[TMP12:%.*]] = load i32, i32* getelementptr inbounds ([32 x i32], [32 x i32]* @arr, i64 0, i64 4), align 16
+; SSE-NEXT:    [[TMP13:%.*]] = icmp sgt i32 [[TMP11]], [[TMP12]]
+; SSE-NEXT:    [[TMP14:%.*]] = select i1 [[TMP13]], i32 [[TMP11]], i32 [[TMP12]]
+; SSE-NEXT:    store i32 0, i32* @var, align 8
+; SSE-NEXT:    [[TMP15:%.*]] = load i32, i32* getelementptr inbounds ([32 x i32], [32 x i32]* @arr, i64 0, i64 5), align 4
+; SSE-NEXT:    [[TMP16:%.*]] = icmp sgt i32 [[TMP14]], [[TMP15]]
+; SSE-NEXT:    [[TMP17:%.*]] = select i1 [[TMP16]], i32 [[TMP14]], i32 [[TMP15]]
+; SSE-NEXT:    [[TMP18:%.*]] = load i32, i32* getelementptr inbounds ([32 x i32], [32 x i32]* @arr, i64 0, i64 6), align 8
+; SSE-NEXT:    [[TMP19:%.*]] = icmp sgt i32 [[TMP17]], [[TMP18]]
+; SSE-NEXT:    [[TMP20:%.*]] = select i1 [[TMP19]], i32 [[TMP17]], i32 [[TMP18]]
+; SSE-NEXT:    [[TMP21:%.*]] = load i32, i32* getelementptr inbounds ([32 x i32], [32 x i32]* @arr, i64 0, i64 7), align 4
+; SSE-NEXT:    [[TMP22:%.*]] = icmp sgt i32 [[TMP20]], [[TMP21]]
+; SSE-NEXT:    [[TMP23:%.*]] = select i1 [[TMP22]], i32 [[TMP20]], i32 [[TMP21]]
+; SSE-NEXT:    ret i32 [[TMP23]]
+;
+; AVX-LABEL: @maxi8_store_in(
+; AVX-NEXT:    [[TMP2:%.*]] = load <8 x i32>, <8 x i32>* bitcast ([32 x i32]* @arr to <8 x i32>*), align 16
+; AVX-NEXT:    store i32 0, i32* @var, align 8
+; AVX-NEXT:    [[TMP3:%.*]] = call i32 @llvm.vector.reduce.smax.v8i32(<8 x i32> [[TMP2]])
+; AVX-NEXT:    ret i32 [[TMP3]]
+;
+; AVX2-LABEL: @maxi8_store_in(
+; AVX2-NEXT:    [[TMP2:%.*]] = load <8 x i32>, <8 x i32>* bitcast ([32 x i32]* @arr to <8 x i32>*), align 16
+; AVX2-NEXT:    store i32 0, i32* @var, align 8
+; AVX2-NEXT:    [[TMP3:%.*]] = call i32 @llvm.vector.reduce.smax.v8i32(<8 x i32> [[TMP2]])
+; AVX2-NEXT:    ret i32 [[TMP3]]
+;
+; THRESH-LABEL: @maxi8_store_in(
+; THRESH-NEXT:    [[TMP2:%.*]] = load <8 x i32>, <8 x i32>* bitcast ([32 x i32]* @arr to <8 x i32>*), align 16
+; THRESH-NEXT:    store i32 0, i32* @var, align 8
+; THRESH-NEXT:    [[TMP3:%.*]] = call i32 @llvm.vector.reduce.smax.v8i32(<8 x i32> [[TMP2]])
+; THRESH-NEXT:    ret i32 [[TMP3]]
 ;
   %2 = load i32, i32* getelementptr inbounds ([32 x i32], [32 x i32]* @arr, i64 0, i64 0), align 16
   %3 = load i32, i32* getelementptr inbounds ([32 x i32], [32 x i32]* @arr, i64 0, i64 1), align 4
@@ -1146,18 +1219,69 @@ define i32* @maxp8(i32) {
 }
 
