[llvm] 4185c55 - [CostModel][X86] Adjust shift SSE4 legalized costs based on llvm-mca reports.
Simon Pilgrim via llvm-commits
llvm-commits at lists.llvm.org
Thu Jul 22 12:10:36 PDT 2021
Author: Simon Pilgrim
Date: 2021-07-22T20:07:32+01:00
New Revision: 4185c5502c81285206ea35172411966c8bd8bd30
URL: https://github.com/llvm/llvm-project/commit/4185c5502c81285206ea35172411966c8bd8bd30
DIFF: https://github.com/llvm/llvm-project/commit/4185c5502c81285206ea35172411966c8bd8bd30.diff
LOG: [CostModel][X86] Adjust shift SSE4 legalized costs based on llvm-mca reports.
Update shl/lshr/ashr costs based on the worst case costs from the script in D103695 - many of the 128-bit shifts (usually where integer multiplies aren't used) have similar behaviour to AVX1 so we can merge them.
Added:
Modified:
llvm/lib/Target/X86/X86TargetTransformInfo.cpp
llvm/test/Analysis/CostModel/X86/div.ll
llvm/test/Analysis/CostModel/X86/fshl.ll
llvm/test/Analysis/CostModel/X86/fshr.ll
llvm/test/Analysis/CostModel/X86/rem.ll
llvm/test/Analysis/CostModel/X86/vshift-ashr-cost-inseltpoison.ll
llvm/test/Analysis/CostModel/X86/vshift-ashr-cost.ll
llvm/test/Analysis/CostModel/X86/vshift-lshr-cost-inseltpoison.ll
llvm/test/Analysis/CostModel/X86/vshift-lshr-cost.ll
llvm/test/Analysis/CostModel/X86/vshift-shl-cost-inseltpoison.ll
llvm/test/Analysis/CostModel/X86/vshift-shl-cost.ll
Removed:
################################################################################
diff --git a/llvm/lib/Target/X86/X86TargetTransformInfo.cpp b/llvm/lib/Target/X86/X86TargetTransformInfo.cpp
index 52bf27933583..28474dfaee9e 100644
--- a/llvm/lib/Target/X86/X86TargetTransformInfo.cpp
+++ b/llvm/lib/Target/X86/X86TargetTransformInfo.cpp
@@ -853,7 +853,6 @@ InstructionCost X86TTIImpl::getArithmeticInstrCost(
{ ISD::SUB, MVT::v4i64, 4 },
{ ISD::ADD, MVT::v4i64, 4 },
- { ISD::SHL, MVT::v16i8, 10 }, // pblendvb sequence .
{ ISD::SHL, MVT::v32i8, 22 }, // pblendvb sequence + split.
{ ISD::SHL, MVT::v8i16, 6 }, // pblendvb sequence.
{ ISD::SHL, MVT::v16i16, 13 }, // pblendvb sequence + split.
@@ -862,18 +861,14 @@ InstructionCost X86TTIImpl::getArithmeticInstrCost(
{ ISD::SHL, MVT::v2i64, 2 }, // Shift each lane + blend.
{ ISD::SHL, MVT::v4i64, 6 }, // Shift each lane + blend + split.
- { ISD::SRL, MVT::v16i8, 11 }, // pblendvb sequence.
{ ISD::SRL, MVT::v32i8, 23 }, // pblendvb sequence + split.
- { ISD::SRL, MVT::v8i16, 13 }, // pblendvb sequence.
{ ISD::SRL, MVT::v16i16, 28 }, // pblendvb sequence + split.
{ ISD::SRL, MVT::v4i32, 6 }, // Shift each lane + blend.
{ ISD::SRL, MVT::v8i32, 14 }, // Shift each lane + blend + split.
{ ISD::SRL, MVT::v2i64, 2 }, // Shift each lane + blend.
{ ISD::SRL, MVT::v4i64, 6 }, // Shift each lane + blend + split.
- { ISD::SRA, MVT::v16i8, 21 }, // pblendvb sequence.
{ ISD::SRA, MVT::v32i8, 44 }, // pblendvb sequence + split.
- { ISD::SRA, MVT::v8i16, 13 }, // pblendvb sequence.
{ ISD::SRA, MVT::v16i16, 28 }, // pblendvb sequence + split.
{ ISD::SRA, MVT::v4i32, 6 }, // Shift each lane + blend.
{ ISD::SRA, MVT::v8i32, 14 }, // Shift each lane + blend + split.
@@ -928,16 +923,16 @@ InstructionCost X86TTIImpl::getArithmeticInstrCost(
return LT.first * Entry->Cost;
static const CostTblEntry SSE41CostTable[] = {
- { ISD::SHL, MVT::v16i8, 11 }, // pblendvb sequence.
- { ISD::SHL, MVT::v8i16, 14 }, // pblendvb sequence.
+ { ISD::SHL, MVT::v16i8, 10 }, // pblendvb sequence.
+ { ISD::SHL, MVT::v8i16, 11 }, // pblendvb sequence.
{ ISD::SHL, MVT::v4i32, 4 }, // pslld/paddd/cvttps2dq/pmulld
- { ISD::SRL, MVT::v16i8, 12 }, // pblendvb sequence.
- { ISD::SRL, MVT::v8i16, 14 }, // pblendvb sequence.
- { ISD::SRL, MVT::v4i32, 11 }, // Shift each lane + blend.
+ { ISD::SRL, MVT::v16i8, 11 }, // pblendvb sequence.
+ { ISD::SRL, MVT::v8i16, 13 }, // pblendvb sequence.
+ { ISD::SRL, MVT::v4i32, 16 }, // Shift each lane + blend.
- { ISD::SRA, MVT::v16i8, 24 }, // pblendvb sequence.
- { ISD::SRA, MVT::v8i16, 14 }, // pblendvb sequence.
+ { ISD::SRA, MVT::v16i8, 21 }, // pblendvb sequence.
+ { ISD::SRA, MVT::v8i16, 13 }, // pblendvb sequence.
{ ISD::MUL, MVT::v4i32, 2 } // pmulld (Nehalem from agner.org)
};
diff --git a/llvm/test/Analysis/CostModel/X86/div.ll b/llvm/test/Analysis/CostModel/X86/div.ll
index 4c22d08d4974..9c9269b8b5c5 100644
--- a/llvm/test/Analysis/CostModel/X86/div.ll
+++ b/llvm/test/Analysis/CostModel/X86/div.ll
@@ -751,17 +751,17 @@ define i32 @sdiv_constpow2() {
; SSE42-NEXT: Cost Model: Found an estimated cost of 42 for instruction: %V4i64 = sdiv <4 x i64> undef, <i64 2, i64 4, i64 8, i64 16>
; SSE42-NEXT: Cost Model: Found an estimated cost of 84 for instruction: %V8i64 = sdiv <8 x i64> undef, <i64 2, i64 4, i64 8, i64 16, i64 32, i64 64, i64 128, i64 256>
; SSE42-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %I32 = sdiv i32 undef, 16
-; SSE42-NEXT: Cost Model: Found an estimated cost of 36 for instruction: %V4i32 = sdiv <4 x i32> undef, <i32 2, i32 4, i32 8, i32 16>
-; SSE42-NEXT: Cost Model: Found an estimated cost of 72 for instruction: %V8i32 = sdiv <8 x i32> undef, <i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256>
-; SSE42-NEXT: Cost Model: Found an estimated cost of 144 for instruction: %V16i32 = sdiv <16 x i32> undef, <i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256, i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256>
+; SSE42-NEXT: Cost Model: Found an estimated cost of 41 for instruction: %V4i32 = sdiv <4 x i32> undef, <i32 2, i32 4, i32 8, i32 16>
+; SSE42-NEXT: Cost Model: Found an estimated cost of 82 for instruction: %V8i32 = sdiv <8 x i32> undef, <i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256>
+; SSE42-NEXT: Cost Model: Found an estimated cost of 164 for instruction: %V16i32 = sdiv <16 x i32> undef, <i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256, i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256>
; SSE42-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %I16 = sdiv i16 undef, 16
-; SSE42-NEXT: Cost Model: Found an estimated cost of 43 for instruction: %V8i16 = sdiv <8 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
-; SSE42-NEXT: Cost Model: Found an estimated cost of 86 for instruction: %V16i16 = sdiv <16 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
-; SSE42-NEXT: Cost Model: Found an estimated cost of 172 for instruction: %V32i16 = sdiv <32 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
+; SSE42-NEXT: Cost Model: Found an estimated cost of 40 for instruction: %V8i16 = sdiv <8 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
+; SSE42-NEXT: Cost Model: Found an estimated cost of 80 for instruction: %V16i16 = sdiv <16 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
+; SSE42-NEXT: Cost Model: Found an estimated cost of 160 for instruction: %V32i16 = sdiv <32 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
; SSE42-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %I8 = sdiv i8 undef, 16
-; SSE42-NEXT: Cost Model: Found an estimated cost of 61 for instruction: %V16i8 = sdiv <16 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
-; SSE42-NEXT: Cost Model: Found an estimated cost of 122 for instruction: %V32i8 = sdiv <32 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
-; SSE42-NEXT: Cost Model: Found an estimated cost of 244 for instruction: %V64i8 = sdiv <64 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
+; SSE42-NEXT: Cost Model: Found an estimated cost of 54 for instruction: %V16i8 = sdiv <16 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
+; SSE42-NEXT: Cost Model: Found an estimated cost of 108 for instruction: %V32i8 = sdiv <32 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
+; SSE42-NEXT: Cost Model: Found an estimated cost of 216 for instruction: %V64i8 = sdiv <64 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
;
; AVX1-LABEL: 'sdiv_constpow2'
@@ -846,17 +846,17 @@ define i32 @sdiv_constpow2() {
; SLM-NEXT: Cost Model: Found an estimated cost of 48 for instruction: %V4i64 = sdiv <4 x i64> undef, <i64 2, i64 4, i64 8, i64 16>
; SLM-NEXT: Cost Model: Found an estimated cost of 96 for instruction: %V8i64 = sdiv <8 x i64> undef, <i64 2, i64 4, i64 8, i64 16, i64 32, i64 64, i64 128, i64 256>
; SLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %I32 = sdiv i32 undef, 16
-; SLM-NEXT: Cost Model: Found an estimated cost of 36 for instruction: %V4i32 = sdiv <4 x i32> undef, <i32 2, i32 4, i32 8, i32 16>
-; SLM-NEXT: Cost Model: Found an estimated cost of 72 for instruction: %V8i32 = sdiv <8 x i32> undef, <i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256>
-; SLM-NEXT: Cost Model: Found an estimated cost of 144 for instruction: %V16i32 = sdiv <16 x i32> undef, <i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256, i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256>
+; SLM-NEXT: Cost Model: Found an estimated cost of 41 for instruction: %V4i32 = sdiv <4 x i32> undef, <i32 2, i32 4, i32 8, i32 16>
+; SLM-NEXT: Cost Model: Found an estimated cost of 82 for instruction: %V8i32 = sdiv <8 x i32> undef, <i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256>
+; SLM-NEXT: Cost Model: Found an estimated cost of 164 for instruction: %V16i32 = sdiv <16 x i32> undef, <i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256, i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256>
; SLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %I16 = sdiv i16 undef, 16
-; SLM-NEXT: Cost Model: Found an estimated cost of 43 for instruction: %V8i16 = sdiv <8 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
-; SLM-NEXT: Cost Model: Found an estimated cost of 86 for instruction: %V16i16 = sdiv <16 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
-; SLM-NEXT: Cost Model: Found an estimated cost of 172 for instruction: %V32i16 = sdiv <32 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
+; SLM-NEXT: Cost Model: Found an estimated cost of 40 for instruction: %V8i16 = sdiv <8 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
+; SLM-NEXT: Cost Model: Found an estimated cost of 80 for instruction: %V16i16 = sdiv <16 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
+; SLM-NEXT: Cost Model: Found an estimated cost of 160 for instruction: %V32i16 = sdiv <32 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
; SLM-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %I8 = sdiv i8 undef, 16
-; SLM-NEXT: Cost Model: Found an estimated cost of 61 for instruction: %V16i8 = sdiv <16 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
-; SLM-NEXT: Cost Model: Found an estimated cost of 122 for instruction: %V32i8 = sdiv <32 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
-; SLM-NEXT: Cost Model: Found an estimated cost of 244 for instruction: %V64i8 = sdiv <64 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
+; SLM-NEXT: Cost Model: Found an estimated cost of 54 for instruction: %V16i8 = sdiv <16 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
+; SLM-NEXT: Cost Model: Found an estimated cost of 108 for instruction: %V32i8 = sdiv <32 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
+; SLM-NEXT: Cost Model: Found an estimated cost of 216 for instruction: %V64i8 = sdiv <64 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
; SLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
;
; BTVER2-LABEL: 'sdiv_constpow2'
@@ -946,17 +946,17 @@ define i32 @udiv_constpow2() {
; SSE42-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4i64 = udiv <4 x i64> undef, <i64 2, i64 4, i64 8, i64 16>
; SSE42-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8i64 = udiv <8 x i64> undef, <i64 2, i64 4, i64 8, i64 16, i64 32, i64 64, i64 128, i64 256>
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I32 = udiv i32 undef, 16
-; SSE42-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %V4i32 = udiv <4 x i32> undef, <i32 2, i32 4, i32 8, i32 16>
-; SSE42-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %V8i32 = udiv <8 x i32> undef, <i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256>
-; SSE42-NEXT: Cost Model: Found an estimated cost of 44 for instruction: %V16i32 = udiv <16 x i32> undef, <i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256, i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256>
+; SSE42-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V4i32 = udiv <4 x i32> undef, <i32 2, i32 4, i32 8, i32 16>
+; SSE42-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %V8i32 = udiv <8 x i32> undef, <i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256>
+; SSE42-NEXT: Cost Model: Found an estimated cost of 64 for instruction: %V16i32 = udiv <16 x i32> undef, <i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256, i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256>
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I16 = udiv i16 undef, 16
-; SSE42-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V8i16 = udiv <8 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
-; SSE42-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V16i16 = udiv <16 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
-; SSE42-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %V32i16 = udiv <32 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
+; SSE42-NEXT: Cost Model: Found an estimated cost of 13 for instruction: %V8i16 = udiv <8 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
+; SSE42-NEXT: Cost Model: Found an estimated cost of 26 for instruction: %V16i16 = udiv <16 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
+; SSE42-NEXT: Cost Model: Found an estimated cost of 52 for instruction: %V32i16 = udiv <32 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I8 = udiv i8 undef, 16
-; SSE42-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %V16i8 = udiv <16 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
-; SSE42-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %V32i8 = udiv <32 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
-; SSE42-NEXT: Cost Model: Found an estimated cost of 48 for instruction: %V64i8 = udiv <64 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
+; SSE42-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %V16i8 = udiv <16 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
+; SSE42-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %V32i8 = udiv <32 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
+; SSE42-NEXT: Cost Model: Found an estimated cost of 44 for instruction: %V64i8 = udiv <64 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
;
; AVX1-LABEL: 'udiv_constpow2'
@@ -1041,17 +1041,17 @@ define i32 @udiv_constpow2() {
; SLM-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4i64 = udiv <4 x i64> undef, <i64 2, i64 4, i64 8, i64 16>
; SLM-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8i64 = udiv <8 x i64> undef, <i64 2, i64 4, i64 8, i64 16, i64 32, i64 64, i64 128, i64 256>
; SLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I32 = udiv i32 undef, 16
-; SLM-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %V4i32 = udiv <4 x i32> undef, <i32 2, i32 4, i32 8, i32 16>
-; SLM-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %V8i32 = udiv <8 x i32> undef, <i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256>
-; SLM-NEXT: Cost Model: Found an estimated cost of 44 for instruction: %V16i32 = udiv <16 x i32> undef, <i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256, i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256>
+; SLM-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V4i32 = udiv <4 x i32> undef, <i32 2, i32 4, i32 8, i32 16>
+; SLM-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %V8i32 = udiv <8 x i32> undef, <i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256>
+; SLM-NEXT: Cost Model: Found an estimated cost of 64 for instruction: %V16i32 = udiv <16 x i32> undef, <i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256, i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256>
; SLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I16 = udiv i16 undef, 16
-; SLM-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V8i16 = udiv <8 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
-; SLM-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V16i16 = udiv <16 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
-; SLM-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %V32i16 = udiv <32 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
