[llvm] 243e588 - [CostModel][X86] Improve accuracy of vXi64 MUL costs on AVX2/AVX512 targets
Simon Pilgrim via llvm-commits
llvm-commits at lists.llvm.org
Mon May 24 01:48:41 PDT 2021
Author: Simon Pilgrim
Date: 2021-05-24T09:48:32+01:00
New Revision: 243e58868176102484c3ff1a338342633ede7361
URL: https://github.com/llvm/llvm-project/commit/243e58868176102484c3ff1a338342633ede7361
DIFF: https://github.com/llvm/llvm-project/commit/243e58868176102484c3ff1a338342633ede7361.diff
LOG: [CostModel][X86] Improve accuracy of vXi64 MUL costs on AVX2/AVX512 targets
By llvm-mca analysis, Haswell/Broadwell has the worst v4i64 recip-throughput cost of the AVX2 targets at 6 (vs the currently used cost of 8). Similarly SkylakeServer (our only AVX512 target model) implements PMULLQ with an average cost of 1.5 (rounded up to 2.0), and the PMULUDQ-sequence (without AVX512DQ) as a cost of 6.
Added:
Modified:
llvm/lib/Target/X86/X86TargetTransformInfo.cpp
llvm/test/Analysis/CostModel/X86/arith-fix.ll
llvm/test/Analysis/CostModel/X86/arith-overflow.ll
llvm/test/Analysis/CostModel/X86/arith.ll
llvm/test/Analysis/CostModel/X86/reduce-mul.ll
llvm/test/Analysis/CostModel/X86/rem.ll
Removed:
################################################################################
diff --git a/llvm/lib/Target/X86/X86TargetTransformInfo.cpp b/llvm/lib/Target/X86/X86TargetTransformInfo.cpp
index bc9b1daca7c21..b3a7bd5c15b90 100644
--- a/llvm/lib/Target/X86/X86TargetTransformInfo.cpp
+++ b/llvm/lib/Target/X86/X86TargetTransformInfo.cpp
@@ -571,9 +571,9 @@ InstructionCost X86TTIImpl::getArithmeticInstrCost(
}
static const CostTblEntry AVX512DQCostTable[] = {
- { ISD::MUL, MVT::v2i64, 1 },
- { ISD::MUL, MVT::v4i64, 1 },
- { ISD::MUL, MVT::v8i64, 1 }
+ { ISD::MUL, MVT::v2i64, 2 }, // pmullq
+ { ISD::MUL, MVT::v4i64, 2 }, // pmullq
+ { ISD::MUL, MVT::v8i64, 2 } // pmullq
};
// Look for AVX512DQ lowering tricks for custom cases.
@@ -612,7 +612,7 @@ InstructionCost X86TTIImpl::getArithmeticInstrCost(
{ ISD::MUL, MVT::v16i32, 1 }, // pmulld (Skylake from agner.org)
{ ISD::MUL, MVT::v8i32, 1 }, // pmulld (Skylake from agner.org)
{ ISD::MUL, MVT::v4i32, 1 }, // pmulld (Skylake from agner.org)
- { ISD::MUL, MVT::v8i64, 8 }, // 3*pmuludq/3*shift/2*add
+ { ISD::MUL, MVT::v8i64, 6 }, // 3*pmuludq/3*shift/2*add
{ ISD::FNEG, MVT::v8f64, 1 }, // Skylake from http://www.agner.org/
{ ISD::FADD, MVT::v8f64, 1 }, // Skylake from http://www.agner.org/
@@ -783,7 +783,7 @@ InstructionCost X86TTIImpl::getArithmeticInstrCost(
{ ISD::MUL, MVT::v16i16, 1 }, // pmullw
{ ISD::MUL, MVT::v8i32, 2 }, // pmulld (Haswell from agner.org)
- { ISD::MUL, MVT::v4i64, 8 }, // 3*pmuludq/3*shift/2*add
+ { ISD::MUL, MVT::v4i64, 6 }, // 3*pmuludq/3*shift/2*add
{ ISD::FNEG, MVT::v4f64, 1 }, // Haswell from http://www.agner.org/
{ ISD::FNEG, MVT::v8f32, 1 }, // Haswell from http://www.agner.org/
diff --git a/llvm/test/Analysis/CostModel/X86/arith-fix.ll b/llvm/test/Analysis/CostModel/X86/arith-fix.ll
index 11fbd53996585..6b32547bb4e57 100644
--- a/llvm/test/Analysis/CostModel/X86/arith-fix.ll
+++ b/llvm/test/Analysis/CostModel/X86/arith-fix.ll
@@ -95,9 +95,9 @@ define i32 @smul(i32 %arg) {
; AVX2-NEXT: Cost Model: Found an estimated cost of 39 for instruction: %V4I64 = call <4 x i64> @llvm.smul.fix.v4i64(<4 x i64> undef, <4 x i64> undef, i32 3)
; AVX2-NEXT: Cost Model: Found an estimated cost of 78 for instruction: %V8I64 = call <8 x i64> @llvm.smul.fix.v8i64(<8 x i64> undef, <8 x i64> undef, i32 3)
; AVX2-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %I32 = call i32 @llvm.smul.fix.i32(i32 undef, i32 undef, i32 3)
-; AVX2-NEXT: Cost Model: Found an estimated cost of 17 for instruction: %V4I32 = call <4 x i32> @llvm.smul.fix.v4i32(<4 x i32> undef, <4 x i32> undef, i32 3)
-; AVX2-NEXT: Cost Model: Found an estimated cost of 31 for instruction: %V8I32 = call <8 x i32> @llvm.smul.fix.v8i32(<8 x i32> undef, <8 x i32> undef, i32 3)
-; AVX2-NEXT: Cost Model: Found an estimated cost of 62 for instruction: %V16I32 = call <16 x i32> @llvm.smul.fix.v16i32(<16 x i32> undef, <16 x i32> undef, i32 3)
+; AVX2-NEXT: Cost Model: Found an estimated cost of 15 for instruction: %V4I32 = call <4 x i32> @llvm.smul.fix.v4i32(<4 x i32> undef, <4 x i32> undef, i32 3)
+; AVX2-NEXT: Cost Model: Found an estimated cost of 27 for instruction: %V8I32 = call <8 x i32> @llvm.smul.fix.v8i32(<8 x i32> undef, <8 x i32> undef, i32 3)
+; AVX2-NEXT: Cost Model: Found an estimated cost of 54 for instruction: %V16I32 = call <16 x i32> @llvm.smul.fix.v16i32(<16 x i32> undef, <16 x i32> undef, i32 3)
; AVX2-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %I16 = call i16 @llvm.smul.fix.i16(i16 undef, i16 undef, i32 3)
; AVX2-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %V8I16 = call <8 x i16> @llvm.smul.fix.v8i16(<8 x i16> undef, <8 x i16> undef, i32 3)
; AVX2-NEXT: Cost Model: Found an estimated cost of 25 for instruction: %V16I16 = call <16 x i16> @llvm.smul.fix.v16i16(<16 x i16> undef, <16 x i16> undef, i32 3)
@@ -114,9 +114,9 @@ define i32 @smul(i32 %arg) {
; AVX512F-NEXT: Cost Model: Found an estimated cost of 39 for instruction: %V4I64 = call <4 x i64> @llvm.smul.fix.v4i64(<4 x i64> undef, <4 x i64> undef, i32 3)
; AVX512F-NEXT: Cost Model: Found an estimated cost of 79 for instruction: %V8I64 = call <8 x i64> @llvm.smul.fix.v8i64(<8 x i64> undef, <8 x i64> undef, i32 3)
; AVX512F-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %I32 = call i32 @llvm.smul.fix.i32(i32 undef, i32 undef, i32 3)
-; AVX512F-NEXT: Cost Model: Found an estimated cost of 15 for instruction: %V4I32 = call <4 x i32> @llvm.smul.fix.v4i32(<4 x i32> undef, <4 x i32> undef, i32 3)
-; AVX512F-NEXT: Cost Model: Found an estimated cost of 15 for instruction: %V8I32 = call <8 x i32> @llvm.smul.fix.v8i32(<8 x i32> undef, <8 x i32> undef, i32 3)