 define i32 @smax_intrinsic_rdx_v8i32(i32* %p0) {
-; CHECK-LABEL: @smax_intrinsic_rdx_v8i32(
-; CHECK-NEXT:    [[P1:%.*]] = getelementptr inbounds i32, i32* [[P0:%.*]], i64 1
-; CHECK-NEXT:    [[P2:%.*]] = getelementptr inbounds i32, i32* [[P0]], i64 2
-; CHECK-NEXT:    [[P3:%.*]] = getelementptr inbounds i32, i32* [[P0]], i64 3
-; CHECK-NEXT:    [[P4:%.*]] = getelementptr inbounds i32, i32* [[P0]], i64 4
-; CHECK-NEXT:    [[P5:%.*]] = getelementptr inbounds i32, i32* [[P0]], i64 5
-; CHECK-NEXT:    [[P6:%.*]] = getelementptr inbounds i32, i32* [[P0]], i64 6
-; CHECK-NEXT:    [[P7:%.*]] = getelementptr inbounds i32, i32* [[P0]], i64 7
-; CHECK-NEXT:    [[TMP1:%.*]] = bitcast i32* [[P0]] to <8 x i32>*
-; CHECK-NEXT:    [[TMP2:%.*]] = load <8 x i32>, <8 x i32>* [[TMP1]], align 4
-; CHECK-NEXT:    [[TMP3:%.*]] = call i32 @llvm.vector.reduce.smax.v8i32(<8 x i32> [[TMP2]])
-; CHECK-NEXT:    ret i32 [[TMP3]]
+; SSE-LABEL: @smax_intrinsic_rdx_v8i32(
+; SSE-NEXT:    [[P1:%.*]] = getelementptr inbounds i32, i32* [[P0:%.*]], i64 1
+; SSE-NEXT:    [[P2:%.*]] = getelementptr inbounds i32, i32* [[P0]], i64 2
+; SSE-NEXT:    [[P3:%.*]] = getelementptr inbounds i32, i32* [[P0]], i64 3
+; SSE-NEXT:    [[P4:%.*]] = getelementptr inbounds i32, i32* [[P0]], i64 4
+; SSE-NEXT:    [[P5:%.*]] = getelementptr inbounds i32, i32* [[P0]], i64 5
+; SSE-NEXT:    [[P6:%.*]] = getelementptr inbounds i32, i32* [[P0]], i64 6
+; SSE-NEXT:    [[P7:%.*]] = getelementptr inbounds i32, i32* [[P0]], i64 7
+; SSE-NEXT:    [[T0:%.*]] = load i32, i32* [[P0]], align 4
+; SSE-NEXT:    [[T1:%.*]] = load i32, i32* [[P1]], align 4
+; SSE-NEXT:    [[T2:%.*]] = load i32, i32* [[P2]], align 4
+; SSE-NEXT:    [[T3:%.*]] = load i32, i32* [[P3]], align 4
+; SSE-NEXT:    [[T4:%.*]] = load i32, i32* [[P4]], align 4
+; SSE-NEXT:    [[T5:%.*]] = load i32, i32* [[P5]], align 4
+; SSE-NEXT:    [[T6:%.*]] = load i32, i32* [[P6]], align 4
+; SSE-NEXT:    [[T7:%.*]] = load i32, i32* [[P7]], align 4
+; SSE-NEXT:    [[M10:%.*]] = tail call i32 @llvm.smax.i32(i32 [[T1]], i32 [[T0]])
+; SSE-NEXT:    [[M32:%.*]] = tail call i32 @llvm.smax.i32(i32 [[T3]], i32 [[T2]])
+; SSE-NEXT:    [[M54:%.*]] = tail call i32 @llvm.smax.i32(i32 [[T5]], i32 [[T4]])
+; SSE-NEXT:    [[M76:%.*]] = tail call i32 @llvm.smax.i32(i32 [[T7]], i32 [[T6]])
+; SSE-NEXT:    [[M3210:%.*]] = tail call i32 @llvm.smax.i32(i32 [[M32]], i32 [[M10]])
+; SSE-NEXT:    [[M7654:%.*]] = tail call i32 @llvm.smax.i32(i32 [[M76]], i32 [[M54]])
+; SSE-NEXT:    [[M:%.*]] = tail call i32 @llvm.smax.i32(i32 [[M7654]], i32 [[M3210]])
+; SSE-NEXT:    ret i32 [[M]]
+;
+; AVX-LABEL: @smax_intrinsic_rdx_v8i32(
+; AVX-NEXT:    [[P1:%.*]] = getelementptr inbounds i32, i32* [[P0:%.*]], i64 1
+; AVX-NEXT:    [[P2:%.*]] = getelementptr inbounds i32, i32* [[P0]], i64 2
+; AVX-NEXT:    [[P3:%.*]] = getelementptr inbounds i32, i32* [[P0]], i64 3
+; AVX-NEXT:    [[P4:%.*]] = getelementptr inbounds i32, i32* [[P0]], i64 4
+; AVX-NEXT:    [[P5:%.*]] = getelementptr inbounds i32, i32* [[P0]], i64 5
+; AVX-NEXT:    [[P6:%.*]] = getelementptr inbounds i32, i32* [[P0]], i64 6
+; AVX-NEXT:    [[P7:%.*]] = getelementptr inbounds i32, i32* [[P0]], i64 7
+; AVX-NEXT:    [[TMP1:%.*]] = bitcast i32* [[P0]] to <8 x i32>*
+; AVX-NEXT:    [[TMP2:%.*]] = load <8 x i32>, <8 x i32>* [[TMP1]], align 4
+; AVX-NEXT:    [[TMP3:%.*]] = call i32 @llvm.vector.reduce.smax.v8i32(<8 x i32> [[TMP2]])
+; AVX-NEXT:    ret i32 [[TMP3]]
+;
+; AVX2-LABEL: @smax_intrinsic_rdx_v8i32(
+; AVX2-NEXT:    [[P1:%.*]] = getelementptr inbounds i32, i32* [[P0:%.*]], i64 1
+; AVX2-NEXT:    [[P2:%.*]] = getelementptr inbounds i32, i32* [[P0]], i64 2
+; AVX2-NEXT:    [[P3:%.*]] = getelementptr inbounds i32, i32* [[P0]], i64 3
+; AVX2-NEXT:    [[P4:%.*]] = getelementptr inbounds i32, i32* [[P0]], i64 4
+; AVX2-NEXT:    [[P5:%.*]] = getelementptr inbounds i32, i32* [[P0]], i64 5
+; AVX2-NEXT:    [[P6:%.*]] = getelementptr inbounds i32, i32* [[P0]], i64 6
+; AVX2-NEXT:    [[P7:%.*]] = getelementptr inbounds i32, i32* [[P0]], i64 7
+; AVX2-NEXT:    [[TMP1:%.*]] = bitcast i32* [[P0]] to <8 x i32>*
+; AVX2-NEXT:    [[TMP2:%.*]] = load <8 x i32>, <8 x i32>* [[TMP1]], align 4
+; AVX2-NEXT:    [[TMP3:%.*]] = call i32 @llvm.vector.reduce.smax.v8i32(<8 x i32> [[TMP2]])
+; AVX2-NEXT:    ret i32 [[TMP3]]
+;
+; THRESH-LABEL: @smax_intrinsic_rdx_v8i32(
+; THRESH-NEXT:    [[P1:%.*]] = getelementptr inbounds i32, i32* [[P0:%.*]], i64 1
+; THRESH-NEXT:    [[P2:%.*]] = getelementptr inbounds i32, i32* [[P0]], i64 2
+; THRESH-NEXT:    [[P3:%.*]] = getelementptr inbounds i32, i32* [[P0]], i64 3
+; THRESH-NEXT:    [[P4:%.*]] = getelementptr inbounds i32, i32* [[P0]], i64 4
+; THRESH-NEXT:    [[P5:%.*]] = getelementptr inbounds i32, i32* [[P0]], i64 5
+; THRESH-NEXT:    [[P6:%.*]] = getelementptr inbounds i32, i32* [[P0]], i64 6
+; THRESH-NEXT:    [[P7:%.*]] = getelementptr inbounds i32, i32* [[P0]], i64 7
+; THRESH-NEXT:    [[TMP1:%.*]] = bitcast i32* [[P0]] to <8 x i32>*
+; THRESH-NEXT:    [[TMP2:%.*]] = load <8 x i32>, <8 x i32>* [[TMP1]], align 4
+; THRESH-NEXT:    [[TMP3:%.*]] = call i32 @llvm.vector.reduce.smax.v8i32(<8 x i32> [[TMP2]])
+; THRESH-NEXT:    ret i32 [[TMP3]]
 ;
   %p1 = getelementptr inbounds i32, i32* %p0, i64 1
   %p2 = getelementptr inbounds i32, i32* %p0, i64 2
@@ -1333,15 +1457,53 @@ define i8 @umin_intrinsic_rdx_v16i8(i8* %p0) {
 ; This should not crash.
 