+; SLM-NEXT: Cost Model: Found an estimated cost of 13 for instruction: %V8i16 = udiv <8 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
+; SLM-NEXT: Cost Model: Found an estimated cost of 26 for instruction: %V16i16 = udiv <16 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
+; SLM-NEXT: Cost Model: Found an estimated cost of 52 for instruction: %V32i16 = udiv <32 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
; SLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I8 = udiv i8 undef, 16
-; SLM-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %V16i8 = udiv <16 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
-; SLM-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %V32i8 = udiv <32 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
-; SLM-NEXT: Cost Model: Found an estimated cost of 48 for instruction: %V64i8 = udiv <64 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
+; SLM-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %V16i8 = udiv <16 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
+; SLM-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %V32i8 = udiv <32 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
+; SLM-NEXT: Cost Model: Found an estimated cost of 44 for instruction: %V64i8 = udiv <64 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
; SLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
;
; BTVER2-LABEL: 'udiv_constpow2'
diff --git a/llvm/test/Analysis/CostModel/X86/fshl.ll b/llvm/test/Analysis/CostModel/X86/fshl.ll
index f445143e0b9b..8ea89ff69d0a 100644
--- a/llvm/test/Analysis/CostModel/X86/fshl.ll
+++ b/llvm/test/Analysis/CostModel/X86/fshl.ll
@@ -93,9 +93,9 @@ define void @var_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16 x i3
;
; 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 20 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 40 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 80 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 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 0 for instruction: ret void
;
; AVX1-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 20 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 40 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 80 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 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 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 20 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 40 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 80 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 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 0 for instruction: ret void
;
; XOP-LABEL: 'var_funnel_i32'
@@ -157,9 +157,9 @@ define void @var_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <32 x i
;
; 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 33 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 66 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 132 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 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 0 for instruction: ret void
;
; AVX1-LABEL: 'var_funnel_i16'
@@ -199,16 +199,16 @@ define void @var_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <32 x i
;
; 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 33 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 66 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 132 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 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 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 33 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 66 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 132 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 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 0 for instruction: ret void
;
; XOP-LABEL: 'var_funnel_i16'
@@ -235,9 +235,9 @@ define void @var_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x i8> %
;
; 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 28 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 56 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 112 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 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 0 for instruction: ret void
;
; AVX1-LABEL: 'var_funnel_i8'
@@ -277,16 +277,16 @@ define void @var_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x i8> %
;
; 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 28 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 56 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 112 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 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 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 28 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 56 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 112 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 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 0 for instruction: ret void
;
; XOP-LABEL: 'var_funnel_i8'
@@ -585,9 +585,9 @@ define void @splatvar_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x
; 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 28 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 56 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 112 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 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 0 for instruction: ret void
;
; AVX1-LABEL: 'splatvar_funnel_i8'
@@ -639,18 +639,18 @@ define void @splatvar_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x
; 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 28 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 56 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 112 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 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 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 28 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 56 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 112 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 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 0 for instruction: ret void
;
; XOP-LABEL: 'splatvar_funnel_i8'
@@ -749,9 +749,9 @@ define void @constant_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16
;
; 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 17 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 34 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 68 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 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 0 for instruction: ret void
;
; AVX1-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 17 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 34 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 68 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 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 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 17 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 34 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 68 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 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 0 for instruction: ret void
;
; XOP-LABEL: 'constant_funnel_i32'
@@ -813,9 +813,9 @@ define void @constant_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <3
;
; 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 19 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 38 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 76 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 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 0 for instruction: ret void
;
; AVX1-LABEL: 'constant_funnel_i16'
@@ -855,16 +855,16 @@ define void @constant_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <3
;
; 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 19 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 38 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 76 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 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 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 19 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 38 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 76 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 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 0 for instruction: ret void
;
; XOP-LABEL: 'constant_funnel_i16'
@@ -891,9 +891,9 @@ define void @constant_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x
;
; 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 27 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 54 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 108 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 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 0 for instruction: ret void
;
; AVX1-LABEL: 'constant_funnel_i8'
@@ -933,16 +933,16 @@ define void @constant_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x
;
; 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 27 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 54 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 108 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 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 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 27 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 54 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 108 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 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 0 for instruction: ret void
;
; XOP-LABEL: 'constant_funnel_i8'
@@ -1318,9 +1318,9 @@ define void @var_rotate_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16 x i3
;
; SSE42-LABEL: 'var_rotate_i32'
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %a32, i32 %c32)
-; SSE42-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %a128, <4 x i32> %c128)
-; SSE42-NEXT: Cost Model: Found an estimated cost of 36 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %a256, <8 x i32> %c256)
-; SSE42-NEXT: Cost Model: Found an estimated cost of 72 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %a512, <16 x i32> %c512)
+; SSE42-NEXT: Cost Model: Found an estimated cost of 23 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %a128, <4 x i32> %c128)
+; SSE42-NEXT: Cost Model: Found an estimated cost of 46 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %a256, <8 x i32> %c256)
+; SSE42-NEXT: Cost Model: Found an estimated cost of 92 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %a512, <16 x i32> %c512)
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; AVX1-LABEL: 'var_rotate_i32'
@@ -1346,16 +1346,16 @@ define void @var_rotate_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16 x i3
;
; SLM-LABEL: 'var_rotate_i32'
; SLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %a32, i32 %c32)
-; SLM-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %a128, <4 x i32> %c128)
-; SLM-NEXT: Cost Model: Found an estimated cost of 36 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %a256, <8 x i32> %c256)
-; SLM-NEXT: Cost Model: Found an estimated cost of 72 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %a512, <16 x i32> %c512)
+; SLM-NEXT: Cost Model: Found an estimated cost of 23 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %a128, <4 x i32> %c128)
+; SLM-NEXT: Cost Model: Found an estimated cost of 46 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %a256, <8 x i32> %c256)
+; SLM-NEXT: Cost Model: Found an estimated cost of 92 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %a512, <16 x i32> %c512)
; SLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; GLM-LABEL: 'var_rotate_i32'
; GLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %a32, i32 %c32)
-; GLM-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %a128, <4 x i32> %c128)
-; GLM-NEXT: Cost Model: Found an estimated cost of 36 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %a256, <8 x i32> %c256)
-; GLM-NEXT: Cost Model: Found an estimated cost of 72 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %a512, <16 x i32> %c512)
+; GLM-NEXT: Cost Model: Found an estimated cost of 23 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %a128, <4 x i32> %c128)
+; GLM-NEXT: Cost Model: Found an estimated cost of 46 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %a256, <8 x i32> %c256)
+; GLM-NEXT: Cost Model: Found an estimated cost of 92 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %a512, <16 x i32> %c512)
; GLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; XOP-LABEL: 'var_rotate_i32'
@@ -1382,9 +1382,9 @@ define void @var_rotate_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <32 x i
;
; SSE42-LABEL: 'var_rotate_i16'
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %a16, i16 %c16)
-; SSE42-NEXT: Cost Model: Found an estimated cost of 31 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %a128, <8 x i16> %c128)
-; SSE42-NEXT: Cost Model: Found an estimated cost of 62 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %a256, <16 x i16> %c256)
-; SSE42-NEXT: Cost Model: Found an estimated cost of 124 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %a512, <32 x i16> %c512)
+; SSE42-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> %a128, <8 x i16> %c128)
+; SSE42-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> %a256, <16 x i16> %c256)
+; SSE42-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> %a512, <32 x i16> %c512)
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; AVX1-LABEL: 'var_rotate_i16'
@@ -1424,16 +1424,16 @@ define void @var_rotate_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <32 x i
;
; SLM-LABEL: 'var_rotate_i16'
; SLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %a16, i16 %c16)
-; SLM-NEXT: Cost Model: Found an estimated cost of 31 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %a128, <8 x i16> %c128)
-; SLM-NEXT: Cost Model: Found an estimated cost of 62 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %a256, <16 x i16> %c256)
-; SLM-NEXT: Cost Model: Found an estimated cost of 124 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %a512, <32 x i16> %c512)
+; SLM-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> %a128, <8 x i16> %c128)
+; SLM-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> %a256, <16 x i16> %c256)
+; SLM-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> %a512, <32 x i16> %c512)
; SLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; GLM-LABEL: 'var_rotate_i16'
; GLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %a16, i16 %c16)
-; GLM-NEXT: Cost Model: Found an estimated cost of 31 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %a128, <8 x i16> %c128)
-; GLM-NEXT: Cost Model: Found an estimated cost of 62 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %a256, <16 x i16> %c256)
-; GLM-NEXT: Cost Model: Found an estimated cost of 124 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %a512, <32 x i16> %c512)
+; GLM-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> %a128, <8 x i16> %c128)
+; GLM-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> %a256, <16 x i16> %c256)
+; GLM-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> %a512, <32 x i16> %c512)
; GLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; XOP-LABEL: 'var_rotate_i16'
@@ -1460,9 +1460,9 @@ define void @var_rotate_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x i8> %
;
; SSE42-LABEL: 'var_rotate_i8'
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %a8, 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> %a128, <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> %a256, <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> %a512, <64 x i8> %c512)
+; SSE42-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %a128, <16 x i8> %c128)
+; SSE42-NEXT: Cost Model: Found an estimated cost of 48 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %a256, <32 x i8> %c256)
+; SSE42-NEXT: Cost Model: Found an estimated cost of 96 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %a512, <64 x i8> %c512)
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; AVX1-LABEL: 'var_rotate_i8'
@@ -1502,16 +1502,16 @@ define void @var_rotate_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x i8> %
;
; SLM-LABEL: 'var_rotate_i8'
; SLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %a8, 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> %a128, <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> %a256, <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> %a512, <64 x i8> %c512)