-; AVX512F-NEXT: Cost Model: Found an estimated cost of 31 for instruction: %V16I32 = call <16 x i32> @llvm.smul.fix.v16i32(<16 x i32> undef, <16 x i32> undef, i32 3)
+; AVX512F-NEXT: Cost Model: Found an estimated cost of 13 for instruction: %V4I32 = call <4 x i32> @llvm.smul.fix.v4i32(<4 x i32> undef, <4 x i32> undef, i32 3)
+; AVX512F-NEXT: Cost Model: Found an estimated cost of 13 for instruction: %V8I32 = call <8 x i32> @llvm.smul.fix.v8i32(<8 x i32> undef, <8 x i32> undef, i32 3)
+; AVX512F-NEXT: Cost Model: Found an estimated cost of 27 for instruction: %V16I32 = call <16 x i32> @llvm.smul.fix.v16i32(<16 x i32> undef, <16 x i32> undef, i32 3)
; AVX512F-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %I16 = call i16 @llvm.smul.fix.i16(i16 undef, i16 undef, i32 3)
; AVX512F-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V8I16 = call <8 x i16> @llvm.smul.fix.v8i16(<8 x i16> undef, <8 x i16> undef, i32 3)
; AVX512F-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V16I16 = call <16 x i16> @llvm.smul.fix.v16i16(<16 x i16> undef, <16 x i16> undef, i32 3)
@@ -133,9 +133,9 @@ define i32 @smul(i32 %arg) {
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 39 for instruction: %V4I64 = call <4 x i64> @llvm.smul.fix.v4i64(<4 x i64> undef, <4 x i64> undef, i32 3)
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 79 for instruction: %V8I64 = call <8 x i64> @llvm.smul.fix.v8i64(<8 x i64> undef, <8 x i64> undef, i32 3)
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %I32 = call i32 @llvm.smul.fix.i32(i32 undef, i32 undef, i32 3)
-; AVX512BW-NEXT: Cost Model: Found an estimated cost of 15 for instruction: %V4I32 = call <4 x i32> @llvm.smul.fix.v4i32(<4 x i32> undef, <4 x i32> undef, i32 3)
-; AVX512BW-NEXT: Cost Model: Found an estimated cost of 15 for instruction: %V8I32 = call <8 x i32> @llvm.smul.fix.v8i32(<8 x i32> undef, <8 x i32> undef, i32 3)
-; AVX512BW-NEXT: Cost Model: Found an estimated cost of 31 for instruction: %V16I32 = call <16 x i32> @llvm.smul.fix.v16i32(<16 x i32> undef, <16 x i32> undef, i32 3)
+; AVX512BW-NEXT: Cost Model: Found an estimated cost of 13 for instruction: %V4I32 = call <4 x i32> @llvm.smul.fix.v4i32(<4 x i32> undef, <4 x i32> undef, i32 3)
+; AVX512BW-NEXT: Cost Model: Found an estimated cost of 13 for instruction: %V8I32 = call <8 x i32> @llvm.smul.fix.v8i32(<8 x i32> undef, <8 x i32> undef, i32 3)
+; AVX512BW-NEXT: Cost Model: Found an estimated cost of 27 for instruction: %V16I32 = call <16 x i32> @llvm.smul.fix.v16i32(<16 x i32> undef, <16 x i32> undef, i32 3)
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %I16 = call i16 @llvm.smul.fix.i16(i16 undef, i16 undef, i32 3)
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V8I16 = call <8 x i16> @llvm.smul.fix.v8i16(<8 x i16> undef, <8 x i16> undef, i32 3)
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V16I16 = call <16 x i16> @llvm.smul.fix.v16i16(<16 x i16> undef, <16 x i16> undef, i32 3)
@@ -152,9 +152,9 @@ define i32 @smul(i32 %arg) {
; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 39 for instruction: %V4I64 = call <4 x i64> @llvm.smul.fix.v4i64(<4 x i64> undef, <4 x i64> undef, i32 3)
; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 79 for instruction: %V8I64 = call <8 x i64> @llvm.smul.fix.v8i64(<8 x i64> undef, <8 x i64> undef, i32 3)
; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %I32 = call i32 @llvm.smul.fix.i32(i32 undef, i32 undef, i32 3)
-; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4I32 = call <4 x i32> @llvm.smul.fix.v4i32(<4 x i32> undef, <4 x i32> undef, i32 3)
-; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V8I32 = call <8 x i32> @llvm.smul.fix.v8i32(<8 x i32> undef, <8 x i32> undef, i32 3)
-; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 17 for instruction: %V16I32 = call <16 x i32> @llvm.smul.fix.v16i32(<16 x i32> undef, <16 x i32> undef, i32 3)
+; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %V4I32 = call <4 x i32> @llvm.smul.fix.v4i32(<4 x i32> undef, <4 x i32> undef, i32 3)
+; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %V8I32 = call <8 x i32> @llvm.smul.fix.v8i32(<8 x i32> undef, <8 x i32> undef, i32 3)
+; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 19 for instruction: %V16I32 = call <16 x i32> @llvm.smul.fix.v16i32(<16 x i32> undef, <16 x i32> undef, i32 3)
; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %I16 = call i16 @llvm.smul.fix.i16(i16 undef, i16 undef, i32 3)
; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V8I16 = call <8 x i16> @llvm.smul.fix.v8i16(<8 x i16> undef, <8 x i16> undef, i32 3)
; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V16I16 = call <16 x i16> @llvm.smul.fix.v16i16(<16 x i16> undef, <16 x i16> undef, i32 3)
@@ -329,9 +329,9 @@ define i32 @umul(i32 %arg) {
; AVX2-NEXT: Cost Model: Found an estimated cost of 39 for instruction: %V4I64 = call <4 x i64> @llvm.umul.fix.v4i64(<4 x i64> undef, <4 x i64> undef, i32 3)
; AVX2-NEXT: Cost Model: Found an estimated cost of 78 for instruction: %V8I64 = call <8 x i64> @llvm.umul.fix.v8i64(<8 x i64> undef, <8 x i64> undef, i32 3)
; AVX2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.umul.fix.i32(i32 undef, i32 undef, i32 3)
-; AVX2-NEXT: Cost Model: Found an estimated cost of 17 for instruction: %V4I32 = call <4 x i32> @llvm.umul.fix.v4i32(<4 x i32> undef, <4 x i32> undef, i32 3)
-; AVX2-NEXT: Cost Model: Found an estimated cost of 31 for instruction: %V8I32 = call <8 x i32> @llvm.umul.fix.v8i32(<8 x i32> undef, <8 x i32> undef, i32 3)
-; AVX2-NEXT: Cost Model: Found an estimated cost of 62 for instruction: %V16I32 = call <16 x i32> @llvm.umul.fix.v16i32(<16 x i32> undef, <16 x i32> undef, i32 3)
+; AVX2-NEXT: Cost Model: Found an estimated cost of 15 for instruction: %V4I32 = call <4 x i32> @llvm.umul.fix.v4i32(<4 x i32> undef, <4 x i32> undef, i32 3)
+; AVX2-NEXT: Cost Model: Found an estimated cost of 27 for instruction: %V8I32 = call <8 x i32> @llvm.umul.fix.v8i32(<8 x i32> undef, <8 x i32> undef, i32 3)