 define void @PR49730() {
-; CHECK-LABEL: @PR49730(
-; CHECK-NEXT:    [[TMP1:%.*]] = call <4 x i32> @llvm.smin.v4i32(<4 x i32> poison, <4 x i32> <i32 2, i32 2, i32 1, i32 1>)
-; CHECK-NEXT:    [[TMP2:%.*]] = sub nsw <4 x i32> poison, [[TMP1]]
-; CHECK-NEXT:    [[T12:%.*]] = sub nsw i32 undef, undef
-; CHECK-NEXT:    [[TMP3:%.*]] = call i32 @llvm.vector.reduce.umin.v4i32(<4 x i32> [[TMP2]])
-; CHECK-NEXT:    [[TMP4:%.*]] = call i32 @llvm.umin.i32(i32 [[TMP3]], i32 [[T12]])
-; CHECK-NEXT:    [[TMP5:%.*]] = call i32 @llvm.umin.i32(i32 [[TMP4]], i32 undef)
-; CHECK-NEXT:    [[T14:%.*]] = call i32 @llvm.umin.i32(i32 [[TMP5]], i32 93)
-; CHECK-NEXT:    ret void
+; SSE-LABEL: @PR49730(
+; SSE-NEXT:    [[T:%.*]] = call i32 @llvm.smin.i32(i32 undef, i32 2)
+; SSE-NEXT:    [[T1:%.*]] = sub nsw i32 undef, [[T]]
+; SSE-NEXT:    [[T2:%.*]] = call i32 @llvm.umin.i32(i32 undef, i32 [[T1]])
+; SSE-NEXT:    [[T3:%.*]] = call i32 @llvm.smin.i32(i32 undef, i32 2)
+; SSE-NEXT:    [[T4:%.*]] = sub nsw i32 undef, [[T3]]
+; SSE-NEXT:    [[T5:%.*]] = call i32 @llvm.umin.i32(i32 [[T2]], i32 [[T4]])
+; SSE-NEXT:    [[T6:%.*]] = call i32 @llvm.smin.i32(i32 undef, i32 1)
+; SSE-NEXT:    [[T7:%.*]] = sub nuw nsw i32 undef, [[T6]]
+; SSE-NEXT:    [[T8:%.*]] = call i32 @llvm.umin.i32(i32 [[T5]], i32 [[T7]])
+; SSE-NEXT:    [[T9:%.*]] = call i32 @llvm.smin.i32(i32 undef, i32 1)
+; SSE-NEXT:    [[T10:%.*]] = sub nsw i32 undef, [[T9]]
+; SSE-NEXT:    [[T11:%.*]] = call i32 @llvm.umin.i32(i32 [[T8]], i32 [[T10]])
+; SSE-NEXT:    [[T12:%.*]] = sub nsw i32 undef, undef
+; SSE-NEXT:    [[T13:%.*]] = call i32 @llvm.umin.i32(i32 [[T11]], i32 [[T12]])
+; SSE-NEXT:    [[T14:%.*]] = call i32 @llvm.umin.i32(i32 [[T13]], i32 93)
+; SSE-NEXT:    ret void
+;
+; AVX-LABEL: @PR49730(
+; AVX-NEXT:    [[TMP1:%.*]] = call <4 x i32> @llvm.smin.v4i32(<4 x i32> poison, <4 x i32> <i32 2, i32 2, i32 1, i32 1>)
+; AVX-NEXT:    [[TMP2:%.*]] = sub nsw <4 x i32> poison, [[TMP1]]
+; AVX-NEXT:    [[T12:%.*]] = sub nsw i32 undef, undef
+; AVX-NEXT:    [[TMP3:%.*]] = call i32 @llvm.vector.reduce.umin.v4i32(<4 x i32> [[TMP2]])
+; AVX-NEXT:    [[TMP4:%.*]] = call i32 @llvm.umin.i32(i32 [[TMP3]], i32 [[T12]])
+; AVX-NEXT:    [[TMP5:%.*]] = call i32 @llvm.umin.i32(i32 [[TMP4]], i32 undef)
+; AVX-NEXT:    [[T14:%.*]] = call i32 @llvm.umin.i32(i32 [[TMP5]], i32 93)
+; AVX-NEXT:    ret void
+;
+; AVX2-LABEL: @PR49730(
+; AVX2-NEXT:    [[TMP1:%.*]] = call <4 x i32> @llvm.smin.v4i32(<4 x i32> poison, <4 x i32> <i32 2, i32 2, i32 1, i32 1>)
+; AVX2-NEXT:    [[TMP2:%.*]] = sub nsw <4 x i32> poison, [[TMP1]]
+; AVX2-NEXT:    [[T12:%.*]] = sub nsw i32 undef, undef
+; AVX2-NEXT:    [[TMP3:%.*]] = call i32 @llvm.vector.reduce.umin.v4i32(<4 x i32> [[TMP2]])
+; AVX2-NEXT:    [[TMP4:%.*]] = call i32 @llvm.umin.i32(i32 [[TMP3]], i32 [[T12]])
+; AVX2-NEXT:    [[TMP5:%.*]] = call i32 @llvm.umin.i32(i32 [[TMP4]], i32 undef)
+; AVX2-NEXT:    [[T14:%.*]] = call i32 @llvm.umin.i32(i32 [[TMP5]], i32 93)
+; AVX2-NEXT:    ret void
+;
+; THRESH-LABEL: @PR49730(
+; THRESH-NEXT:    [[TMP1:%.*]] = call <4 x i32> @llvm.smin.v4i32(<4 x i32> poison, <4 x i32> <i32 2, i32 2, i32 1, i32 1>)
+; THRESH-NEXT:    [[TMP2:%.*]] = sub nsw <4 x i32> poison, [[TMP1]]
+; THRESH-NEXT:    [[T12:%.*]] = sub nsw i32 undef, undef
+; THRESH-NEXT:    [[TMP3:%.*]] = call i32 @llvm.vector.reduce.umin.v4i32(<4 x i32> [[TMP2]])
+; THRESH-NEXT:    [[TMP4:%.*]] = call i32 @llvm.umin.i32(i32 [[TMP3]], i32 [[T12]])
+; THRESH-NEXT:    [[TMP5:%.*]] = call i32 @llvm.umin.i32(i32 [[TMP4]], i32 undef)
+; THRESH-NEXT:    [[T14:%.*]] = call i32 @llvm.umin.i32(i32 [[TMP5]], i32 93)
+; THRESH-NEXT:    ret void
 ;
   %t = call i32 @llvm.smin.i32(i32 undef, i32 2)
   %t1 = sub nsw i32 undef, %t

diff  --git a/llvm/test/Transforms/SLPVectorizer/X86/horizontal-smax.ll b/llvm/test/Transforms/SLPVectorizer/X86/horizontal-smax.ll
index b96e262504e2d..7ec349b7ace79 100644
--- a/llvm/test/Transforms/SLPVectorizer/X86/horizontal-smax.ll
+++ b/llvm/test/Transforms/SLPVectorizer/X86/horizontal-smax.ll
@@ -47,10 +47,28 @@ define i32 @smax_v4i32(i32) {
 }
 