+; SLM-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %a128, <16 x i8> %c128)
+; SLM-NEXT: Cost Model: Found an estimated cost of 48 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %a256, <32 x i8> %c256)
+; SLM-NEXT: Cost Model: Found an estimated cost of 96 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %a512, <64 x i8> %c512)
; SLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; GLM-LABEL: 'var_rotate_i8'
; GLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %a8, 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> %a128, <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> %a256, <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> %a512, <64 x i8> %c512)
+; GLM-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %a128, <16 x i8> %c128)
+; GLM-NEXT: Cost Model: Found an estimated cost of 48 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %a256, <32 x i8> %c256)
+; GLM-NEXT: Cost Model: Found an estimated cost of 96 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %a512, <64 x i8> %c512)
; GLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; XOP-LABEL: 'var_rotate_i8'
@@ -1783,9 +1783,9 @@ define void @splatvar_rotate_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x
; 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> %a128, <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> %a256, <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> %a512, <64 x i8> %u512)
+; SSE42-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %a128, <16 x i8> %u128)
+; SSE42-NEXT: Cost Model: Found an estimated cost of 48 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %a256, <32 x i8> %u256)
+; SSE42-NEXT: Cost Model: Found an estimated cost of 96 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %a512, <64 x i8> %u512)
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; AVX1-LABEL: 'splatvar_rotate_i8'
@@ -1837,18 +1837,18 @@ define void @splatvar_rotate_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x
; 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> %a128, <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> %a256, <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> %a512, <64 x i8> %u512)
+; SLM-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %a128, <16 x i8> %u128)
+; SLM-NEXT: Cost Model: Found an estimated cost of 48 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %a256, <32 x i8> %u256)
+; SLM-NEXT: Cost Model: Found an estimated cost of 96 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %a512, <64 x i8> %u512)
; SLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; GLM-LABEL: 'splatvar_rotate_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> %a128, <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> %a256, <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> %a512, <64 x i8> %u512)
+; GLM-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %a128, <16 x i8> %u128)
+; GLM-NEXT: Cost Model: Found an estimated cost of 48 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %a256, <32 x i8> %u256)
+; GLM-NEXT: Cost Model: Found an estimated cost of 96 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %a512, <64 x i8> %u512)
; GLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; XOP-LABEL: 'splatvar_rotate_i8'
@@ -1931,19 +1931,12 @@ define void @constant_rotate_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256, <8
}
define void @constant_rotate_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16 x i32> %a512) {
-; SSSE3-LABEL: 'constant_rotate_i32'
-; SSSE3-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %a32, i32 7)
-; SSSE3-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %a128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
-; SSSE3-NEXT: Cost Model: Found an estimated cost of 40 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %a256, <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 80 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %a512, <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_rotate_i32'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %a32, i32 7)
-; SSE42-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> %a128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
-; SSE42-NEXT: Cost Model: Found an estimated cost of 30 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %a256, <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 60 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %a512, <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
+; SSE-LABEL: 'constant_rotate_i32'
+; SSE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %a32, i32 7)
+; SSE-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %a128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
+; SSE-NEXT: Cost Model: Found an estimated cost of 40 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %a256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
+; SSE-NEXT: Cost Model: Found an estimated cost of 80 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %a512, <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>)
+; SSE-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; AVX1-LABEL: 'constant_rotate_i32'
; AVX1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %a32, i32 7)
@@ -1968,16 +1961,16 @@ define void @constant_rotate_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16
;
; SLM-LABEL: 'constant_rotate_i32'
; SLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %a32, i32 7)
-; SLM-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> %a128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
-; SLM-NEXT: Cost Model: Found an estimated cost of 30 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %a256, <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 60 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %a512, <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 20 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %a128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
+; SLM-NEXT: Cost Model: Found an estimated cost of 40 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %a256, <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 80 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %a512, <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_rotate_i32'
; GLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %a32, i32 7)
-; GLM-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> %a128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
-; GLM-NEXT: Cost Model: Found an estimated cost of 30 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %a256, <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 60 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %a512, <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 20 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %a128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
+; GLM-NEXT: Cost Model: Found an estimated cost of 40 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %a256, <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 80 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %a512, <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_rotate_i32'
@@ -2004,9 +1997,9 @@ define void @constant_rotate_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <3
;
; SSE42-LABEL: 'constant_rotate_i16'
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %a16, i16 7)
-; SSE42-NEXT: Cost Model: Found an estimated cost of 17 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %a128, <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 34 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %a256, <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 68 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %a512, <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 16 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %a128, <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 32 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %a256, <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 64 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %a512, <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_rotate_i16'
@@ -2046,16 +2039,16 @@ define void @constant_rotate_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <3
;
; SLM-LABEL: 'constant_rotate_i16'
; SLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %a16, i16 7)
-; SLM-NEXT: Cost Model: Found an estimated cost of 17 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %a128, <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 34 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %a256, <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 68 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %a512, <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 16 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %a128, <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 32 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %a256, <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 64 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %a512, <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_rotate_i16'
; GLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %a16, i16 7)
-; GLM-NEXT: Cost Model: Found an estimated cost of 17 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %a128, <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 34 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %a256, <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 68 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %a512, <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 16 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %a128, <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 32 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %a256, <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 64 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %a512, <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_rotate_i16'
@@ -2082,9 +2075,9 @@ define void @constant_rotate_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x
;
; SSE42-LABEL: 'constant_rotate_i8'
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %a8, 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> %a128, <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> %a256, <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> %a512, <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 23 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %a128, <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 46 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %a256, <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 92 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %a512, <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_rotate_i8'
@@ -2124,16 +2117,16 @@ define void @constant_rotate_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x
;
; SLM-LABEL: 'constant_rotate_i8'
; SLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %a8, 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> %a128, <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> %a256, <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> %a512, <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 23 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %a128, <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 46 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %a256, <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 92 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %a512, <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_rotate_i8'
; GLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %a8, 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> %a128, <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> %a256, <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> %a512, <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 23 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %a128, <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 46 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %a256, <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 92 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %a512, <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_rotate_i8'
diff --git a/llvm/test/Analysis/CostModel/X86/fshr.ll b/llvm/test/Analysis/CostModel/X86/fshr.ll
index 07d7d6f85590..9d4f529be062 100644
--- a/llvm/test/Analysis/CostModel/X86/fshr.ll
+++ b/llvm/test/Analysis/CostModel/X86/fshr.ll
@@ -93,9 +93,9 @@ define void @var_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16 x i3
;
; 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 20 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 40 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 80 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 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 0 for instruction: ret void
;
; AVX1-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 20 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 40 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 80 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 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 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 20 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 40 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 80 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 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 0 for instruction: ret void
;
; XOP-LABEL: 'var_funnel_i32'
@@ -157,9 +157,9 @@ define void @var_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <32 x i
;
; 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 33 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 66 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 132 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 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 0 for instruction: ret void
;
; AVX1-LABEL: 'var_funnel_i16'
@@ -199,16 +199,16 @@ define void @var_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <32 x i
;
; 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 33 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 66 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 132 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 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 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 33 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 66 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 132 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 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 0 for instruction: ret void
;
; XOP-LABEL: 'var_funnel_i16'
@@ -235,9 +235,9 @@ define void @var_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x i8> %
;
; 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 28 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 56 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 112 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 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 0 for instruction: ret void
;
; AVX1-LABEL: 'var_funnel_i8'
@@ -277,16 +277,16 @@ define void @var_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x i8> %
;
; 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 28 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 56 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 112 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 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 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 28 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 56 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 112 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 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 0 for instruction: ret void
;
; XOP-LABEL: 'var_funnel_i8'
@@ -585,9 +585,9 @@ define void @splatvar_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x
; 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 28 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 56 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 112 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 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 0 for instruction: ret void
;
; AVX1-LABEL: 'splatvar_funnel_i8'