+; AVX2-NEXT: Cost Model: Found an estimated cost of 54 for instruction: %V16I32 = call <16 x i32> @llvm.umul.fix.v16i32(<16 x i32> undef, <16 x i32> undef, i32 3)
; AVX2-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %I16 = call i16 @llvm.umul.fix.i16(i16 undef, i16 undef, i32 3)
; AVX2-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %V8I16 = call <8 x i16> @llvm.umul.fix.v8i16(<8 x i16> undef, <8 x i16> undef, i32 3)
; AVX2-NEXT: Cost Model: Found an estimated cost of 25 for instruction: %V16I16 = call <16 x i16> @llvm.umul.fix.v16i16(<16 x i16> undef, <16 x i16> undef, i32 3)
@@ -348,9 +348,9 @@ define i32 @umul(i32 %arg) {
; AVX512F-NEXT: Cost Model: Found an estimated cost of 39 for instruction: %V4I64 = call <4 x i64> @llvm.umul.fix.v4i64(<4 x i64> undef, <4 x i64> undef, i32 3)
; AVX512F-NEXT: Cost Model: Found an estimated cost of 79 for instruction: %V8I64 = call <8 x i64> @llvm.umul.fix.v8i64(<8 x i64> undef, <8 x i64> undef, i32 3)
; AVX512F-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.umul.fix.i32(i32 undef, i32 undef, i32 3)
-; AVX512F-NEXT: Cost Model: Found an estimated cost of 15 for instruction: %V4I32 = call <4 x i32> @llvm.umul.fix.v4i32(<4 x i32> undef, <4 x i32> undef, i32 3)
-; AVX512F-NEXT: Cost Model: Found an estimated cost of 15 for instruction: %V8I32 = call <8 x i32> @llvm.umul.fix.v8i32(<8 x i32> undef, <8 x i32> undef, i32 3)
-; AVX512F-NEXT: Cost Model: Found an estimated cost of 31 for instruction: %V16I32 = call <16 x i32> @llvm.umul.fix.v16i32(<16 x i32> undef, <16 x i32> undef, i32 3)
+; AVX512F-NEXT: Cost Model: Found an estimated cost of 13 for instruction: %V4I32 = call <4 x i32> @llvm.umul.fix.v4i32(<4 x i32> undef, <4 x i32> undef, i32 3)
+; AVX512F-NEXT: Cost Model: Found an estimated cost of 13 for instruction: %V8I32 = call <8 x i32> @llvm.umul.fix.v8i32(<8 x i32> undef, <8 x i32> undef, i32 3)
+; AVX512F-NEXT: Cost Model: Found an estimated cost of 27 for instruction: %V16I32 = call <16 x i32> @llvm.umul.fix.v16i32(<16 x i32> undef, <16 x i32> undef, i32 3)
; AVX512F-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %I16 = call i16 @llvm.umul.fix.i16(i16 undef, i16 undef, i32 3)
; AVX512F-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V8I16 = call <8 x i16> @llvm.umul.fix.v8i16(<8 x i16> undef, <8 x i16> undef, i32 3)
; AVX512F-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V16I16 = call <16 x i16> @llvm.umul.fix.v16i16(<16 x i16> undef, <16 x i16> undef, i32 3)
@@ -367,9 +367,9 @@ define i32 @umul(i32 %arg) {
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 39 for instruction: %V4I64 = call <4 x i64> @llvm.umul.fix.v4i64(<4 x i64> undef, <4 x i64> undef, i32 3)
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 79 for instruction: %V8I64 = call <8 x i64> @llvm.umul.fix.v8i64(<8 x i64> undef, <8 x i64> undef, i32 3)
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.umul.fix.i32(i32 undef, i32 undef, i32 3)
-; AVX512BW-NEXT: Cost Model: Found an estimated cost of 15 for instruction: %V4I32 = call <4 x i32> @llvm.umul.fix.v4i32(<4 x i32> undef, <4 x i32> undef, i32 3)
-; AVX512BW-NEXT: Cost Model: Found an estimated cost of 15 for instruction: %V8I32 = call <8 x i32> @llvm.umul.fix.v8i32(<8 x i32> undef, <8 x i32> undef, i32 3)
-; AVX512BW-NEXT: Cost Model: Found an estimated cost of 31 for instruction: %V16I32 = call <16 x i32> @llvm.umul.fix.v16i32(<16 x i32> undef, <16 x i32> undef, i32 3)
+; AVX512BW-NEXT: Cost Model: Found an estimated cost of 13 for instruction: %V4I32 = call <4 x i32> @llvm.umul.fix.v4i32(<4 x i32> undef, <4 x i32> undef, i32 3)
+; AVX512BW-NEXT: Cost Model: Found an estimated cost of 13 for instruction: %V8I32 = call <8 x i32> @llvm.umul.fix.v8i32(<8 x i32> undef, <8 x i32> undef, i32 3)
+; AVX512BW-NEXT: Cost Model: Found an estimated cost of 27 for instruction: %V16I32 = call <16 x i32> @llvm.umul.fix.v16i32(<16 x i32> undef, <16 x i32> undef, i32 3)
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %I16 = call i16 @llvm.umul.fix.i16(i16 undef, i16 undef, i32 3)
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V8I16 = call <8 x i16> @llvm.umul.fix.v8i16(<8 x i16> undef, <8 x i16> undef, i32 3)
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V16I16 = call <16 x i16> @llvm.umul.fix.v16i16(<16 x i16> undef, <16 x i16> undef, i32 3)
@@ -386,9 +386,9 @@ define i32 @umul(i32 %arg) {
; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 39 for instruction: %V4I64 = call <4 x i64> @llvm.umul.fix.v4i64(<4 x i64> undef, <4 x i64> undef, i32 3)
; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 79 for instruction: %V8I64 = call <8 x i64> @llvm.umul.fix.v8i64(<8 x i64> undef, <8 x i64> undef, i32 3)
; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.umul.fix.i32(i32 undef, i32 undef, i32 3)
-; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4I32 = call <4 x i32> @llvm.umul.fix.v4i32(<4 x i32> undef, <4 x i32> undef, i32 3)
-; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V8I32 = call <8 x i32> @llvm.umul.fix.v8i32(<8 x i32> undef, <8 x i32> undef, i32 3)
-; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 17 for instruction: %V16I32 = call <16 x i32> @llvm.umul.fix.v16i32(<16 x i32> undef, <16 x i32> undef, i32 3)
+; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %V4I32 = call <4 x i32> @llvm.umul.fix.v4i32(<4 x i32> undef, <4 x i32> undef, i32 3)
+; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %V8I32 = call <8 x i32> @llvm.umul.fix.v8i32(<8 x i32> undef, <8 x i32> undef, i32 3)
+; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 19 for instruction: %V16I32 = call <16 x i32> @llvm.umul.fix.v16i32(<16 x i32> undef, <16 x i32> undef, i32 3)
; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %I16 = call i16 @llvm.umul.fix.i16(i16 undef, i16 undef, i32 3)
; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V8I16 = call <8 x i16> @llvm.umul.fix.v8i16(<8 x i16> undef, <8 x i16> undef, i32 3)
; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V16I16 = call <16 x i16> @llvm.umul.fix.v16i16(<16 x i16> undef, <16 x i16> undef, i32 3)
diff --git a/llvm/test/Analysis/CostModel/X86/arith-overflow.ll b/llvm/test/Analysis/CostModel/X86/arith-overflow.ll
index 2e852d88742f2..05d7550f394f6 100644
--- a/llvm/test/Analysis/CostModel/X86/arith-overflow.ll
+++ b/llvm/test/Analysis/CostModel/X86/arith-overflow.ll
@@ -1051,9 +1051,9 @@ define i32 @smul(i32 %arg) {
; AVX2-NEXT: Cost Model: Found an estimated cost of 42 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.smul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
; AVX2-NEXT: Cost Model: Found an estimated cost of 84 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.smul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
; AVX2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %I32 = call { i32, i1 } @llvm.smul.with.overflow.i32(i32 undef, i32 undef)
-; AVX2-NEXT: Cost Model: Found an estimated cost of 17 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.smul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; AVX2-NEXT: Cost Model: Found an estimated cost of 31 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.smul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; AVX2-NEXT: Cost Model: Found an estimated cost of 62 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.smul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; AVX2-NEXT: Cost Model: Found an estimated cost of 15 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.smul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
+; AVX2-NEXT: Cost Model: Found an estimated cost of 27 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.smul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; AVX2-NEXT: Cost Model: Found an estimated cost of 54 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.smul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
; AVX2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %I16 = call { i16, i1 } @llvm.smul.with.overflow.i16(i16 undef, i16 undef)
; AVX2-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.smul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
; AVX2-NEXT: Cost Model: Found an estimated cost of 25 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.smul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
@@ -1070,9 +1070,9 @@ define i32 @smul(i32 %arg) {
; AVX512F-NEXT: Cost Model: Found an estimated cost of 39 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.smul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
; AVX512F-NEXT: Cost Model: Found an estimated cost of 79 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.smul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
; AVX512F-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %I32 = call { i32, i1 } @llvm.smul.with.overflow.i32(i32 undef, i32 undef)
-; AVX512F-NEXT: Cost Model: Found an estimated cost of 15 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.smul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; AVX512F-NEXT: Cost Model: Found an estimated cost of 15 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.smul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; AVX512F-NEXT: Cost Model: Found an estimated cost of 31 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.smul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; AVX512F-NEXT: Cost Model: Found an estimated cost of 13 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.smul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
+; AVX512F-NEXT: Cost Model: Found an estimated cost of 13 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.smul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; AVX512F-NEXT: Cost Model: Found an estimated cost of 27 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.smul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
; AVX512F-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %I16 = call { i16, i1 } @llvm.smul.with.overflow.i16(i16 undef, i16 undef)
; AVX512F-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.smul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
; AVX512F-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.smul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
@@ -1089,9 +1089,9 @@ define i32 @smul(i32 %arg) {
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 39 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.smul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 79 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.smul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %I32 = call { i32, i1 } @llvm.smul.with.overflow.i32(i32 undef, i32 undef)
-; AVX512BW-NEXT: Cost Model: Found an estimated cost of 15 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.smul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; AVX512BW-NEXT: Cost Model: Found an estimated cost of 15 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.smul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; AVX512BW-NEXT: Cost Model: Found an estimated cost of 31 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.smul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; AVX512BW-NEXT: Cost Model: Found an estimated cost of 13 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.smul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
+; AVX512BW-NEXT: Cost Model: Found an estimated cost of 13 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.smul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; AVX512BW-NEXT: Cost Model: Found an estimated cost of 27 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.smul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %I16 = call { i16, i1 } @llvm.smul.with.overflow.i16(i16 undef, i16 undef)