 define i32 @smax_v8i32(i32) {
-; CHECK-LABEL: @smax_v8i32(
-; CHECK-NEXT:    [[TMP2:%.*]] = load <8 x i32>, <8 x i32>* bitcast ([32 x i32]* @arr to <8 x i32>*), align 16
-; CHECK-NEXT:    [[TMP3:%.*]] = call i32 @llvm.vector.reduce.smax.v8i32(<8 x i32> [[TMP2]])
-; CHECK-NEXT:    ret i32 [[TMP3]]
+; SSE-LABEL: @smax_v8i32(
+; SSE-NEXT:    [[TMP2:%.*]] = load i32, i32* getelementptr inbounds ([32 x i32], [32 x i32]* @arr, i64 0, i64 0), align 16
+; SSE-NEXT:    [[TMP3:%.*]] = load i32, i32* getelementptr inbounds ([32 x i32], [32 x i32]* @arr, i64 0, i64 1), align 4
+; SSE-NEXT:    [[TMP4:%.*]] = load i32, i32* getelementptr inbounds ([32 x i32], [32 x i32]* @arr, i64 0, i64 2), align 8
+; SSE-NEXT:    [[TMP5:%.*]] = load i32, i32* getelementptr inbounds ([32 x i32], [32 x i32]* @arr, i64 0, i64 3), align 4
+; SSE-NEXT:    [[TMP6:%.*]] = load i32, i32* getelementptr inbounds ([32 x i32], [32 x i32]* @arr, i64 0, i64 4), align 16
+; SSE-NEXT:    [[TMP7:%.*]] = load i32, i32* getelementptr inbounds ([32 x i32], [32 x i32]* @arr, i64 0, i64 5), align 4
+; SSE-NEXT:    [[TMP8:%.*]] = load i32, i32* getelementptr inbounds ([32 x i32], [32 x i32]* @arr, i64 0, i64 6), align 8
+; SSE-NEXT:    [[TMP9:%.*]] = load i32, i32* getelementptr inbounds ([32 x i32], [32 x i32]* @arr, i64 0, i64 7), align 4
+; SSE-NEXT:    [[TMP10:%.*]] = call i32 @llvm.smax.i32(i32 [[TMP2]], i32 [[TMP3]])
+; SSE-NEXT:    [[TMP11:%.*]] = call i32 @llvm.smax.i32(i32 [[TMP10]], i32 [[TMP4]])
+; SSE-NEXT:    [[TMP12:%.*]] = call i32 @llvm.smax.i32(i32 [[TMP11]], i32 [[TMP5]])
+; SSE-NEXT:    [[TMP13:%.*]] = call i32 @llvm.smax.i32(i32 [[TMP12]], i32 [[TMP6]])
+; SSE-NEXT:    [[TMP14:%.*]] = call i32 @llvm.smax.i32(i32 [[TMP13]], i32 [[TMP7]])
+; SSE-NEXT:    [[TMP15:%.*]] = call i32 @llvm.smax.i32(i32 [[TMP14]], i32 [[TMP8]])
+; SSE-NEXT:    [[TMP16:%.*]] = call i32 @llvm.smax.i32(i32 [[TMP15]], i32 [[TMP9]])
+; SSE-NEXT:    ret i32 [[TMP16]]
+;
+; AVX-LABEL: @smax_v8i32(
+; AVX-NEXT:    [[TMP2:%.*]] = load <8 x i32>, <8 x i32>* bitcast ([32 x i32]* @arr to <8 x i32>*), align 16
+; AVX-NEXT:    [[TMP3:%.*]] = call i32 @llvm.vector.reduce.smax.v8i32(<8 x i32> [[TMP2]])
+; AVX-NEXT:    ret i32 [[TMP3]]
 ;
   %2 = load i32, i32* getelementptr inbounds ([32 x i32], [32 x i32]* @arr, i64 0, i64 0), align 16
   %3 = load i32, i32* getelementptr inbounds ([32 x i32], [32 x i32]* @arr, i64 0, i64 1), align 4

diff  --git a/llvm/test/Transforms/VectorCombine/X86/extract-cmp-binop.ll b/llvm/test/Transforms/VectorCombine/X86/extract-cmp-binop.ll
index 73e52c13a465f..bb2734284c86a 100644
--- a/llvm/test/Transforms/VectorCombine/X86/extract-cmp-binop.ll
+++ b/llvm/test/Transforms/VectorCombine/X86/extract-cmp-binop.ll
@@ -1,22 +1,15 @@
 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
-; RUN: opt < %s -vector-combine -S -mtriple=x86_64-- -mattr=sse2 | FileCheck %s --check-prefixes=CHECK,SSE
-; RUN: opt < %s -vector-combine -S -mtriple=x86_64-- -mattr=avx2 | FileCheck %s --check-prefixes=CHECK,AVX
+; RUN: opt < %s -vector-combine -S -mtriple=x86_64-- -mattr=sse2 | FileCheck %s --check-prefixes=CHECK
+; RUN: opt < %s -vector-combine -S -mtriple=x86_64-- -mattr=avx2 | FileCheck %s --check-prefixes=CHECK
 
 define i1 @fcmp_and_v2f64(<2 x double> %a) {
-; SSE-LABEL: @fcmp_and_v2f64(
-; SSE-NEXT:    [[E1:%.*]] = extractelement <2 x double> [[A:%.*]], i32 0
-; SSE-NEXT:    [[E2:%.*]] = extractelement <2 x double> [[A]], i32 1
-; SSE-NEXT:    [[CMP1:%.*]] = fcmp olt double [[E1]], 4.200000e+01
-; SSE-NEXT:    [[CMP2:%.*]] = fcmp olt double [[E2]], -8.000000e+00
-; SSE-NEXT:    [[R:%.*]] = and i1 [[CMP1]], [[CMP2]]
-; SSE-NEXT:    ret i1 [[R]]
-;
-; AVX-LABEL: @fcmp_and_v2f64(
-; AVX-NEXT:    [[TMP1:%.*]] = fcmp olt <2 x double> [[A:%.*]], <double 4.200000e+01, double -8.000000e+00>
-; AVX-NEXT:    [[SHIFT:%.*]] = shufflevector <2 x i1> [[TMP1]], <2 x i1> poison, <2 x i32> <i32 1, i32 undef>
-; AVX-NEXT:    [[TMP2:%.*]] = and <2 x i1> [[TMP1]], [[SHIFT]]
-; AVX-NEXT:    [[R:%.*]] = extractelement <2 x i1> [[TMP2]], i64 0
-; AVX-NEXT:    ret i1 [[R]]
+; CHECK-LABEL: @fcmp_and_v2f64(
+; CHECK-NEXT:    [[E1:%.*]] = extractelement <2 x double> [[A:%.*]], i32 0
+; CHECK-NEXT:    [[E2:%.*]] = extractelement <2 x double> [[A]], i32 1
+; CHECK-NEXT:    [[CMP1:%.*]] = fcmp olt double [[E1]], 4.200000e+01
+; CHECK-NEXT:    [[CMP2:%.*]] = fcmp olt double [[E2]], -8.000000e+00
+; CHECK-NEXT:    [[R:%.*]] = and i1 [[CMP1]], [[CMP2]]
+; CHECK-NEXT:    ret i1 [[R]]
 ;
   %e1 = extractelement <2 x double> %a, i32 0
   %e2 = extractelement <2 x double> %a, i32 1
@@ -27,20 +20,13 @@ define i1 @fcmp_and_v2f64(<2 x double> %a) {
 }
 