@@ -639,18 +639,18 @@ define void @splatvar_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x
; 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 28 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 56 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 112 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 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 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 28 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 56 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 112 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 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 0 for instruction: ret void
;
; XOP-LABEL: 'splatvar_funnel_i8'
@@ -749,9 +749,9 @@ define void @constant_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16
;
; 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 17 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 34 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 68 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 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 0 for instruction: ret void
;
; AVX1-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 17 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 34 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 68 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 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 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 17 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 34 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 68 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 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 0 for instruction: ret void
;
; XOP-LABEL: 'constant_funnel_i32'
@@ -813,9 +813,9 @@ define void @constant_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <3
;
; 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 19 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 38 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 76 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 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 0 for instruction: ret void
;
; AVX1-LABEL: 'constant_funnel_i16'
@@ -855,16 +855,16 @@ define void @constant_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <3
;
; 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 19 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 38 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 76 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 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 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 19 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 38 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 76 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 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 0 for instruction: ret void
;
; XOP-LABEL: 'constant_funnel_i16'
@@ -891,9 +891,9 @@ define void @constant_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x
;
; 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 27 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 54 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 108 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 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 0 for instruction: ret void
;
; AVX1-LABEL: 'constant_funnel_i8'
@@ -933,16 +933,16 @@ define void @constant_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x
;
; 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 27 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 54 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 108 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 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 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 27 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 54 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 108 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 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 0 for instruction: ret void
;
; XOP-LABEL: 'constant_funnel_i8'
@@ -1319,9 +1319,9 @@ define void @var_rotate_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16 x i3
;
; SSE42-LABEL: 'var_rotate_i32'
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %a32, i32 %c32)
-; SSE42-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %a128, <4 x i32> %c128)
-; SSE42-NEXT: Cost Model: Found an estimated cost of 36 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %a256, <8 x i32> %c256)
-; SSE42-NEXT: Cost Model: Found an estimated cost of 72 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %a512, <16 x i32> %c512)
+; SSE42-NEXT: Cost Model: Found an estimated cost of 23 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %a128, <4 x i32> %c128)
+; SSE42-NEXT: Cost Model: Found an estimated cost of 46 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %a256, <8 x i32> %c256)
+; SSE42-NEXT: Cost Model: Found an estimated cost of 92 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %a512, <16 x i32> %c512)
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; AVX1-LABEL: 'var_rotate_i32'
@@ -1347,16 +1347,16 @@ define void @var_rotate_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16 x i3
;
; SLM-LABEL: 'var_rotate_i32'
; SLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %a32, i32 %c32)
-; SLM-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %a128, <4 x i32> %c128)
-; SLM-NEXT: Cost Model: Found an estimated cost of 36 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %a256, <8 x i32> %c256)
-; SLM-NEXT: Cost Model: Found an estimated cost of 72 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %a512, <16 x i32> %c512)
+; SLM-NEXT: Cost Model: Found an estimated cost of 23 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %a128, <4 x i32> %c128)
+; SLM-NEXT: Cost Model: Found an estimated cost of 46 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %a256, <8 x i32> %c256)
+; SLM-NEXT: Cost Model: Found an estimated cost of 92 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %a512, <16 x i32> %c512)
; SLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; GLM-LABEL: 'var_rotate_i32'
; GLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %a32, i32 %c32)
-; GLM-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %a128, <4 x i32> %c128)
-; GLM-NEXT: Cost Model: Found an estimated cost of 36 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %a256, <8 x i32> %c256)
-; GLM-NEXT: Cost Model: Found an estimated cost of 72 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %a512, <16 x i32> %c512)
+; GLM-NEXT: Cost Model: Found an estimated cost of 23 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %a128, <4 x i32> %c128)
+; GLM-NEXT: Cost Model: Found an estimated cost of 46 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %a256, <8 x i32> %c256)
+; GLM-NEXT: Cost Model: Found an estimated cost of 92 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %a512, <16 x i32> %c512)
; GLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; XOP-LABEL: 'var_rotate_i32'
@@ -1383,9 +1383,9 @@ define void @var_rotate_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <32 x i
;
; SSE42-LABEL: 'var_rotate_i16'
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %a16, i16 %c16)
-; SSE42-NEXT: Cost Model: Found an estimated cost of 31 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %a128, <8 x i16> %c128)
-; SSE42-NEXT: Cost Model: Found an estimated cost of 62 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %a256, <16 x i16> %c256)
-; SSE42-NEXT: Cost Model: Found an estimated cost of 124 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %a512, <32 x i16> %c512)
+; SSE42-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> %a128, <8 x i16> %c128)
+; SSE42-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> %a256, <16 x i16> %c256)
+; SSE42-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> %a512, <32 x i16> %c512)
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; AVX1-LABEL: 'var_rotate_i16'
@@ -1425,16 +1425,16 @@ define void @var_rotate_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <32 x i
;
; SLM-LABEL: 'var_rotate_i16'
; SLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %a16, i16 %c16)
-; SLM-NEXT: Cost Model: Found an estimated cost of 31 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %a128, <8 x i16> %c128)
-; SLM-NEXT: Cost Model: Found an estimated cost of 62 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %a256, <16 x i16> %c256)
-; SLM-NEXT: Cost Model: Found an estimated cost of 124 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %a512, <32 x i16> %c512)
+; SLM-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> %a128, <8 x i16> %c128)
+; SLM-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> %a256, <16 x i16> %c256)
+; SLM-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> %a512, <32 x i16> %c512)
; SLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; GLM-LABEL: 'var_rotate_i16'
; GLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %a16, i16 %c16)
-; GLM-NEXT: Cost Model: Found an estimated cost of 31 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %a128, <8 x i16> %c128)
-; GLM-NEXT: Cost Model: Found an estimated cost of 62 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %a256, <16 x i16> %c256)
-; GLM-NEXT: Cost Model: Found an estimated cost of 124 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %a512, <32 x i16> %c512)
+; GLM-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> %a128, <8 x i16> %c128)
+; GLM-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> %a256, <16 x i16> %c256)
+; GLM-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> %a512, <32 x i16> %c512)
; GLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; XOP-LABEL: 'var_rotate_i16'
@@ -1461,9 +1461,9 @@ define void @var_rotate_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x i8> %
;
; SSE42-LABEL: 'var_rotate_i8'
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %a8, 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> %a128, <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> %a256, <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> %a512, <64 x i8> %c512)
+; SSE42-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %a128, <16 x i8> %c128)
+; SSE42-NEXT: Cost Model: Found an estimated cost of 48 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %a256, <32 x i8> %c256)
+; SSE42-NEXT: Cost Model: Found an estimated cost of 96 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %a512, <64 x i8> %c512)
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; AVX1-LABEL: 'var_rotate_i8'
@@ -1503,16 +1503,16 @@ define void @var_rotate_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x i8> %
;
; SLM-LABEL: 'var_rotate_i8'
; SLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %a8, 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> %a128, <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> %a256, <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> %a512, <64 x i8> %c512)
+; SLM-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %a128, <16 x i8> %c128)
+; SLM-NEXT: Cost Model: Found an estimated cost of 48 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %a256, <32 x i8> %c256)
+; SLM-NEXT: Cost Model: Found an estimated cost of 96 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %a512, <64 x i8> %c512)
; SLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; GLM-LABEL: 'var_rotate_i8'
; GLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %a8, 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> %a128, <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> %a256, <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> %a512, <64 x i8> %c512)
+; GLM-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %a128, <16 x i8> %c128)
+; GLM-NEXT: Cost Model: Found an estimated cost of 48 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %a256, <32 x i8> %c256)
+; GLM-NEXT: Cost Model: Found an estimated cost of 96 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %a512, <64 x i8> %c512)
; GLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; XOP-LABEL: 'var_rotate_i8'
@@ -1784,9 +1784,9 @@ define void @splatvar_rotate_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x
; 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> %a128, <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> %a256, <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> %a512, <64 x i8> %u512)
+; SSE42-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %a128, <16 x i8> %u128)
+; SSE42-NEXT: Cost Model: Found an estimated cost of 48 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %a256, <32 x i8> %u256)
+; SSE42-NEXT: Cost Model: Found an estimated cost of 96 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %a512, <64 x i8> %u512)
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; AVX1-LABEL: 'splatvar_rotate_i8'
@@ -1838,18 +1838,18 @@ define void @splatvar_rotate_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x
; 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> %a128, <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> %a256, <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> %a512, <64 x i8> %u512)
+; SLM-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %a128, <16 x i8> %u128)
+; SLM-NEXT: Cost Model: Found an estimated cost of 48 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %a256, <32 x i8> %u256)
+; SLM-NEXT: Cost Model: Found an estimated cost of 96 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %a512, <64 x i8> %u512)
; SLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; GLM-LABEL: 'splatvar_rotate_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> %a128, <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> %a256, <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> %a512, <64 x i8> %u512)
+; GLM-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %a128, <16 x i8> %u128)
+; GLM-NEXT: Cost Model: Found an estimated cost of 48 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %a256, <32 x i8> %u256)
+; GLM-NEXT: Cost Model: Found an estimated cost of 96 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %a512, <64 x i8> %u512)
; GLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; XOP-LABEL: 'splatvar_rotate_i8'
@@ -1932,19 +1932,12 @@ define void @constant_rotate_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256, <8
}
define void @constant_rotate_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16 x i32> %a512) {
-; SSSE3-LABEL: 'constant_rotate_i32'
-; SSSE3-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %a32, i32 7)
-; SSSE3-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %a128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
-; SSSE3-NEXT: Cost Model: Found an estimated cost of 40 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %a256, <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 80 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %a512, <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_rotate_i32'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %a32, i32 7)
-; SSE42-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> %a128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
-; SSE42-NEXT: Cost Model: Found an estimated cost of 30 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %a256, <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 60 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %a512, <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
+; SSE-LABEL: 'constant_rotate_i32'
+; SSE-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %a32, i32 7)
+; SSE-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %a128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
+; SSE-NEXT: Cost Model: Found an estimated cost of 40 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %a256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
+; SSE-NEXT: Cost Model: Found an estimated cost of 80 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %a512, <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>)
+; SSE-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; AVX1-LABEL: 'constant_rotate_i32'
; AVX1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %a32, i32 7)
@@ -1969,16 +1962,16 @@ define void @constant_rotate_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16