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.smul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.smul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
@@ -1108,9 +1108,9 @@ define i32 @smul(i32 %arg) {
; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 39 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.smul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 79 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.smul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %I32 = call { i32, i1 } @llvm.smul.with.overflow.i32(i32 undef, i32 undef)
-; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.smul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.smul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 17 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.smul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.smul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
+; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.smul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 19 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.smul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %I16 = call { i16, i1 } @llvm.smul.with.overflow.i16(i16 undef, i16 undef)
; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.smul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.smul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
@@ -1289,9 +1289,9 @@ define i32 @umul(i32 %arg) {
; AVX2-NEXT: Cost Model: Found an estimated cost of 38 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.umul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
; AVX2-NEXT: Cost Model: Found an estimated cost of 76 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.umul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
; AVX2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %I32 = call { i32, i1 } @llvm.umul.with.overflow.i32(i32 undef, i32 undef)
-; AVX2-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.umul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; AVX2-NEXT: Cost Model: Found an estimated cost of 30 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.umul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; AVX2-NEXT: Cost Model: Found an estimated cost of 60 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.umul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; AVX2-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.umul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
+; AVX2-NEXT: Cost Model: Found an estimated cost of 26 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.umul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; AVX2-NEXT: Cost Model: Found an estimated cost of 52 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.umul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
; AVX2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %I16 = call { i16, i1 } @llvm.umul.with.overflow.i16(i16 undef, i16 undef)
; AVX2-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.umul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
; AVX2-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.umul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
@@ -1308,9 +1308,9 @@ define i32 @umul(i32 %arg) {
; AVX512F-NEXT: Cost Model: Found an estimated cost of 38 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.umul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
; AVX512F-NEXT: Cost Model: Found an estimated cost of 78 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.umul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
; AVX512F-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %I32 = call { i32, i1 } @llvm.umul.with.overflow.i32(i32 undef, i32 undef)
-; AVX512F-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.umul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; AVX512F-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.umul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; AVX512F-NEXT: Cost Model: Found an estimated cost of 30 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.umul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; AVX512F-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.umul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
+; AVX512F-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.umul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; AVX512F-NEXT: Cost Model: Found an estimated cost of 26 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.umul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
; AVX512F-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %I16 = call { i16, i1 } @llvm.umul.with.overflow.i16(i16 undef, i16 undef)
; AVX512F-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.umul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
; AVX512F-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.umul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
@@ -1327,9 +1327,9 @@ define i32 @umul(i32 %arg) {
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 38 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.umul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 78 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.umul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %I32 = call { i32, i1 } @llvm.umul.with.overflow.i32(i32 undef, i32 undef)
-; AVX512BW-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.umul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; AVX512BW-NEXT: Cost Model: Found an estimated cost of 14 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.umul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; AVX512BW-NEXT: Cost Model: Found an estimated cost of 30 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.umul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; AVX512BW-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.umul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