 define i1 @fcmp_or_v4f64(<4 x double> %a) {
-; SSE-LABEL: @fcmp_or_v4f64(
-; SSE-NEXT:    [[E1:%.*]] = extractelement <4 x double> [[A:%.*]], i32 0
-; SSE-NEXT:    [[E2:%.*]] = extractelement <4 x double> [[A]], i64 2
-; SSE-NEXT:    [[CMP1:%.*]] = fcmp olt double [[E1]], 4.200000e+01
-; SSE-NEXT:    [[CMP2:%.*]] = fcmp olt double [[E2]], -8.000000e+00
-; SSE-NEXT:    [[R:%.*]] = or i1 [[CMP1]], [[CMP2]]
-; SSE-NEXT:    ret i1 [[R]]
-;
-; AVX-LABEL: @fcmp_or_v4f64(
-; AVX-NEXT:    [[TMP1:%.*]] = fcmp olt <4 x double> [[A:%.*]], <double 4.200000e+01, double undef, double -8.000000e+00, double undef>
-; AVX-NEXT:    [[SHIFT:%.*]] = shufflevector <4 x i1> [[TMP1]], <4 x i1> poison, <4 x i32> <i32 2, i32 undef, i32 undef, i32 undef>
-; AVX-NEXT:    [[TMP2:%.*]] = or <4 x i1> [[TMP1]], [[SHIFT]]
-; AVX-NEXT:    [[R:%.*]] = extractelement <4 x i1> [[TMP2]], i64 0
-; AVX-NEXT:    ret i1 [[R]]
+; CHECK-LABEL: @fcmp_or_v4f64(
+; CHECK-NEXT:    [[E1:%.*]] = extractelement <4 x double> [[A:%.*]], i32 0
+; CHECK-NEXT:    [[E2:%.*]] = extractelement <4 x double> [[A]], i64 2
+; CHECK-NEXT:    [[CMP1:%.*]] = fcmp olt double [[E1]], 4.200000e+01
+; CHECK-NEXT:    [[CMP2:%.*]] = fcmp olt double [[E2]], -8.000000e+00
+; CHECK-NEXT:    [[R:%.*]] = or i1 [[CMP1]], [[CMP2]]
+; CHECK-NEXT:    ret i1 [[R]]
 ;
   %e1 = extractelement <4 x double> %a, i32 0
   %e2 = extractelement <4 x double> %a, i64 2
@@ -52,10 +38,11 @@ define i1 @fcmp_or_v4f64(<4 x double> %a) {
 
 define i1 @icmp_xor_v4i32(<4 x i32> %a) {
 ; CHECK-LABEL: @icmp_xor_v4i32(
-; CHECK-NEXT:    [[TMP1:%.*]] = icmp sgt <4 x i32> [[A:%.*]], <i32 undef, i32 -8, i32 undef, i32 42>
-; CHECK-NEXT:    [[SHIFT:%.*]] = shufflevector <4 x i1> [[TMP1]], <4 x i1> poison, <4 x i32> <i32 undef, i32 3, i32 undef, i32 undef>
-; CHECK-NEXT:    [[TMP2:%.*]] = xor <4 x i1> [[TMP1]], [[SHIFT]]
-; CHECK-NEXT:    [[R:%.*]] = extractelement <4 x i1> [[TMP2]], i64 1
+; CHECK-NEXT:    [[E1:%.*]] = extractelement <4 x i32> [[A:%.*]], i32 3
+; CHECK-NEXT:    [[E2:%.*]] = extractelement <4 x i32> [[A]], i32 1
+; CHECK-NEXT:    [[CMP1:%.*]] = icmp sgt i32 [[E1]], 42
+; CHECK-NEXT:    [[CMP2:%.*]] = icmp sgt i32 [[E2]], -8
+; CHECK-NEXT:    [[R:%.*]] = xor i1 [[CMP1]], [[CMP2]]
 ; CHECK-NEXT:    ret i1 [[R]]
 ;
   %e1 = extractelement <4 x i32> %a, i32 3
@@ -69,20 +56,13 @@ define i1 @icmp_xor_v4i32(<4 x i32> %a) {
 ; add is not canonical (should be xor), but that is ok.
 
 define i1 @icmp_add_v8i32(<8 x i32> %a) {
-; SSE-LABEL: @icmp_add_v8i32(
-; SSE-NEXT:    [[E1:%.*]] = extractelement <8 x i32> [[A:%.*]], i32 7
-; SSE-NEXT:    [[E2:%.*]] = extractelement <8 x i32> [[A]], i32 2
-; SSE-NEXT:    [[CMP1:%.*]] = icmp eq i32 [[E1]], 42
-; SSE-NEXT:    [[CMP2:%.*]] = icmp eq i32 [[E2]], -8
-; SSE-NEXT:    [[R:%.*]] = add i1 [[CMP1]], [[CMP2]]
-; SSE-NEXT:    ret i1 [[R]]
-;
-; AVX-LABEL: @icmp_add_v8i32(
-; AVX-NEXT:    [[TMP1:%.*]] = icmp eq <8 x i32> [[A:%.*]], <i32 undef, i32 undef, i32 -8, i32 undef, i32 undef, i32 undef, i32 undef, i32 42>
-; AVX-NEXT:    [[SHIFT:%.*]] = shufflevector <8 x i1> [[TMP1]], <8 x i1> poison, <8 x i32> <i32 undef, i32 undef, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
-; AVX-NEXT:    [[TMP2:%.*]] = add <8 x i1> [[TMP1]], [[SHIFT]]
-; AVX-NEXT:    [[R:%.*]] = extractelement <8 x i1> [[TMP2]], i64 2
-; AVX-NEXT:    ret i1 [[R]]
+; CHECK-LABEL: @icmp_add_v8i32(
+; CHECK-NEXT:    [[E1:%.*]] = extractelement <8 x i32> [[A:%.*]], i32 7
+; CHECK-NEXT:    [[E2:%.*]] = extractelement <8 x i32> [[A]], i32 2
+; CHECK-NEXT:    [[CMP1:%.*]] = icmp eq i32 [[E1]], 42
+; CHECK-NEXT:    [[CMP2:%.*]] = icmp eq i32 [[E2]], -8
+; CHECK-NEXT:    [[R:%.*]] = add i1 [[CMP1]], [[CMP2]]
+; CHECK-NEXT:    ret i1 [[R]]
 ;
   %e1 = extractelement <8 x i32> %a, i32 7
   %e2 = extractelement <8 x i32> %a, i32 2