;
; SLM-LABEL: 'constant_rotate_i32'
; SLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %a32, i32 7)
-; SLM-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> %a128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
-; SLM-NEXT: Cost Model: Found an estimated cost of 30 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %a256, <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 60 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %a512, <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 20 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %a128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
+; SLM-NEXT: Cost Model: Found an estimated cost of 40 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %a256, <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 80 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %a512, <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_rotate_i32'
; GLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %a32, i32 7)
-; GLM-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> %a128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
-; GLM-NEXT: Cost Model: Found an estimated cost of 30 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %a256, <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 60 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %a512, <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 20 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %a128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
+; GLM-NEXT: Cost Model: Found an estimated cost of 40 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %a256, <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 80 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %a512, <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_rotate_i32'
@@ -2005,9 +1998,9 @@ define void @constant_rotate_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <3
;
; SSE42-LABEL: 'constant_rotate_i16'
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %a16, i16 7)
-; SSE42-NEXT: Cost Model: Found an estimated cost of 17 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %a128, <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 34 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %a256, <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 68 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %a512, <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 16 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %a128, <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 32 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %a256, <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 64 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %a512, <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_rotate_i16'
@@ -2047,16 +2040,16 @@ define void @constant_rotate_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <3
;
; SLM-LABEL: 'constant_rotate_i16'
; SLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %a16, i16 7)
-; SLM-NEXT: Cost Model: Found an estimated cost of 17 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %a128, <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 34 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %a256, <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 68 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %a512, <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 16 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %a128, <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 32 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %a256, <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 64 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %a512, <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_rotate_i16'
; GLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %a16, i16 7)
-; GLM-NEXT: Cost Model: Found an estimated cost of 17 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %a128, <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 34 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %a256, <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 68 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %a512, <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 16 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %a128, <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 32 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %a256, <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 64 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %a512, <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_rotate_i16'
@@ -2083,9 +2076,9 @@ define void @constant_rotate_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x
;
; SSE42-LABEL: 'constant_rotate_i8'
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %a8, 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> %a128, <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> %a256, <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> %a512, <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 23 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %a128, <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 46 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %a256, <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 92 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %a512, <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_rotate_i8'
@@ -2125,16 +2118,16 @@ define void @constant_rotate_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x
;
; SLM-LABEL: 'constant_rotate_i8'
; SLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %a8, 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> %a128, <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> %a256, <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> %a512, <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 23 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %a128, <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 46 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %a256, <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 92 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %a512, <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_rotate_i8'
; GLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %a8, 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> %a128, <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> %a256, <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> %a512, <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 23 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %a128, <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 46 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %a256, <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 92 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %a512, <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_rotate_i8'
diff --git a/llvm/test/Analysis/CostModel/X86/rem.ll b/llvm/test/Analysis/CostModel/X86/rem.ll
index e764c18850e7..45625434bf62 100644
--- a/llvm/test/Analysis/CostModel/X86/rem.ll
+++ b/llvm/test/Analysis/CostModel/X86/rem.ll
@@ -770,17 +770,17 @@ define i32 @srem_constpow2() {
; SSE42-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %V4i64 = srem <4 x i64> undef, <i64 2, i64 4, i64 8, i64 16>
; SSE42-NEXT: Cost Model: Found an estimated cost of 112 for instruction: %V8i64 = srem <8 x i64> undef, <i64 2, i64 4, i64 8, i64 16, i64 32, i64 64, i64 128, i64 256>
; SSE42-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %I32 = srem i32 undef, 16
-; SSE42-NEXT: Cost Model: Found an estimated cost of 39 for instruction: %V4i32 = srem <4 x i32> undef, <i32 2, i32 4, i32 8, i32 16>
-; SSE42-NEXT: Cost Model: Found an estimated cost of 78 for instruction: %V8i32 = srem <8 x i32> undef, <i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256>
-; SSE42-NEXT: Cost Model: Found an estimated cost of 156 for instruction: %V16i32 = srem <16 x i32> undef, <i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256, i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256>
+; SSE42-NEXT: Cost Model: Found an estimated cost of 44 for instruction: %V4i32 = srem <4 x i32> undef, <i32 2, i32 4, i32 8, i32 16>
+; SSE42-NEXT: Cost Model: Found an estimated cost of 88 for instruction: %V8i32 = srem <8 x i32> undef, <i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256>
+; SSE42-NEXT: Cost Model: Found an estimated cost of 176 for instruction: %V16i32 = srem <16 x i32> undef, <i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256, i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256>
; SSE42-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %I16 = srem i16 undef, 16
-; SSE42-NEXT: Cost Model: Found an estimated cost of 45 for instruction: %V8i16 = srem <8 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
-; SSE42-NEXT: Cost Model: Found an estimated cost of 90 for instruction: %V16i16 = srem <16 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
-; SSE42-NEXT: Cost Model: Found an estimated cost of 180 for instruction: %V32i16 = srem <32 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
+; SSE42-NEXT: Cost Model: Found an estimated cost of 42 for instruction: %V8i16 = srem <8 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
+; SSE42-NEXT: Cost Model: Found an estimated cost of 84 for instruction: %V16i16 = srem <16 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
+; SSE42-NEXT: Cost Model: Found an estimated cost of 168 for instruction: %V32i16 = srem <32 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
; SSE42-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %I8 = srem i8 undef, 16
-; SSE42-NEXT: Cost Model: Found an estimated cost of 69 for instruction: %V16i8 = srem <16 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
-; SSE42-NEXT: Cost Model: Found an estimated cost of 140 for instruction: %V32i8 = srem <32 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
-; SSE42-NEXT: Cost Model: Found an estimated cost of 280 for instruction: %V64i8 = srem <64 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
+; SSE42-NEXT: Cost Model: Found an estimated cost of 62 for instruction: %V16i8 = srem <16 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
+; SSE42-NEXT: Cost Model: Found an estimated cost of 126 for instruction: %V32i8 = srem <32 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
+; SSE42-NEXT: Cost Model: Found an estimated cost of 252 for instruction: %V64i8 = srem <64 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
;
; AVX1-LABEL: 'srem_constpow2'
@@ -865,17 +865,17 @@ define i32 @srem_constpow2() {
; SLM-NEXT: Cost Model: Found an estimated cost of 90 for instruction: %V4i64 = srem <4 x i64> undef, <i64 2, i64 4, i64 8, i64 16>
; SLM-NEXT: Cost Model: Found an estimated cost of 180 for instruction: %V8i64 = srem <8 x i64> undef, <i64 2, i64 4, i64 8, i64 16, i64 32, i64 64, i64 128, i64 256>
; SLM-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %I32 = srem i32 undef, 16
-; SLM-NEXT: Cost Model: Found an estimated cost of 48 for instruction: %V4i32 = srem <4 x i32> undef, <i32 2, i32 4, i32 8, i32 16>
-; SLM-NEXT: Cost Model: Found an estimated cost of 96 for instruction: %V8i32 = srem <8 x i32> undef, <i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256>
-; SLM-NEXT: Cost Model: Found an estimated cost of 192 for instruction: %V16i32 = srem <16 x i32> undef, <i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256, i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256>
+; SLM-NEXT: Cost Model: Found an estimated cost of 53 for instruction: %V4i32 = srem <4 x i32> undef, <i32 2, i32 4, i32 8, i32 16>
+; SLM-NEXT: Cost Model: Found an estimated cost of 106 for instruction: %V8i32 = srem <8 x i32> undef, <i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256>
+; SLM-NEXT: Cost Model: Found an estimated cost of 212 for instruction: %V16i32 = srem <16 x i32> undef, <i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256, i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256>
; SLM-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %I16 = srem i16 undef, 16
-; SLM-NEXT: Cost Model: Found an estimated cost of 46 for instruction: %V8i16 = srem <8 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
-; SLM-NEXT: Cost Model: Found an estimated cost of 92 for instruction: %V16i16 = srem <16 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
-; SLM-NEXT: Cost Model: Found an estimated cost of 184 for instruction: %V32i16 = srem <32 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
+; SLM-NEXT: Cost Model: Found an estimated cost of 43 for instruction: %V8i16 = srem <8 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
+; SLM-NEXT: Cost Model: Found an estimated cost of 86 for instruction: %V16i16 = srem <16 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
+; SLM-NEXT: Cost Model: Found an estimated cost of 172 for instruction: %V32i16 = srem <32 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
; SLM-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %I8 = srem i8 undef, 16
-; SLM-NEXT: Cost Model: Found an estimated cost of 71 for instruction: %V16i8 = srem <16 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
-; SLM-NEXT: Cost Model: Found an estimated cost of 144 for instruction: %V32i8 = srem <32 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
-; SLM-NEXT: Cost Model: Found an estimated cost of 288 for instruction: %V64i8 = srem <64 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
+; SLM-NEXT: Cost Model: Found an estimated cost of 64 for instruction: %V16i8 = srem <16 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
+; SLM-NEXT: Cost Model: Found an estimated cost of 130 for instruction: %V32i8 = srem <32 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
+; SLM-NEXT: Cost Model: Found an estimated cost of 260 for instruction: %V64i8 = srem <64 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
; SLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
;
; GLM-LABEL: 'srem_constpow2'
@@ -884,17 +884,17 @@ define i32 @srem_constpow2() {
; GLM-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %V4i64 = srem <4 x i64> undef, <i64 2, i64 4, i64 8, i64 16>
; GLM-NEXT: Cost Model: Found an estimated cost of 112 for instruction: %V8i64 = srem <8 x i64> undef, <i64 2, i64 4, i64 8, i64 16, i64 32, i64 64, i64 128, i64 256>
; GLM-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %I32 = srem i32 undef, 16
-; GLM-NEXT: Cost Model: Found an estimated cost of 39 for instruction: %V4i32 = srem <4 x i32> undef, <i32 2, i32 4, i32 8, i32 16>
-; GLM-NEXT: Cost Model: Found an estimated cost of 78 for instruction: %V8i32 = srem <8 x i32> undef, <i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256>
-; GLM-NEXT: Cost Model: Found an estimated cost of 156 for instruction: %V16i32 = srem <16 x i32> undef, <i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256, i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256>
+; GLM-NEXT: Cost Model: Found an estimated cost of 44 for instruction: %V4i32 = srem <4 x i32> undef, <i32 2, i32 4, i32 8, i32 16>
+; GLM-NEXT: Cost Model: Found an estimated cost of 88 for instruction: %V8i32 = srem <8 x i32> undef, <i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256>
+; GLM-NEXT: Cost Model: Found an estimated cost of 176 for instruction: %V16i32 = srem <16 x i32> undef, <i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256, i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256>
; GLM-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %I16 = srem i16 undef, 16
-; GLM-NEXT: Cost Model: Found an estimated cost of 45 for instruction: %V8i16 = srem <8 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
-; GLM-NEXT: Cost Model: Found an estimated cost of 90 for instruction: %V16i16 = srem <16 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
-; GLM-NEXT: Cost Model: Found an estimated cost of 180 for instruction: %V32i16 = srem <32 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
+; GLM-NEXT: Cost Model: Found an estimated cost of 42 for instruction: %V8i16 = srem <8 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
+; GLM-NEXT: Cost Model: Found an estimated cost of 84 for instruction: %V16i16 = srem <16 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