+; AVX512BW-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.umul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; AVX512BW-NEXT: Cost Model: Found an estimated cost of 26 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.umul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %I16 = call { i16, i1 } @llvm.umul.with.overflow.i16(i16 undef, i16 undef)
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.umul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.umul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
@@ -1346,9 +1346,9 @@ define i32 @umul(i32 %arg) {
; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 38 for instruction: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.umul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 78 for instruction: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.umul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %I32 = call { i32, i1 } @llvm.umul.with.overflow.i32(i32 undef, i32 undef)
-; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.umul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.umul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.umul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.umul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
+; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.umul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.umul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %I16 = call { i16, i1 } @llvm.umul.with.overflow.i16(i16 undef, i16 undef)
; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.umul.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.umul.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
diff --git a/llvm/test/Analysis/CostModel/X86/arith.ll b/llvm/test/Analysis/CostModel/X86/arith.ll
index 8cb8b627a0270..d9a714f7caeef 100644
--- a/llvm/test/Analysis/CostModel/X86/arith.ll
+++ b/llvm/test/Analysis/CostModel/X86/arith.ll
@@ -998,8 +998,8 @@ define i32 @mul(i32 %arg) {
; AVX2-LABEL: 'mul'
; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I64 = mul i64 undef, undef
; AVX2-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = mul <2 x i64> undef, undef
-; AVX2-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4I64 = mul <4 x i64> undef, undef
-; AVX2-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8I64 = mul <8 x i64> undef, undef
+; AVX2-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V4I64 = mul <4 x i64> undef, undef
+; AVX2-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %V8I64 = mul <8 x i64> undef, undef
; AVX2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I32 = mul i32 undef, undef
; AVX2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = mul <4 x i32> undef, undef
; AVX2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8I32 = mul <8 x i32> undef, undef
@@ -1020,8 +1020,8 @@ define i32 @mul(i32 %arg) {
; AVX512F-LABEL: 'mul'
; AVX512F-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I64 = mul i64 undef, undef
; AVX512F-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = mul <2 x i64> undef, undef
-; AVX512F-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4I64 = mul <4 x i64> undef, undef
-; AVX512F-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V8I64 = mul <8 x i64> undef, undef
+; AVX512F-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V4I64 = mul <4 x i64> undef, undef
+; AVX512F-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V8I64 = mul <8 x i64> undef, undef
; AVX512F-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I32 = mul i32 undef, undef
; AVX512F-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = mul <4 x i32> undef, undef
; AVX512F-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = mul <8 x i32> undef, undef
@@ -1042,8 +1042,8 @@ define i32 @mul(i32 %arg) {
; AVX512BW-LABEL: 'mul'
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I64 = mul i64 undef, undef
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = mul <2 x i64> undef, undef
-; AVX512BW-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4I64 = mul <4 x i64> undef, undef
-; AVX512BW-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V8I64 = mul <8 x i64> undef, undef
+; AVX512BW-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V4I64 = mul <4 x i64> undef, undef
+; AVX512BW-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V8I64 = mul <8 x i64> undef, undef
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I32 = mul i32 undef, undef
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = mul <4 x i32> undef, undef
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = mul <8 x i32> undef, undef
@@ -1063,9 +1063,9 @@ define i32 @mul(i32 %arg) {
;
; AVX512DQ-LABEL: 'mul'
; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I64 = mul i64 undef, undef
-; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = mul <2 x i64> undef, undef
-; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4I64 = mul <4 x i64> undef, undef
-; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8I64 = mul <8 x i64> undef, undef
+; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = mul <2 x i64> undef, undef
+; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4I64 = mul <4 x i64> undef, undef
+; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8I64 = mul <8 x i64> undef, undef
; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %I32 = mul i32 undef, undef
; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = mul <4 x i32> undef, undef
; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8I32 = mul <8 x i32> undef, undef
diff --git a/llvm/test/Analysis/CostModel/X86/reduce-mul.ll b/llvm/test/Analysis/CostModel/X86/reduce-mul.ll
index 084b091be8b04..1033d3e02f8fc 100644
--- a/llvm/test/Analysis/CostModel/X86/reduce-mul.ll
+++ b/llvm/test/Analysis/CostModel/X86/reduce-mul.ll
@@ -45,32 +45,32 @@ define i32 @reduce_i64(i32 %arg) {