diff  --git a/llvm/test/Transforms/VectorCombine/X86/extract-cmp.ll b/llvm/test/Transforms/VectorCombine/X86/extract-cmp.ll
index ade08844e1f87..5bc70aeabfbcf 100644
--- a/llvm/test/Transforms/VectorCombine/X86/extract-cmp.ll
+++ b/llvm/test/Transforms/VectorCombine/X86/extract-cmp.ll
@@ -1,26 +1,30 @@
 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
-; RUN: opt < %s -vector-combine -S -mtriple=x86_64-- -mattr=SSE2 | FileCheck %s --check-prefixes=CHECK,SSE
-; RUN: opt < %s -vector-combine -S -mtriple=x86_64-- -mattr=AVX2 | FileCheck %s --check-prefixes=CHECK,AVX
+; RUN: opt < %s -vector-combine -S -mtriple=x86_64-- -mattr=SSE2 | FileCheck %s --check-prefixes=CHECK
+; RUN: opt < %s -vector-combine -S -mtriple=x86_64-- -mattr=AVX2 | FileCheck %s --check-prefixes=CHECK
 
 define i1 @cmp_v4i32(<4 x float> %arg, <4 x float> %arg1) {
 ; CHECK-LABEL: @cmp_v4i32(
 ; CHECK-NEXT:  bb:
 ; CHECK-NEXT:    [[T:%.*]] = bitcast <4 x float> [[ARG:%.*]] to <4 x i32>
+; CHECK-NEXT:    [[T2:%.*]] = extractelement <4 x i32> [[T]], i32 0
 ; CHECK-NEXT:    [[T3:%.*]] = bitcast <4 x float> [[ARG1:%.*]] to <4 x i32>
-; CHECK-NEXT:    [[TMP0:%.*]] = icmp eq <4 x i32> [[T]], [[T3]]
-; CHECK-NEXT:    [[T5:%.*]] = extractelement <4 x i1> [[TMP0]], i32 0
+; CHECK-NEXT:    [[T4:%.*]] = extractelement <4 x i32> [[T3]], i32 0
+; CHECK-NEXT:    [[T5:%.*]] = icmp eq i32 [[T2]], [[T4]]
 ; CHECK-NEXT:    br i1 [[T5]], label [[BB6:%.*]], label [[BB18:%.*]]
 ; CHECK:       bb6:
-; CHECK-NEXT:    [[TMP1:%.*]] = icmp eq <4 x i32> [[T]], [[T3]]
-; CHECK-NEXT:    [[T9:%.*]] = extractelement <4 x i1> [[TMP1]], i32 1
+; CHECK-NEXT:    [[T7:%.*]] = extractelement <4 x i32> [[T]], i32 1
+; CHECK-NEXT:    [[T8:%.*]] = extractelement <4 x i32> [[T3]], i32 1
+; CHECK-NEXT:    [[T9:%.*]] = icmp eq i32 [[T7]], [[T8]]
 ; CHECK-NEXT:    br i1 [[T9]], label [[BB10:%.*]], label [[BB18]]
 ; CHECK:       bb10:
-; CHECK-NEXT:    [[TMP2:%.*]] = icmp eq <4 x i32> [[T]], [[T3]]
-; CHECK-NEXT:    [[T13:%.*]] = extractelement <4 x i1> [[TMP2]], i32 2
+; CHECK-NEXT:    [[T11:%.*]] = extractelement <4 x i32> [[T]], i32 2
+; CHECK-NEXT:    [[T12:%.*]] = extractelement <4 x i32> [[T3]], i32 2
+; CHECK-NEXT:    [[T13:%.*]] = icmp eq i32 [[T11]], [[T12]]
 ; CHECK-NEXT:    br i1 [[T13]], label [[BB14:%.*]], label [[BB18]]
 ; CHECK:       bb14:
-; CHECK-NEXT:    [[TMP3:%.*]] = icmp eq <4 x i32> [[T]], [[T3]]
-; CHECK-NEXT:    [[T17:%.*]] = extractelement <4 x i1> [[TMP3]], i32 3
+; CHECK-NEXT:    [[T15:%.*]] = extractelement <4 x i32> [[T]], i32 3
+; CHECK-NEXT:    [[T16:%.*]] = extractelement <4 x i32> [[T3]], i32 3
+; CHECK-NEXT:    [[T17:%.*]] = icmp eq i32 [[T15]], [[T16]]
 ; CHECK-NEXT:    br label [[BB18]]
 ; CHECK:       bb18:
 ; CHECK-NEXT:    [[T19:%.*]] = phi i1 [ false, [[BB10]] ], [ false, [[BB6]] ], [ false, [[BB:%.*]] ], [ [[T17]], [[BB14]] ]
@@ -58,32 +62,19 @@ bb18:
 }
 