+; GLM-NEXT: Cost Model: Found an estimated cost of 168 for instruction: %V32i16 = srem <32 x i16> undef, <i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256, i16 2, i16 4, i16 8, i16 16, i16 32, i16 64, i16 128, i16 256>
; GLM-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %I8 = srem i8 undef, 16
-; GLM-NEXT: Cost Model: Found an estimated cost of 69 for instruction: %V16i8 = srem <16 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
-; GLM-NEXT: Cost Model: Found an estimated cost of 140 for instruction: %V32i8 = srem <32 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
-; GLM-NEXT: Cost Model: Found an estimated cost of 280 for instruction: %V64i8 = srem <64 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
+; GLM-NEXT: Cost Model: Found an estimated cost of 62 for instruction: %V16i8 = srem <16 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
+; GLM-NEXT: Cost Model: Found an estimated cost of 126 for instruction: %V32i8 = srem <32 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
+; GLM-NEXT: Cost Model: Found an estimated cost of 252 for instruction: %V64i8 = srem <64 x i8> undef, <i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16, i8 2, i8 4, i8 8, i8 16>
; GLM-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
;
; BTVER2-LABEL: 'srem_constpow2'
diff --git a/llvm/test/Analysis/CostModel/X86/vshift-ashr-cost-inseltpoison.ll b/llvm/test/Analysis/CostModel/X86/vshift-ashr-cost-inseltpoison.ll
index 5594df8753d4..562d29358db9 100644
--- a/llvm/test/Analysis/CostModel/X86/vshift-ashr-cost-inseltpoison.ll
+++ b/llvm/test/Analysis/CostModel/X86/vshift-ashr-cost-inseltpoison.ll
@@ -210,7 +210,7 @@ define <8 x i16> @var_shift_v8i16(<8 x i16> %a, <8 x i16> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shift
;
; SSE42-LABEL: 'var_shift_v8i16'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %shift = ashr <8 x i16> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 13 for instruction: %shift = ashr <8 x i16> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shift
;
; AVX1-LABEL: 'var_shift_v8i16'
@@ -255,7 +255,7 @@ define <16 x i16> @var_shift_v16i16(<16 x i16> %a, <16 x i16> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
;
; SSE42-LABEL: 'var_shift_v16i16'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %shift = ashr <16 x i16> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 26 for instruction: %shift = ashr <16 x i16> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
;
; AVX1-LABEL: 'var_shift_v16i16'
@@ -300,7 +300,7 @@ define <32 x i16> @var_shift_v32i16(<32 x i16> %a, <32 x i16> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
;
; SSE42-LABEL: 'var_shift_v32i16'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %shift = ashr <32 x i16> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 52 for instruction: %shift = ashr <32 x i16> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
;
; AVX1-LABEL: 'var_shift_v32i16'
@@ -345,7 +345,7 @@ define <16 x i8> @var_shift_v16i8(<16 x i8> %a, <16 x i8> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
;
; SSE42-LABEL: 'var_shift_v16i8'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %shift = ashr <16 x i8> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 21 for instruction: %shift = ashr <16 x i8> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
;
; AVX1-LABEL: 'var_shift_v16i8'
@@ -390,7 +390,7 @@ define <32 x i8> @var_shift_v32i8(<32 x i8> %a, <32 x i8> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
;
; SSE42-LABEL: 'var_shift_v32i8'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 48 for instruction: %shift = ashr <32 x i8> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 42 for instruction: %shift = ashr <32 x i8> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
;
; AVX1-LABEL: 'var_shift_v32i8'
@@ -435,7 +435,7 @@ define <64 x i8> @var_shift_v64i8(<64 x i8> %a, <64 x i8> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
;
; SSE42-LABEL: 'var_shift_v64i8'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 96 for instruction: %shift = ashr <64 x i8> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 84 for instruction: %shift = ashr <64 x i8> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
;
; AVX1-LABEL: 'var_shift_v64i8'
@@ -947,7 +947,7 @@ define <16 x i8> @splatvar_shift_v16i8(<16 x i8> %a, i8 %b) {
; SSE42-LABEL: 'splatvar_shift_v16i8'
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i8> poison, i8 %b, i32 0
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i8> %insert, <16 x i8> poison, <16 x i32> zeroinitializer
-; SSE42-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %shift = ashr <16 x i8> %a, %splat
+; SSE42-NEXT: Cost Model: Found an estimated cost of 21 for instruction: %shift = ashr <16 x i8> %a, %splat
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
;
; AVX1-LABEL: 'splatvar_shift_v16i8'
@@ -1014,7 +1014,7 @@ define <32 x i8> @splatvar_shift_v32i8(<32 x i8> %a, i8 %b) {
; SSE42-LABEL: 'splatvar_shift_v32i8'
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <32 x i8> poison, i8 %b, i32 0
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <32 x i8> %insert, <32 x i8> poison, <32 x i32> zeroinitializer
-; SSE42-NEXT: Cost Model: Found an estimated cost of 48 for instruction: %shift = ashr <32 x i8> %a, %splat
+; SSE42-NEXT: Cost Model: Found an estimated cost of 42 for instruction: %shift = ashr <32 x i8> %a, %splat
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
;
; AVX1-LABEL: 'splatvar_shift_v32i8'
@@ -1087,7 +1087,7 @@ define <64 x i8> @splatvar_shift_v64i8(<64 x i8> %a, i8 %b) {
; SSE42-LABEL: 'splatvar_shift_v64i8'
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <64 x i8> poison, i8 %b, i32 0
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <64 x i8> %insert, <64 x i8> poison, <64 x i32> zeroinitializer
-; SSE42-NEXT: Cost Model: Found an estimated cost of 96 for instruction: %shift = ashr <64 x i8> %a, %splat
+; SSE42-NEXT: Cost Model: Found an estimated cost of 84 for instruction: %shift = ashr <64 x i8> %a, %splat
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
;
; AVX1-LABEL: 'splatvar_shift_v64i8'
@@ -1342,7 +1342,7 @@ define <8 x i16> @constant_shift_v8i16(<8 x i16> %a) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shift
;
; SSE42-LABEL: 'constant_shift_v8i16'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %shift = ashr <8 x i16> %a, <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 13 for instruction: %shift = ashr <8 x i16> %a, <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 <8 x i16> %shift
;
; AVX1-LABEL: 'constant_shift_v8i16'
@@ -1387,7 +1387,7 @@ define <16 x i16> @constant_shift_v16i16(<16 x i16> %a) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
;
; SSE42-LABEL: 'constant_shift_v16i16'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %shift = ashr <16 x i16> %a, <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 26 for instruction: %shift = ashr <16 x i16> %a, <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 <16 x i16> %shift
;
; AVX1-LABEL: 'constant_shift_v16i16'
@@ -1432,7 +1432,7 @@ define <32 x i16> @constant_shift_v32i16(<32 x i16> %a) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
;
; SSE42-LABEL: 'constant_shift_v32i16'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %shift = ashr <32 x i16> %a, <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 52 for instruction: %shift = ashr <32 x i16> %a, <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 <32 x i16> %shift
;
; AVX1-LABEL: 'constant_shift_v32i16'
@@ -1477,7 +1477,7 @@ define <16 x i8> @constant_shift_v16i8(<16 x i8> %a) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
;
; SSE42-LABEL: 'constant_shift_v16i8'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %shift = ashr <16 x i8> %a, <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 21 for instruction: %shift = ashr <16 x i8> %a, <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 <16 x i8> %shift
;
; AVX1-LABEL: 'constant_shift_v16i8'
@@ -1522,7 +1522,7 @@ define <32 x i8> @constant_shift_v32i8(<32 x i8> %a) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
;
; SSE42-LABEL: 'constant_shift_v32i8'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 48 for instruction: %shift = ashr <32 x i8> %a, <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 42 for instruction: %shift = ashr <32 x i8> %a, <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 <32 x i8> %shift
;
; AVX1-LABEL: 'constant_shift_v32i8'
@@ -1567,7 +1567,7 @@ define <64 x i8> @constant_shift_v64i8(<64 x i8> %a) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
;
; SSE42-LABEL: 'constant_shift_v64i8'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 96 for instruction: %shift = ashr <64 x i8> %a, <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 84 for instruction: %shift = ashr <64 x i8> %a, <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 <64 x i8> %shift
;
; AVX1-LABEL: 'constant_shift_v64i8'
diff --git a/llvm/test/Analysis/CostModel/X86/vshift-ashr-cost.ll b/llvm/test/Analysis/CostModel/X86/vshift-ashr-cost.ll
index 0d1d893de635..007841463100 100644
--- a/llvm/test/Analysis/CostModel/X86/vshift-ashr-cost.ll
+++ b/llvm/test/Analysis/CostModel/X86/vshift-ashr-cost.ll
@@ -210,7 +210,7 @@ define <8 x i16> @var_shift_v8i16(<8 x i16> %a, <8 x i16> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shift
;
; SSE42-LABEL: 'var_shift_v8i16'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %shift = ashr <8 x i16> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 13 for instruction: %shift = ashr <8 x i16> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shift
;
; AVX1-LABEL: 'var_shift_v8i16'
@@ -255,7 +255,7 @@ define <16 x i16> @var_shift_v16i16(<16 x i16> %a, <16 x i16> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
;
; SSE42-LABEL: 'var_shift_v16i16'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %shift = ashr <16 x i16> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 26 for instruction: %shift = ashr <16 x i16> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
;
; AVX1-LABEL: 'var_shift_v16i16'
@@ -300,7 +300,7 @@ define <32 x i16> @var_shift_v32i16(<32 x i16> %a, <32 x i16> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
;
; SSE42-LABEL: 'var_shift_v32i16'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %shift = ashr <32 x i16> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 52 for instruction: %shift = ashr <32 x i16> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
;
; AVX1-LABEL: 'var_shift_v32i16'
@@ -345,7 +345,7 @@ define <16 x i8> @var_shift_v16i8(<16 x i8> %a, <16 x i8> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
;
; SSE42-LABEL: 'var_shift_v16i8'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %shift = ashr <16 x i8> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 21 for instruction: %shift = ashr <16 x i8> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
;
; AVX1-LABEL: 'var_shift_v16i8'
@@ -390,7 +390,7 @@ define <32 x i8> @var_shift_v32i8(<32 x i8> %a, <32 x i8> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
;
; SSE42-LABEL: 'var_shift_v32i8'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 48 for instruction: %shift = ashr <32 x i8> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 42 for instruction: %shift = ashr <32 x i8> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
;
; AVX1-LABEL: 'var_shift_v32i8'
@@ -435,7 +435,7 @@ define <64 x i8> @var_shift_v64i8(<64 x i8> %a, <64 x i8> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
;
; SSE42-LABEL: 'var_shift_v64i8'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 96 for instruction: %shift = ashr <64 x i8> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 84 for instruction: %shift = ashr <64 x i8> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
;
; AVX1-LABEL: 'var_shift_v64i8'
@@ -947,7 +947,7 @@ define <16 x i8> @splatvar_shift_v16i8(<16 x i8> %a, i8 %b) {
; SSE42-LABEL: 'splatvar_shift_v16i8'
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i8> undef, i8 %b, i32 0
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i8> %insert, <16 x i8> undef, <16 x i32> zeroinitializer
-; SSE42-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %shift = ashr <16 x i8> %a, %splat
+; SSE42-NEXT: Cost Model: Found an estimated cost of 21 for instruction: %shift = ashr <16 x i8> %a, %splat
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
;
; AVX1-LABEL: 'splatvar_shift_v16i8'
@@ -1014,7 +1014,7 @@ define <32 x i8> @splatvar_shift_v32i8(<32 x i8> %a, i8 %b) {
; SSE42-LABEL: 'splatvar_shift_v32i8'
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <32 x i8> undef, i8 %b, i32 0
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <32 x i8> %insert, <32 x i8> undef, <32 x i32> zeroinitializer
-; SSE42-NEXT: Cost Model: Found an estimated cost of 48 for instruction: %shift = ashr <32 x i8> %a, %splat
+; SSE42-NEXT: Cost Model: Found an estimated cost of 42 for instruction: %shift = ashr <32 x i8> %a, %splat
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
;
; AVX1-LABEL: 'splatvar_shift_v32i8'
@@ -1087,7 +1087,7 @@ define <64 x i8> @splatvar_shift_v64i8(<64 x i8> %a, i8 %b) {
; SSE42-LABEL: 'splatvar_shift_v64i8'
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <64 x i8> undef, i8 %b, i32 0
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <64 x i8> %insert, <64 x i8> undef, <64 x i32> zeroinitializer
-; SSE42-NEXT: Cost Model: Found an estimated cost of 96 for instruction: %shift = ashr <64 x i8> %a, %splat
+; SSE42-NEXT: Cost Model: Found an estimated cost of 84 for instruction: %shift = ashr <64 x i8> %a, %splat
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
;
; AVX1-LABEL: 'splatvar_shift_v64i8'
@@ -1342,7 +1342,7 @@ define <8 x i16> @constant_shift_v8i16(<8 x i16> %a) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shift
;
; SSE42-LABEL: 'constant_shift_v8i16'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %shift = ashr <8 x i16> %a, <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 13 for instruction: %shift = ashr <8 x i16> %a, <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 <8 x i16> %shift
;
; AVX1-LABEL: 'constant_shift_v8i16'
@@ -1387,7 +1387,7 @@ define <16 x i16> @constant_shift_v16i16(<16 x i16> %a) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
;
; SSE42-LABEL: 'constant_shift_v16i16'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %shift = ashr <16 x i16> %a, <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 26 for instruction: %shift = ashr <16 x i16> %a, <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 <16 x i16> %shift
;
; AVX1-LABEL: 'constant_shift_v16i16'
@@ -1432,7 +1432,7 @@ define <32 x i16> @constant_shift_v32i16(<32 x i16> %a) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
;
; SSE42-LABEL: 'constant_shift_v32i16'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %shift = ashr <32 x i16> %a, <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 52 for instruction: %shift = ashr <32 x i16> %a, <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 <32 x i16> %shift
;
; AVX1-LABEL: 'constant_shift_v32i16'
@@ -1477,7 +1477,7 @@ define <16 x i8> @constant_shift_v16i8(<16 x i8> %a) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
;
; SSE42-LABEL: 'constant_shift_v16i8'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %shift = ashr <16 x i8> %a, <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 21 for instruction: %shift = ashr <16 x i8> %a, <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 <16 x i8> %shift
;
; AVX1-LABEL: 'constant_shift_v16i8'
@@ -1522,7 +1522,7 @@ define <32 x i8> @constant_shift_v32i8(<32 x i8> %a) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
;
; SSE42-LABEL: 'constant_shift_v32i8'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 48 for instruction: %shift = ashr <32 x i8> %a, <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 42 for instruction: %shift = ashr <32 x i8> %a, <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 <32 x i8> %shift
;
; AVX1-LABEL: 'constant_shift_v32i8'
@@ -1567,7 +1567,7 @@ define <64 x i8> @constant_shift_v64i8(<64 x i8> %a) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
;
; SSE42-LABEL: 'constant_shift_v64i8'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 96 for instruction: %shift = ashr <64 x i8> %a, <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 84 for instruction: %shift = ashr <64 x i8> %a, <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 <64 x i8> %shift