; AVX2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.vector.reduce.mul.v1i64(<1 x i64> undef)
; AVX2-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V2 = call i64 @llvm.vector.reduce.mul.v2i64(<2 x i64> undef)
; AVX2-NEXT: Cost Model: Found an estimated cost of 15 for instruction: %V4 = call i64 @llvm.vector.reduce.mul.v4i64(<4 x i64> undef)
-; AVX2-NEXT: Cost Model: Found an estimated cost of 23 for instruction: %V8 = call i64 @llvm.vector.reduce.mul.v8i64(<8 x i64> undef)
-; AVX2-NEXT: Cost Model: Found an estimated cost of 39 for instruction: %V16 = call i64 @llvm.vector.reduce.mul.v16i64(<16 x i64> undef)
+; AVX2-NEXT: Cost Model: Found an estimated cost of 21 for instruction: %V8 = call i64 @llvm.vector.reduce.mul.v8i64(<8 x i64> undef)
+; AVX2-NEXT: Cost Model: Found an estimated cost of 33 for instruction: %V16 = call i64 @llvm.vector.reduce.mul.v16i64(<16 x i64> undef)
; AVX2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
;
; AVX512F-LABEL: 'reduce_i64'
; AVX512F-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.vector.reduce.mul.v1i64(<1 x i64> undef)
; AVX512F-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V2 = call i64 @llvm.vector.reduce.mul.v2i64(<2 x i64> undef)
; AVX512F-NEXT: Cost Model: Found an estimated cost of 15 for instruction: %V4 = call i64 @llvm.vector.reduce.mul.v4i64(<4 x i64> undef)
-; AVX512F-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %V8 = call i64 @llvm.vector.reduce.mul.v8i64(<8 x i64> undef)
-; AVX512F-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %V16 = call i64 @llvm.vector.reduce.mul.v16i64(<16 x i64> undef)
+; AVX512F-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %V8 = call i64 @llvm.vector.reduce.mul.v8i64(<8 x i64> undef)
+; AVX512F-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V16 = call i64 @llvm.vector.reduce.mul.v16i64(<16 x i64> undef)
; AVX512F-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
;
; AVX512BW-LABEL: 'reduce_i64'
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.vector.reduce.mul.v1i64(<1 x i64> undef)
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V2 = call i64 @llvm.vector.reduce.mul.v2i64(<2 x i64> undef)
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 15 for instruction: %V4 = call i64 @llvm.vector.reduce.mul.v4i64(<4 x i64> undef)
-; AVX512BW-NEXT: Cost Model: Found an estimated cost of 24 for instruction: %V8 = call i64 @llvm.vector.reduce.mul.v8i64(<8 x i64> undef)
-; AVX512BW-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %V16 = call i64 @llvm.vector.reduce.mul.v16i64(<16 x i64> undef)
+; AVX512BW-NEXT: Cost Model: Found an estimated cost of 22 for instruction: %V8 = call i64 @llvm.vector.reduce.mul.v8i64(<8 x i64> undef)
+; AVX512BW-NEXT: Cost Model: Found an estimated cost of 28 for instruction: %V16 = call i64 @llvm.vector.reduce.mul.v16i64(<16 x i64> undef)
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
;
; AVX512DQ-LABEL: 'reduce_i64'
; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.vector.reduce.mul.v1i64(<1 x i64> undef)
-; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V2 = call i64 @llvm.vector.reduce.mul.v2i64(<2 x i64> undef)
-; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %V4 = call i64 @llvm.vector.reduce.mul.v4i64(<4 x i64> undef)
-; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %V8 = call i64 @llvm.vector.reduce.mul.v8i64(<8 x i64> undef)
-; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V16 = call i64 @llvm.vector.reduce.mul.v16i64(<16 x i64> undef)
+; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2 = call i64 @llvm.vector.reduce.mul.v2i64(<2 x i64> undef)
+; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %V4 = call i64 @llvm.vector.reduce.mul.v4i64(<4 x i64> undef)
+; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V8 = call i64 @llvm.vector.reduce.mul.v8i64(<8 x i64> undef)
+; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %V16 = call i64 @llvm.vector.reduce.mul.v16i64(<16 x i64> undef)
; AVX512DQ-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
;
%V1 = call i64 @llvm.vector.reduce.mul.v1i64(<1 x i64> undef)
diff --git a/llvm/test/Analysis/CostModel/X86/rem.ll b/llvm/test/Analysis/CostModel/X86/rem.ll
index 8248a3147c896..d5c063a564eaa 100644
--- a/llvm/test/Analysis/CostModel/X86/rem.ll
+++ b/llvm/test/Analysis/CostModel/X86/rem.ll
@@ -805,8 +805,8 @@ define i32 @srem_constpow2() {
; AVX2-LABEL: 'srem_constpow2'
; AVX2-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %I64 = srem i64 undef, 16
; AVX2-NEXT: Cost Model: Found an estimated cost of 17 for instruction: %V2i64 = srem <2 x i64> undef, <i64 8, i64 16>
-; AVX2-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %V4i64 = srem <4 x i64> undef, <i64 2, i64 4, i64 8, i64 16>
-; AVX2-NEXT: Cost Model: Found an estimated cost of 40 for instruction: %V8i64 = srem <8 x i64> undef, <i64 2, i64 4, i64 8, i64 16, i64 32, i64 64, i64 128, i64 256>
+; AVX2-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %V4i64 = srem <4 x i64> undef, <i64 2, i64 4, i64 8, i64 16>
+; AVX2-NEXT: Cost Model: Found an estimated cost of 36 for instruction: %V8i64 = srem <8 x i64> undef, <i64 2, i64 4, i64 8, i64 16, i64 32, i64 64, i64 128, i64 256>
; AVX2-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %I32 = srem i32 undef, 16
; AVX2-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %V4i32 = srem <4 x i32> undef, <i32 2, i32 4, i32 8, i32 16>
; AVX2-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V8i32 = srem <8 x i32> undef, <i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256>
@@ -824,8 +824,8 @@ define i32 @srem_constpow2() {
; AVX512F-LABEL: 'srem_constpow2'
; AVX512F-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %I64 = srem i64 undef, 16
; AVX512F-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %V2i64 = srem <2 x i64> undef, <i64 8, i64 16>
-; AVX512F-NEXT: Cost Model: Found an estimated cost of 13 for instruction: %V4i64 = srem <4 x i64> undef, <i64 2, i64 4, i64 8, i64 16>