 define i32 @cmp_v2f64(<2 x double> %x, <2 x double> %y, <2 x double> %z) {
-; SSE-LABEL: @cmp_v2f64(
-; SSE-NEXT:  entry:
-; SSE-NEXT:    [[X1:%.*]] = extractelement <2 x double> [[X:%.*]], i32 1
-; SSE-NEXT:    [[Y1:%.*]] = extractelement <2 x double> [[Y:%.*]], i32 1
-; SSE-NEXT:    [[CMP1:%.*]] = fcmp oeq double [[X1]], [[Y1]]
-; SSE-NEXT:    br i1 [[CMP1]], label [[T:%.*]], label [[F:%.*]]
-; SSE:       t:
-; SSE-NEXT:    [[Z1:%.*]] = extractelement <2 x double> [[Z:%.*]], i32 1
-; SSE-NEXT:    [[CMP2:%.*]] = fcmp ogt double [[Y1]], [[Z1]]
-; SSE-NEXT:    [[E:%.*]] = select i1 [[CMP2]], i32 42, i32 99
-; SSE-NEXT:    ret i32 [[E]]
-; SSE:       f:
-; SSE-NEXT:    ret i32 0
-;
-; AVX-LABEL: @cmp_v2f64(
-; AVX-NEXT:  entry:
-; AVX-NEXT:    [[TMP0:%.*]] = fcmp oeq <2 x double> [[X:%.*]], [[Y:%.*]]
-; AVX-NEXT:    [[CMP1:%.*]] = extractelement <2 x i1> [[TMP0]], i32 1
-; AVX-NEXT:    br i1 [[CMP1]], label [[T:%.*]], label [[F:%.*]]
-; AVX:       t:
-; AVX-NEXT:    [[TMP1:%.*]] = fcmp ogt <2 x double> [[Y]], [[Z:%.*]]
-; AVX-NEXT:    [[CMP2:%.*]] = extractelement <2 x i1> [[TMP1]], i32 1
-; AVX-NEXT:    [[E:%.*]] = select i1 [[CMP2]], i32 42, i32 99
-; AVX-NEXT:    ret i32 [[E]]
-; AVX:       f:
-; AVX-NEXT:    ret i32 0
+; CHECK-LABEL: @cmp_v2f64(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[X1:%.*]] = extractelement <2 x double> [[X:%.*]], i32 1
+; CHECK-NEXT:    [[Y1:%.*]] = extractelement <2 x double> [[Y:%.*]], i32 1
+; CHECK-NEXT:    [[CMP1:%.*]] = fcmp oeq double [[X1]], [[Y1]]
+; CHECK-NEXT:    br i1 [[CMP1]], label [[T:%.*]], label [[F:%.*]]
+; CHECK:       t:
+; CHECK-NEXT:    [[Z1:%.*]] = extractelement <2 x double> [[Z:%.*]], i32 1
+; CHECK-NEXT:    [[CMP2:%.*]] = fcmp ogt double [[Y1]], [[Z1]]
+; CHECK-NEXT:    [[E:%.*]] = select i1 [[CMP2]], i32 42, i32 99
+; CHECK-NEXT:    ret i32 [[E]]
+; CHECK:       f:
+; CHECK-NEXT:    ret i32 0
 ;
 entry:
   %x1 = extractelement <2 x double> %x, i32 1
@@ -102,17 +93,11 @@ f:
 }
 
 define i1 @cmp01_v2f64(<2 x double> %x, <2 x double> %y) {
-; SSE-LABEL: @cmp01_v2f64(
-; SSE-NEXT:    [[X0:%.*]] = extractelement <2 x double> [[X:%.*]], i32 0
-; SSE-NEXT:    [[Y1:%.*]] = extractelement <2 x double> [[Y:%.*]], i32 1
-; SSE-NEXT:    [[CMP:%.*]] = fcmp oge double [[X0]], [[Y1]]
-; SSE-NEXT:    ret i1 [[CMP]]
-;
-; AVX-LABEL: @cmp01_v2f64(
-; AVX-NEXT:    [[SHIFT:%.*]] = shufflevector <2 x double> [[Y:%.*]], <2 x double> poison, <2 x i32> <i32 1, i32 undef>
-; AVX-NEXT:    [[TMP1:%.*]] = fcmp oge <2 x double> [[X:%.*]], [[SHIFT]]
-; AVX-NEXT:    [[CMP:%.*]] = extractelement <2 x i1> [[TMP1]], i32 0
-; AVX-NEXT:    ret i1 [[CMP]]
+; CHECK-LABEL: @cmp01_v2f64(
+; CHECK-NEXT:    [[X0:%.*]] = extractelement <2 x double> [[X:%.*]], i32 0
+; CHECK-NEXT:    [[Y1:%.*]] = extractelement <2 x double> [[Y:%.*]], i32 1
+; CHECK-NEXT:    [[CMP:%.*]] = fcmp oge double [[X0]], [[Y1]]
+; CHECK-NEXT:    ret i1 [[CMP]]
 ;
   %x0 = extractelement <2 x double> %x, i32 0
   %y1 = extractelement <2 x double> %y, i32 1
@@ -121,17 +106,11 @@ define i1 @cmp01_v2f64(<2 x double> %x, <2 x double> %y) {
 }
 
 define i1 @cmp10_v2f64(<2 x double> %x, <2 x double> %y) {
-; SSE-LABEL: @cmp10_v2f64(
-; SSE-NEXT:    [[X1:%.*]] = extractelement <2 x double> [[X:%.*]], i32 1
-; SSE-NEXT:    [[Y0:%.*]] = extractelement <2 x double> [[Y:%.*]], i32 0
-; SSE-NEXT:    [[CMP:%.*]] = fcmp ule double [[X1]], [[Y0]]
-; SSE-NEXT:    ret i1 [[CMP]]
-;
-; AVX-LABEL: @cmp10_v2f64(
-; AVX-NEXT:    [[SHIFT:%.*]] = shufflevector <2 x double> [[X:%.*]], <2 x double> poison, <2 x i32> <i32 1, i32 undef>
-; AVX-NEXT:    [[TMP1:%.*]] = fcmp ule <2 x double> [[SHIFT]], [[Y:%.*]]
-; AVX-NEXT:    [[CMP:%.*]] = extractelement <2 x i1> [[TMP1]], i64 0
-; AVX-NEXT:    ret i1 [[CMP]]
+; CHECK-LABEL: @cmp10_v2f64(
+; CHECK-NEXT:    [[X1:%.*]] = extractelement <2 x double> [[X:%.*]], i32 1
+; CHECK-NEXT:    [[Y0:%.*]] = extractelement <2 x double> [[Y:%.*]], i32 0
+; CHECK-NEXT:    [[CMP:%.*]] = fcmp ule double [[X1]], [[Y0]]
+; CHECK-NEXT:    ret i1 [[CMP]]
 ;
   %x1 = extractelement <2 x double> %x, i32 1
   %y0 = extractelement <2 x double> %y, i32 0
@@ -141,9 +120,9 @@ define i1 @cmp10_v2f64(<2 x double> %x, <2 x double> %y) {
 
 define i1 @cmp12_v4i32(<4 x i32> %x, <4 x i32> %y) {
 ; CHECK-LABEL: @cmp12_v4i32(
-; CHECK-NEXT:    [[SHIFT:%.*]] = shufflevector <4 x i32> [[Y:%.*]], <4 x i32> poison, <4 x i32> <i32 undef, i32 2, i32 undef, i32 undef>
-; CHECK-NEXT:    [[TMP1:%.*]] = icmp sgt <4 x i32> [[X:%.*]], [[SHIFT]]
-; CHECK-NEXT:    [[CMP:%.*]] = extractelement <4 x i1> [[TMP1]], i32 1
+; CHECK-NEXT:    [[X1:%.*]] = extractelement <4 x i32> [[X:%.*]], i32 1
+; CHECK-NEXT:    [[Y2:%.*]] = extractelement <4 x i32> [[Y:%.*]], i32 2
+; CHECK-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[X1]], [[Y2]]
 ; CHECK-NEXT:    ret i1 [[CMP]]
 ;
   %x1 = extractelement <4 x i32> %x, i32 1
@@ -153,19 +132,12 @@ define i1 @cmp12_v4i32(<4 x i32> %x, <4 x i32> %y) {
 }
 