;
; AVX1-LABEL: 'constant_shift_v64i8'
diff --git a/llvm/test/Analysis/CostModel/X86/vshift-lshr-cost-inseltpoison.ll b/llvm/test/Analysis/CostModel/X86/vshift-lshr-cost-inseltpoison.ll
index 37d3b8137669..7b9fb1a49124 100644
--- a/llvm/test/Analysis/CostModel/X86/vshift-lshr-cost-inseltpoison.ll
+++ b/llvm/test/Analysis/CostModel/X86/vshift-lshr-cost-inseltpoison.ll
@@ -127,7 +127,7 @@ define <4 x i32> @var_shift_v4i32(<4 x i32> %a, <4 x i32> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
;
; SSE42-LABEL: 'var_shift_v4i32'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %shift = lshr <4 x i32> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %shift = lshr <4 x i32> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
;
; AVX1-LABEL: 'var_shift_v4i32'
@@ -160,7 +160,7 @@ define <8 x i32> @var_shift_v8i32(<8 x i32> %a, <8 x i32> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
;
; SSE42-LABEL: 'var_shift_v8i32'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %shift = lshr <8 x i32> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %shift = lshr <8 x i32> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
;
; AVX1-LABEL: 'var_shift_v8i32'
@@ -197,7 +197,7 @@ define <16 x i32> @var_shift_v16i32(<16 x i32> %a, <16 x i32> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
;
; SSE42-LABEL: 'var_shift_v16i32'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 44 for instruction: %shift = lshr <16 x i32> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 64 for instruction: %shift = lshr <16 x i32> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
;
; AVX1-LABEL: 'var_shift_v16i32'
@@ -234,7 +234,7 @@ define <8 x i16> @var_shift_v8i16(<8 x i16> %a, <8 x i16> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shift
;
; SSE42-LABEL: 'var_shift_v8i16'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %shift = lshr <8 x i16> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 13 for instruction: %shift = lshr <8 x i16> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shift
;
; AVX1-LABEL: 'var_shift_v8i16'
@@ -279,7 +279,7 @@ define <16 x i16> @var_shift_v16i16(<16 x i16> %a, <16 x i16> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
;
; SSE42-LABEL: 'var_shift_v16i16'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %shift = lshr <16 x i16> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 26 for instruction: %shift = lshr <16 x i16> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
;
; AVX1-LABEL: 'var_shift_v16i16'
@@ -324,7 +324,7 @@ define <32 x i16> @var_shift_v32i16(<32 x i16> %a, <32 x i16> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
;
; SSE42-LABEL: 'var_shift_v32i16'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %shift = lshr <32 x i16> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 52 for instruction: %shift = lshr <32 x i16> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
;
; AVX1-LABEL: 'var_shift_v32i16'
@@ -369,7 +369,7 @@ define <16 x i8> @var_shift_v16i8(<16 x i8> %a, <16 x i8> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
;
; SSE42-LABEL: 'var_shift_v16i8'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %shift = lshr <16 x i8> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %shift = lshr <16 x i8> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
;
; AVX1-LABEL: 'var_shift_v16i8'
@@ -414,7 +414,7 @@ define <32 x i8> @var_shift_v32i8(<32 x i8> %a, <32 x i8> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
;
; SSE42-LABEL: 'var_shift_v32i8'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %shift = lshr <32 x i8> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %shift = lshr <32 x i8> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
;
; AVX1-LABEL: 'var_shift_v32i8'
@@ -459,7 +459,7 @@ define <64 x i8> @var_shift_v64i8(<64 x i8> %a, <64 x i8> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
;
; SSE42-LABEL: 'var_shift_v64i8'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 48 for instruction: %shift = lshr <64 x i8> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 44 for instruction: %shift = lshr <64 x i8> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
;
; AVX1-LABEL: 'var_shift_v64i8'
@@ -971,7 +971,7 @@ define <16 x i8> @splatvar_shift_v16i8(<16 x i8> %a, i8 %b) {
; SSE42-LABEL: 'splatvar_shift_v16i8'
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i8> poison, i8 %b, i32 0
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i8> %insert, <16 x i8> poison, <16 x i32> zeroinitializer
-; SSE42-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %shift = lshr <16 x i8> %a, %splat
+; SSE42-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %shift = lshr <16 x i8> %a, %splat
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
;
; AVX1-LABEL: 'splatvar_shift_v16i8'
@@ -1038,7 +1038,7 @@ define <32 x i8> @splatvar_shift_v32i8(<32 x i8> %a, i8 %b) {
; SSE42-LABEL: 'splatvar_shift_v32i8'
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <32 x i8> poison, i8 %b, i32 0
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <32 x i8> %insert, <32 x i8> poison, <32 x i32> zeroinitializer
-; SSE42-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %shift = lshr <32 x i8> %a, %splat
+; SSE42-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %shift = lshr <32 x i8> %a, %splat
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
;
; AVX1-LABEL: 'splatvar_shift_v32i8'
@@ -1111,7 +1111,7 @@ define <64 x i8> @splatvar_shift_v64i8(<64 x i8> %a, i8 %b) {
; SSE42-LABEL: 'splatvar_shift_v64i8'
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <64 x i8> poison, i8 %b, i32 0
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <64 x i8> %insert, <64 x i8> poison, <64 x i32> zeroinitializer
-; SSE42-NEXT: Cost Model: Found an estimated cost of 48 for instruction: %shift = lshr <64 x i8> %a, %splat
+; SSE42-NEXT: Cost Model: Found an estimated cost of 44 for instruction: %shift = lshr <64 x i8> %a, %splat
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
;
; AVX1-LABEL: 'splatvar_shift_v64i8'
@@ -1279,7 +1279,7 @@ define <4 x i32> @constant_shift_v4i32(<4 x i32> %a) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
;
; SSE42-LABEL: 'constant_shift_v4i32'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %shift = lshr <4 x i32> %a, <i32 4, i32 5, i32 6, i32 7>
+; SSE42-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %shift = lshr <4 x i32> %a, <i32 4, i32 5, i32 6, i32 7>
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
;
; AVX1-LABEL: 'constant_shift_v4i32'
@@ -1312,7 +1312,7 @@ define <8 x i32> @constant_shift_v8i32(<8 x i32> %a) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
;
; SSE42-LABEL: 'constant_shift_v8i32'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %shift = lshr <8 x i32> %a, <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 32 for instruction: %shift = lshr <8 x i32> %a, <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 <8 x i32> %shift
;
; AVX1-LABEL: 'constant_shift_v8i32'
@@ -1349,7 +1349,7 @@ define <16 x i32> @constant_shift_v16i32(<16 x i32> %a) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
;
; SSE42-LABEL: 'constant_shift_v16i32'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 44 for instruction: %shift = lshr <16 x i32> %a, <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 64 for instruction: %shift = lshr <16 x i32> %a, <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 <16 x i32> %shift
;
; AVX1-LABEL: 'constant_shift_v16i32'
@@ -1386,7 +1386,7 @@ define <8 x i16> @constant_shift_v8i16(<8 x i16> %a) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shift
;
; SSE42-LABEL: 'constant_shift_v8i16'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %shift = lshr <8 x i16> %a, <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 13 for instruction: %shift = lshr <8 x i16> %a, <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 <8 x i16> %shift
;
; AVX1-LABEL: 'constant_shift_v8i16'
@@ -1431,7 +1431,7 @@ define <16 x i16> @constant_shift_v16i16(<16 x i16> %a) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
;
; SSE42-LABEL: 'constant_shift_v16i16'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %shift = lshr <16 x i16> %a, <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 26 for instruction: %shift = lshr <16 x i16> %a, <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 <16 x i16> %shift
;
; AVX1-LABEL: 'constant_shift_v16i16'
@@ -1476,7 +1476,7 @@ define <32 x i16> @constant_shift_v32i16(<32 x i16> %a) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
;
; SSE42-LABEL: 'constant_shift_v32i16'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %shift = lshr <32 x i16> %a, <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 52 for instruction: %shift = lshr <32 x i16> %a, <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 <32 x i16> %shift
;
; AVX1-LABEL: 'constant_shift_v32i16'
@@ -1521,7 +1521,7 @@ define <16 x i8> @constant_shift_v16i8(<16 x i8> %a) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
;
; SSE42-LABEL: 'constant_shift_v16i8'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %shift = lshr <16 x i8> %a, <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 11 for instruction: %shift = lshr <16 x i8> %a, <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 <16 x i8> %shift
;
; AVX1-LABEL: 'constant_shift_v16i8'
@@ -1566,7 +1566,7 @@ define <32 x i8> @constant_shift_v32i8(<32 x i8> %a) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
;
; SSE42-LABEL: 'constant_shift_v32i8'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %shift = lshr <32 x i8> %a, <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 22 for instruction: %shift = lshr <32 x i8> %a, <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 <32 x i8> %shift
;
; AVX1-LABEL: 'constant_shift_v32i8'
@@ -1611,7 +1611,7 @@ define <64 x i8> @constant_shift_v64i8(<64 x i8> %a) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
;
; SSE42-LABEL: 'constant_shift_v64i8'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 48 for instruction: %shift = lshr <64 x i8> %a, <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 44 for instruction: %shift = lshr <64 x i8> %a, <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 <64 x i8> %shift
;
; AVX1-LABEL: 'constant_shift_v64i8'
diff --git a/llvm/test/Analysis/CostModel/X86/vshift-lshr-cost.ll b/llvm/test/Analysis/CostModel/X86/vshift-lshr-cost.ll
index ad4db43651f4..c5f326a53cb1 100644
--- a/llvm/test/Analysis/CostModel/X86/vshift-lshr-cost.ll
+++ b/llvm/test/Analysis/CostModel/X86/vshift-lshr-cost.ll
@@ -127,7 +127,7 @@ define <4 x i32> @var_shift_v4i32(<4 x i32> %a, <4 x i32> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
;
; SSE42-LABEL: 'var_shift_v4i32'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %shift = lshr <4 x i32> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %shift = lshr <4 x i32> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
;
; AVX1-LABEL: 'var_shift_v4i32'
@@ -160,7 +160,7 @@ define <8 x i32> @var_shift_v8i32(<8 x i32> %a, <8 x i32> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
;
; SSE42-LABEL: 'var_shift_v8i32'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %shift = lshr <8 x i32> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %shift = lshr <8 x i32> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
;
; AVX1-LABEL: 'var_shift_v8i32'
@@ -197,7 +197,7 @@ define <16 x i32> @var_shift_v16i32(<16 x i32> %a, <16 x i32> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
;
; SSE42-LABEL: 'var_shift_v16i32'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 44 for instruction: %shift = lshr <16 x i32> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 64 for instruction: %shift = lshr <16 x i32> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
;
; AVX1-LABEL: 'var_shift_v16i32'
@@ -234,7 +234,7 @@ define <8 x i16> @var_shift_v8i16(<8 x i16> %a, <8 x i16> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shift
;
; SSE42-LABEL: 'var_shift_v8i16'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %shift = lshr <8 x i16> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 13 for instruction: %shift = lshr <8 x i16> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shift
;
; AVX1-LABEL: 'var_shift_v8i16'
@@ -279,7 +279,7 @@ define <16 x i16> @var_shift_v16i16(<16 x i16> %a, <16 x i16> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
;
; SSE42-LABEL: 'var_shift_v16i16'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %shift = lshr <16 x i16> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 26 for instruction: %shift = lshr <16 x i16> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
;
; AVX1-LABEL: 'var_shift_v16i16'
@@ -324,7 +324,7 @@ define <32 x i16> @var_shift_v32i16(<32 x i16> %a, <32 x i16> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
;
; SSE42-LABEL: 'var_shift_v32i16'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %shift = lshr <32 x i16> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 52 for instruction: %shift = lshr <32 x i16> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
;
; AVX1-LABEL: 'var_shift_v32i16'
@@ -369,7 +369,7 @@ define <16 x i8> @var_shift_v16i8(<16 x i8> %a, <16 x i8> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
;
; SSE42-LABEL: 'var_shift_v16i8'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %shift = lshr <16 x i8> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %shift = lshr <16 x i8> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
;
; AVX1-LABEL: 'var_shift_v16i8'
@@ -414,7 +414,7 @@ define <32 x i8> @var_shift_v32i8(<32 x i8> %a, <32 x i8> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
;
; SSE42-LABEL: 'var_shift_v32i8'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %shift = lshr <32 x i8> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %shift = lshr <32 x i8> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
;
; AVX1-LABEL: 'var_shift_v32i8'
@@ -459,7 +459,7 @@ define <64 x i8> @var_shift_v64i8(<64 x i8> %a, <64 x i8> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
;
; SSE42-LABEL: 'var_shift_v64i8'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 48 for instruction: %shift = lshr <64 x i8> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 44 for instruction: %shift = lshr <64 x i8> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
;
; AVX1-LABEL: 'var_shift_v64i8'
@@ -971,7 +971,7 @@ define <16 x i8> @splatvar_shift_v16i8(<16 x i8> %a, i8 %b) {
; SSE42-LABEL: 'splatvar_shift_v16i8'
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i8> undef, i8 %b, i32 0
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i8> %insert, <16 x i8> undef, <16 x i32> zeroinitializer
-; SSE42-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %shift = lshr <16 x i8> %a, %splat
+; SSE42-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %shift = lshr <16 x i8> %a, %splat
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
;
; AVX1-LABEL: 'splatvar_shift_v16i8'
@@ -1038,7 +1038,7 @@ define <32 x i8> @splatvar_shift_v32i8(<32 x i8> %a, i8 %b) {
; SSE42-LABEL: 'splatvar_shift_v32i8'
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <32 x i8> undef, i8 %b, i32 0
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <32 x i8> %insert, <32 x i8> undef, <32 x i32> zeroinitializer
-; SSE42-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %shift = lshr <32 x i8> %a, %splat
+; SSE42-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %shift = lshr <32 x i8> %a, %splat
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
;
; AVX1-LABEL: 'splatvar_shift_v32i8'
@@ -1111,7 +1111,7 @@ define <64 x i8> @splatvar_shift_v64i8(<64 x i8> %a, i8 %b) {
; SSE42-LABEL: 'splatvar_shift_v64i8'
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <64 x i8> undef, i8 %b, i32 0
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <64 x i8> %insert, <64 x i8> undef, <64 x i32> zeroinitializer
-; SSE42-NEXT: Cost Model: Found an estimated cost of 48 for instruction: %shift = lshr <64 x i8> %a, %splat
+; SSE42-NEXT: Cost Model: Found an estimated cost of 44 for instruction: %shift = lshr <64 x i8> %a, %splat
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
;
; AVX1-LABEL: 'splatvar_shift_v64i8'
@@ -1279,7 +1279,7 @@ define <4 x i32> @constant_shift_v4i32(<4 x i32> %a) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
;