-; AVX512F-NEXT: Cost Model: Found an estimated cost of 13 for instruction: %V8i64 = srem <8 x i64> undef, <i64 2, i64 4, i64 8, i64 16, i64 32, i64 64, i64 128, i64 256>
+; AVX512F-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %V4i64 = srem <4 x i64> undef, <i64 2, i64 4, i64 8, i64 16>
+; AVX512F-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %V8i64 = srem <8 x i64> undef, <i64 2, i64 4, i64 8, i64 16, i64 32, i64 64, i64 128, i64 256>
; AVX512F-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %I32 = srem i32 undef, 16
; AVX512F-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V4i32 = srem <4 x i32> undef, <i32 2, i32 4, i32 8, i32 16>
; AVX512F-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V8i32 = srem <8 x i32> undef, <i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256>
@@ -843,8 +843,8 @@ define i32 @srem_constpow2() {
; AVX512BW-LABEL: 'srem_constpow2'
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %I64 = srem i64 undef, 16
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %V2i64 = srem <2 x i64> undef, <i64 8, i64 16>
-; AVX512BW-NEXT: Cost Model: Found an estimated cost of 13 for instruction: %V4i64 = srem <4 x i64> undef, <i64 2, i64 4, i64 8, i64 16>
-; AVX512BW-NEXT: Cost Model: Found an estimated cost of 13 for instruction: %V8i64 = srem <8 x i64> undef, <i64 2, i64 4, i64 8, i64 16, i64 32, i64 64, i64 128, i64 256>
+; AVX512BW-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %V4i64 = srem <4 x i64> undef, <i64 2, i64 4, i64 8, i64 16>
+; AVX512BW-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %V8i64 = srem <8 x i64> undef, <i64 2, i64 4, i64 8, i64 16, i64 32, i64 64, i64 128, i64 256>
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %I32 = srem i32 undef, 16
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V4i32 = srem <4 x i32> undef, <i32 2, i32 4, i32 8, i32 16>
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V8i32 = srem <8 x i32> undef, <i32 2, i32 4, i32 8, i32 16, i32 32, i32 64, i32 128, i32 256>
@@ -1100,8 +1100,8 @@ define i32 @srem_uniformconstpow2() {
; AVX2-LABEL: 'srem_uniformconstpow2'
; AVX2-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %I64 = srem i64 undef, 16
; AVX2-NEXT: Cost Model: Found an estimated cost of 17 for instruction: %V2i64 = srem <2 x i64> undef, <i64 16, i64 16>
-; AVX2-NEXT: Cost Model: Found an estimated cost of 19 for instruction: %V4i64 = srem <4 x i64> undef, <i64 16, i64 16, i64 16, i64 16>
-; AVX2-NEXT: Cost Model: Found an estimated cost of 38 for instruction: %V8i64 = srem <8 x i64> undef, <i64 16, i64 16, i64 16, i64 16, i64 16, i64 16, i64 16, i64 16>
+; AVX2-NEXT: Cost Model: Found an estimated cost of 17 for instruction: %V4i64 = srem <4 x i64> undef, <i64 16, i64 16, i64 16, i64 16>
+; AVX2-NEXT: Cost Model: Found an estimated cost of 34 for instruction: %V8i64 = srem <8 x i64> undef, <i64 16, i64 16, i64 16, i64 16, i64 16, i64 16, i64 16, i64 16>
; AVX2-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %I32 = srem i32 undef, 16
; AVX2-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %V4i32 = srem <4 x i32> undef, <i32 16, i32 16, i32 16, i32 16>
; AVX2-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %V8i32 = srem <8 x i32> undef, <i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16>
@@ -1119,8 +1119,8 @@ define i32 @srem_uniformconstpow2() {
; AVX512F-LABEL: 'srem_uniformconstpow2'
; AVX512F-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %I64 = srem i64 undef, 16
; AVX512F-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %V2i64 = srem <2 x i64> undef, <i64 16, i64 16>
-; AVX512F-NEXT: Cost Model: Found an estimated cost of 13 for instruction: %V4i64 = srem <4 x i64> undef, <i64 16, i64 16, i64 16, i64 16>
-; AVX512F-NEXT: Cost Model: Found an estimated cost of 13 for instruction: %V8i64 = srem <8 x i64> undef, <i64 16, i64 16, i64 16, i64 16, i64 16, i64 16, i64 16, i64 16>
+; AVX512F-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %V4i64 = srem <4 x i64> undef, <i64 16, i64 16, i64 16, i64 16>
+; AVX512F-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %V8i64 = srem <8 x i64> undef, <i64 16, i64 16, i64 16, i64 16, i64 16, i64 16, i64 16, i64 16>
; AVX512F-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %I32 = srem i32 undef, 16
; AVX512F-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V4i32 = srem <4 x i32> undef, <i32 16, i32 16, i32 16, i32 16>
; AVX512F-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V8i32 = srem <8 x i32> undef, <i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16>
@@ -1138,8 +1138,8 @@ define i32 @srem_uniformconstpow2() {
; AVX512BW-LABEL: 'srem_uniformconstpow2'
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %I64 = srem i64 undef, 16
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %V2i64 = srem <2 x i64> undef, <i64 16, i64 16>
-; AVX512BW-NEXT: Cost Model: Found an estimated cost of 13 for instruction: %V4i64 = srem <4 x i64> undef, <i64 16, i64 16, i64 16, i64 16>
-; AVX512BW-NEXT: Cost Model: Found an estimated cost of 13 for instruction: %V8i64 = srem <8 x i64> undef, <i64 16, i64 16, i64 16, i64 16, i64 16, i64 16, i64 16, i64 16>
+; AVX512BW-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %V4i64 = srem <4 x i64> undef, <i64 16, i64 16, i64 16, i64 16>
+; AVX512BW-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %V8i64 = srem <8 x i64> undef, <i64 16, i64 16, i64 16, i64 16, i64 16, i64 16, i64 16, i64 16>
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %I32 = srem i32 undef, 16
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V4i32 = srem <4 x i32> undef, <i32 16, i32 16, i32 16, i32 16>
; AVX512BW-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V8i32 = srem <8 x i32> undef, <i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16>
More information about the llvm-commits
mailing list