 define <4 x i1> @ins_fcmp_ext_ext(<4 x float> %a, <4 x i1> %b) {
-; SSE-LABEL: @ins_fcmp_ext_ext(
-; SSE-NEXT:    [[A1:%.*]] = extractelement <4 x float> [[A:%.*]], i32 1
-; SSE-NEXT:    [[A2:%.*]] = extractelement <4 x float> [[A]], i32 2
-; SSE-NEXT:    [[A21:%.*]] = fcmp ugt float [[A2]], [[A1]]
-; SSE-NEXT:    [[R:%.*]] = insertelement <4 x i1> [[B:%.*]], i1 [[A21]], i32 2
-; SSE-NEXT:    ret <4 x i1> [[R]]
-;
-; AVX-LABEL: @ins_fcmp_ext_ext(
-; AVX-NEXT:    [[SHIFT:%.*]] = shufflevector <4 x float> [[A:%.*]], <4 x float> poison, <4 x i32> <i32 undef, i32 undef, i32 1, i32 undef>
-; AVX-NEXT:    [[TMP1:%.*]] = fcmp ugt <4 x float> [[A]], [[SHIFT]]
-; AVX-NEXT:    [[A21:%.*]] = extractelement <4 x i1> [[TMP1]], i32 2
-; AVX-NEXT:    [[R:%.*]] = insertelement <4 x i1> [[B:%.*]], i1 [[A21]], i32 2
-; AVX-NEXT:    ret <4 x i1> [[R]]
+; CHECK-LABEL: @ins_fcmp_ext_ext(
+; CHECK-NEXT:    [[A1:%.*]] = extractelement <4 x float> [[A:%.*]], i32 1
+; CHECK-NEXT:    [[A2:%.*]] = extractelement <4 x float> [[A]], i32 2
+; CHECK-NEXT:    [[A21:%.*]] = fcmp ugt float [[A2]], [[A1]]
+; CHECK-NEXT:    [[R:%.*]] = insertelement <4 x i1> [[B:%.*]], i1 [[A21]], i32 2
+; CHECK-NEXT:    ret <4 x i1> [[R]]
 ;
   %a1 = extractelement <4 x float> %a, i32 1
   %a2 = extractelement <4 x float> %a, i32 2
@@ -176,9 +148,9 @@ define <4 x i1> @ins_fcmp_ext_ext(<4 x float> %a, <4 x i1> %b) {
 
 define <4 x i1> @ins_icmp_ext_ext(<4 x i32> %a, <4 x i1> %b) {
 ; CHECK-LABEL: @ins_icmp_ext_ext(
-; CHECK-NEXT:    [[SHIFT:%.*]] = shufflevector <4 x i32> [[A:%.*]], <4 x i32> poison, <4 x i32> <i32 undef, i32 undef, i32 undef, i32 2>
-; CHECK-NEXT:    [[TMP1:%.*]] = icmp ule <4 x i32> [[SHIFT]], [[A]]
-; CHECK-NEXT:    [[A23:%.*]] = extractelement <4 x i1> [[TMP1]], i64 3
+; CHECK-NEXT:    [[A3:%.*]] = extractelement <4 x i32> [[A:%.*]], i32 3
+; CHECK-NEXT:    [[A2:%.*]] = extractelement <4 x i32> [[A]], i32 2
+; CHECK-NEXT:    [[A23:%.*]] = icmp ule i32 [[A2]], [[A3]]
 ; CHECK-NEXT:    [[R:%.*]] = insertelement <4 x i1> [[B:%.*]], i1 [[A23]], i32 3
 ; CHECK-NEXT:    ret <4 x i1> [[R]]
 ;

diff  --git a/llvm/test/Transforms/VectorCombine/X86/scalarize-cmp-inseltpoison.ll b/llvm/test/Transforms/VectorCombine/X86/scalarize-cmp-inseltpoison.ll
index a9214aa26e767..c82c42d8d3eae 100644
--- a/llvm/test/Transforms/VectorCombine/X86/scalarize-cmp-inseltpoison.ll
+++ b/llvm/test/Transforms/VectorCombine/X86/scalarize-cmp-inseltpoison.ll
@@ -132,7 +132,8 @@ define <4 x i1> @ins2_ins2_f32_uses(float %x, float %y) {
 ; CHECK-NEXT:    call void @usef(<4 x float> [[I0]])
 ; CHECK-NEXT:    [[I1:%.*]] = insertelement <4 x float> poison, float [[Y:%.*]], i32 2
 ; CHECK-NEXT:    call void @usef(<4 x float> [[I1]])
-; CHECK-NEXT:    [[R:%.*]] = fcmp oeq <4 x float> [[I0]], [[I1]]
+; CHECK-NEXT:    [[R_SCALAR:%.*]] = fcmp oeq float [[X]], [[Y]]
+; CHECK-NEXT:    [[R:%.*]] = insertelement <4 x i1> poison, i1 [[R_SCALAR]], i64 2
 ; CHECK-NEXT:    ret <4 x i1> [[R]]
 ;
   %i0 = insertelement <4 x float> poison, float %x, i32 2

diff  --git a/llvm/test/Transforms/VectorCombine/X86/scalarize-cmp.ll b/llvm/test/Transforms/VectorCombine/X86/scalarize-cmp.ll
index 85f521e7d0d14..3d32113771457 100644
--- a/llvm/test/Transforms/VectorCombine/X86/scalarize-cmp.ll
+++ b/llvm/test/Transforms/VectorCombine/X86/scalarize-cmp.ll
@@ -132,7 +132,8 @@ define <4 x i1> @ins2_ins2_f32_uses(float %x, float %y) {
 ; CHECK-NEXT:    call void @usef(<4 x float> [[I0]])
 ; CHECK-NEXT:    [[I1:%.*]] = insertelement <4 x float> undef, float [[Y:%.*]], i32 2
 ; CHECK-NEXT:    call void @usef(<4 x float> [[I1]])
-; CHECK-NEXT:    [[R:%.*]] = fcmp oeq <4 x float> [[I0]], [[I1]]
+; CHECK-NEXT:    [[R_SCALAR:%.*]] = fcmp oeq float [[X]], [[Y]]
+; CHECK-NEXT:    [[R:%.*]] = insertelement <4 x i1> zeroinitializer, i1 [[R_SCALAR]], i64 2
 ; CHECK-NEXT:    ret <4 x i1> [[R]]
 ;
   %i0 = insertelement <4 x float> undef, float %x, i32 2


        