; SSE42-LABEL: 'constant_shift_v4i32'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %shift = lshr <4 x i32> %a, <i32 4, i32 5, i32 6, i32 7>
+; SSE42-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %shift = lshr <4 x i32> %a, <i32 4, i32 5, i32 6, i32 7>
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %shift
;
; AVX1-LABEL: 'constant_shift_v4i32'
@@ -1312,7 +1312,7 @@ define <8 x i32> @constant_shift_v8i32(<8 x i32> %a) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %shift
;
; SSE42-LABEL: 'constant_shift_v8i32'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %shift = lshr <8 x i32> %a, <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 32 for instruction: %shift = lshr <8 x i32> %a, <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 <8 x i32> %shift
;
; AVX1-LABEL: 'constant_shift_v8i32'
@@ -1349,7 +1349,7 @@ define <16 x i32> @constant_shift_v16i32(<16 x i32> %a) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i32> %shift
;
; SSE42-LABEL: 'constant_shift_v16i32'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 44 for instruction: %shift = lshr <16 x i32> %a, <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 64 for instruction: %shift = lshr <16 x i32> %a, <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 <16 x i32> %shift
;
; AVX1-LABEL: 'constant_shift_v16i32'
@@ -1386,7 +1386,7 @@ define <8 x i16> @constant_shift_v8i16(<8 x i16> %a) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shift
;
; SSE42-LABEL: 'constant_shift_v8i16'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %shift = lshr <8 x i16> %a, <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 13 for instruction: %shift = lshr <8 x i16> %a, <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 <8 x i16> %shift
;
; AVX1-LABEL: 'constant_shift_v8i16'
@@ -1431,7 +1431,7 @@ define <16 x i16> @constant_shift_v16i16(<16 x i16> %a) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
;
; SSE42-LABEL: 'constant_shift_v16i16'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %shift = lshr <16 x i16> %a, <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 26 for instruction: %shift = lshr <16 x i16> %a, <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 <16 x i16> %shift
;
; AVX1-LABEL: 'constant_shift_v16i16'
@@ -1476,7 +1476,7 @@ define <32 x i16> @constant_shift_v32i16(<32 x i16> %a) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
;
; SSE42-LABEL: 'constant_shift_v32i16'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %shift = lshr <32 x i16> %a, <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 52 for instruction: %shift = lshr <32 x i16> %a, <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 <32 x i16> %shift
;
; AVX1-LABEL: 'constant_shift_v32i16'
@@ -1521,7 +1521,7 @@ define <16 x i8> @constant_shift_v16i8(<16 x i8> %a) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
;
; SSE42-LABEL: 'constant_shift_v16i8'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %shift = lshr <16 x i8> %a, <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 11 for instruction: %shift = lshr <16 x i8> %a, <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 <16 x i8> %shift
;
; AVX1-LABEL: 'constant_shift_v16i8'
@@ -1566,7 +1566,7 @@ define <32 x i8> @constant_shift_v32i8(<32 x i8> %a) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
;
; SSE42-LABEL: 'constant_shift_v32i8'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %shift = lshr <32 x i8> %a, <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 22 for instruction: %shift = lshr <32 x i8> %a, <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 <32 x i8> %shift
;
; AVX1-LABEL: 'constant_shift_v32i8'
@@ -1611,7 +1611,7 @@ define <64 x i8> @constant_shift_v64i8(<64 x i8> %a) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
;
; SSE42-LABEL: 'constant_shift_v64i8'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 48 for instruction: %shift = lshr <64 x i8> %a, <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 44 for instruction: %shift = lshr <64 x i8> %a, <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 <64 x i8> %shift
;
; AVX1-LABEL: 'constant_shift_v64i8'
diff --git a/llvm/test/Analysis/CostModel/X86/vshift-shl-cost-inseltpoison.ll b/llvm/test/Analysis/CostModel/X86/vshift-shl-cost-inseltpoison.ll
index 1b751c66a040..a0b27e550b30 100644
--- a/llvm/test/Analysis/CostModel/X86/vshift-shl-cost-inseltpoison.ll
+++ b/llvm/test/Analysis/CostModel/X86/vshift-shl-cost-inseltpoison.ll
@@ -230,7 +230,7 @@ define <8 x i16> @var_shift_v8i16(<8 x i16> %a, <8 x i16> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shift
;
; SSE42-LABEL: 'var_shift_v8i16'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %shift = shl <8 x i16> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %shift = shl <8 x i16> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shift
;
; AVX1-LABEL: 'var_shift_v8i16'
@@ -275,7 +275,7 @@ define <16 x i16> @var_shift_v16i16(<16 x i16> %a, <16 x i16> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
;
; SSE42-LABEL: 'var_shift_v16i16'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %shift = shl <16 x i16> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %shift = shl <16 x i16> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
;
; AVX1-LABEL: 'var_shift_v16i16'
@@ -320,7 +320,7 @@ define <32 x i16> @var_shift_v32i16(<32 x i16> %a, <32 x i16> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
;
; SSE42-LABEL: 'var_shift_v32i16'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %shift = shl <32 x i16> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 44 for instruction: %shift = shl <32 x i16> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
;
; AVX1-LABEL: 'var_shift_v32i16'
@@ -365,7 +365,7 @@ define <16 x i8> @var_shift_v16i8(<16 x i8> %a, <16 x i8> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
;
; SSE42-LABEL: 'var_shift_v16i8'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %shift = shl <16 x i8> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %shift = shl <16 x i8> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
;
; AVX1-LABEL: 'var_shift_v16i8'
@@ -410,7 +410,7 @@ define <32 x i8> @var_shift_v32i8(<32 x i8> %a, <32 x i8> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
;
; SSE42-LABEL: 'var_shift_v32i8'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %shift = shl <32 x i8> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %shift = shl <32 x i8> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
;
; AVX1-LABEL: 'var_shift_v32i8'
@@ -455,7 +455,7 @@ define <64 x i8> @var_shift_v64i8(<64 x i8> %a, <64 x i8> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
;
; SSE42-LABEL: 'var_shift_v64i8'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 44 for instruction: %shift = shl <64 x i8> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 40 for instruction: %shift = shl <64 x i8> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
;
; AVX1-LABEL: 'var_shift_v64i8'
@@ -967,7 +967,7 @@ define <16 x i8> @splatvar_shift_v16i8(<16 x i8> %a, i8 %b) {
; SSE42-LABEL: 'splatvar_shift_v16i8'
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i8> poison, i8 %b, i32 0
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i8> %insert, <16 x i8> poison, <16 x i32> zeroinitializer
-; SSE42-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %shift = shl <16 x i8> %a, %splat
+; SSE42-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %shift = shl <16 x i8> %a, %splat
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
;
; AVX1-LABEL: 'splatvar_shift_v16i8'
@@ -1034,7 +1034,7 @@ define <32 x i8> @splatvar_shift_v32i8(<32 x i8> %a, i8 %b) {
; SSE42-LABEL: 'splatvar_shift_v32i8'
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <32 x i8> poison, i8 %b, i32 0
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <32 x i8> %insert, <32 x i8> poison, <32 x i32> zeroinitializer
-; SSE42-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %shift = shl <32 x i8> %a, %splat
+; SSE42-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %shift = shl <32 x i8> %a, %splat
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
;
; AVX1-LABEL: 'splatvar_shift_v32i8'
@@ -1107,7 +1107,7 @@ define <64 x i8> @splatvar_shift_v64i8(<64 x i8> %a, i8 %b) {
; SSE42-LABEL: 'splatvar_shift_v64i8'
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <64 x i8> poison, i8 %b, i32 0
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <64 x i8> %insert, <64 x i8> poison, <64 x i32> zeroinitializer
-; SSE42-NEXT: Cost Model: Found an estimated cost of 44 for instruction: %shift = shl <64 x i8> %a, %splat
+; SSE42-NEXT: Cost Model: Found an estimated cost of 40 for instruction: %shift = shl <64 x i8> %a, %splat
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
;
; AVX1-LABEL: 'splatvar_shift_v64i8'
@@ -1461,7 +1461,7 @@ define <16 x i8> @constant_shift_v16i8(<16 x i8> %a) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
;
; SSE42-LABEL: 'constant_shift_v16i8'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %shift = shl <16 x i8> %a, <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 10 for instruction: %shift = shl <16 x i8> %a, <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 <16 x i8> %shift
;
; AVX1-LABEL: 'constant_shift_v16i8'
@@ -1506,7 +1506,7 @@ define <32 x i8> @constant_shift_v32i8(<32 x i8> %a) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
;
; SSE42-LABEL: 'constant_shift_v32i8'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %shift = shl <32 x i8> %a, <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 20 for instruction: %shift = shl <32 x i8> %a, <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 <32 x i8> %shift
;
; AVX1-LABEL: 'constant_shift_v32i8'
@@ -1551,7 +1551,7 @@ define <64 x i8> @constant_shift_v64i8(<64 x i8> %a) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
;
; SSE42-LABEL: 'constant_shift_v64i8'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 44 for instruction: %shift = shl <64 x i8> %a, <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 40 for instruction: %shift = shl <64 x i8> %a, <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 <64 x i8> %shift
;
; AVX1-LABEL: 'constant_shift_v64i8'
diff --git a/llvm/test/Analysis/CostModel/X86/vshift-shl-cost.ll b/llvm/test/Analysis/CostModel/X86/vshift-shl-cost.ll
index 70f10cb22420..cecd5c446fbd 100644
--- a/llvm/test/Analysis/CostModel/X86/vshift-shl-cost.ll
+++ b/llvm/test/Analysis/CostModel/X86/vshift-shl-cost.ll
@@ -230,7 +230,7 @@ define <8 x i16> @var_shift_v8i16(<8 x i16> %a, <8 x i16> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shift
;
; SSE42-LABEL: 'var_shift_v8i16'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %shift = shl <8 x i16> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %shift = shl <8 x i16> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i16> %shift
;
; AVX1-LABEL: 'var_shift_v8i16'
@@ -275,7 +275,7 @@ define <16 x i16> @var_shift_v16i16(<16 x i16> %a, <16 x i16> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
;
; SSE42-LABEL: 'var_shift_v16i16'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %shift = shl <16 x i16> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %shift = shl <16 x i16> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %shift
;
; AVX1-LABEL: 'var_shift_v16i16'
@@ -320,7 +320,7 @@ define <32 x i16> @var_shift_v32i16(<32 x i16> %a, <32 x i16> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
;
; SSE42-LABEL: 'var_shift_v32i16'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 56 for instruction: %shift = shl <32 x i16> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 44 for instruction: %shift = shl <32 x i16> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i16> %shift
;
; AVX1-LABEL: 'var_shift_v32i16'
@@ -365,7 +365,7 @@ define <16 x i8> @var_shift_v16i8(<16 x i8> %a, <16 x i8> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
;
; SSE42-LABEL: 'var_shift_v16i8'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %shift = shl <16 x i8> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %shift = shl <16 x i8> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
;
; AVX1-LABEL: 'var_shift_v16i8'
@@ -410,7 +410,7 @@ define <32 x i8> @var_shift_v32i8(<32 x i8> %a, <32 x i8> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
;
; SSE42-LABEL: 'var_shift_v32i8'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %shift = shl <32 x i8> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %shift = shl <32 x i8> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
;
; AVX1-LABEL: 'var_shift_v32i8'
@@ -455,7 +455,7 @@ define <64 x i8> @var_shift_v64i8(<64 x i8> %a, <64 x i8> %b) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
;
; SSE42-LABEL: 'var_shift_v64i8'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 44 for instruction: %shift = shl <64 x i8> %a, %b
+; SSE42-NEXT: Cost Model: Found an estimated cost of 40 for instruction: %shift = shl <64 x i8> %a, %b
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
;
; AVX1-LABEL: 'var_shift_v64i8'
@@ -967,7 +967,7 @@ define <16 x i8> @splatvar_shift_v16i8(<16 x i8> %a, i8 %b) {
; SSE42-LABEL: 'splatvar_shift_v16i8'
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <16 x i8> undef, i8 %b, i32 0
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <16 x i8> %insert, <16 x i8> undef, <16 x i32> zeroinitializer
-; SSE42-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %shift = shl <16 x i8> %a, %splat
+; SSE42-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %shift = shl <16 x i8> %a, %splat
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
;
; AVX1-LABEL: 'splatvar_shift_v16i8'
@@ -1034,7 +1034,7 @@ define <32 x i8> @splatvar_shift_v32i8(<32 x i8> %a, i8 %b) {
; SSE42-LABEL: 'splatvar_shift_v32i8'
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <32 x i8> undef, i8 %b, i32 0
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <32 x i8> %insert, <32 x i8> undef, <32 x i32> zeroinitializer
-; SSE42-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %shift = shl <32 x i8> %a, %splat
+; SSE42-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %shift = shl <32 x i8> %a, %splat
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
;
; AVX1-LABEL: 'splatvar_shift_v32i8'
@@ -1107,7 +1107,7 @@ define <64 x i8> @splatvar_shift_v64i8(<64 x i8> %a, i8 %b) {
; SSE42-LABEL: 'splatvar_shift_v64i8'
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %insert = insertelement <64 x i8> undef, i8 %b, i32 0
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %splat = shufflevector <64 x i8> %insert, <64 x i8> undef, <64 x i32> zeroinitializer
-; SSE42-NEXT: Cost Model: Found an estimated cost of 44 for instruction: %shift = shl <64 x i8> %a, %splat
+; SSE42-NEXT: Cost Model: Found an estimated cost of 40 for instruction: %shift = shl <64 x i8> %a, %splat
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
;
; AVX1-LABEL: 'splatvar_shift_v64i8'
@@ -1461,7 +1461,7 @@ define <16 x i8> @constant_shift_v16i8(<16 x i8> %a) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %shift
;
; SSE42-LABEL: 'constant_shift_v16i8'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %shift = shl <16 x i8> %a, <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 10 for instruction: %shift = shl <16 x i8> %a, <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 <16 x i8> %shift
;
; AVX1-LABEL: 'constant_shift_v16i8'
@@ -1506,7 +1506,7 @@ define <32 x i8> @constant_shift_v32i8(<32 x i8> %a) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %shift
;
; SSE42-LABEL: 'constant_shift_v32i8'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %shift = shl <32 x i8> %a, <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 20 for instruction: %shift = shl <32 x i8> %a, <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 <32 x i8> %shift
;
; AVX1-LABEL: 'constant_shift_v32i8'
@@ -1551,7 +1551,7 @@ define <64 x i8> @constant_shift_v64i8(<64 x i8> %a) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <64 x i8> %shift
;
; SSE42-LABEL: 'constant_shift_v64i8'
-; SSE42-NEXT: Cost Model: Found an estimated cost of 44 for instruction: %shift = shl <64 x i8> %a, <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 40 for instruction: %shift = shl <64 x i8> %a, <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 <64 x i8> %shift
;
; AVX1-LABEL: 'constant_shift_v64i8'
More information about the llvm-commits
mailing list