[CostModel] Handle all cost kinds in getCmpSelInstrCost #148233
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
davemgreen
requested review from
arsenm,
RKSimon,
c-rhodes,
hassnaaHamdi,
nasherm and
sjoerdmeijer
llvmbot
added
backend:AMDGPU
backend:X86
llvm:analysis
|
@llvm/pr-subscribers-llvm-transforms @llvm/pr-subscribers-backend-amdgpu Author: David Green (davemgreen) ChangesCurrently we always produce a cost of 1 for all CostKinds that are not RecipThroughput, which can underestimate the cost if the type has a higher legalization cost (like larger vectors). This relaxes it to cover all cost kinds. Patch is 333.01 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/148233.diff 24 Files Affected:
diff --git a/llvm/include/llvm/CodeGen/BasicTTIImpl.h b/llvm/include/llvm/CodeGen/BasicTTIImpl.h
index fefd36ec54ae2..bff9979493125 100644
--- a/llvm/include/llvm/CodeGen/BasicTTIImpl.h
+++ b/llvm/include/llvm/CodeGen/BasicTTIImpl.h
@@ -1369,8 +1369,7 @@ class BasicTTIImplBase : public TargetTransformInfoImplCRTPBase<T> {
int ISD = TLI->InstructionOpcodeToISD(Opcode);
assert(ISD && "Invalid opcode");
- // TODO: Handle other cost kinds.
- if (CostKind != TTI::TCK_RecipThroughput)
+ if (getTLI()->getValueType(DL, ValTy, true) == MVT::Other)
return BaseT::getCmpSelInstrCost(Opcode, ValTy, CondTy, VecPred, CostKind,
Op1Info, Op2Info, I);
diff --git a/llvm/test/Analysis/CostModel/AArch64/arith-overflow.ll b/llvm/test/Analysis/CostModel/AArch64/arith-overflow.ll
index 8b52b123d6d25..3fbd12ac8813c 100644
--- a/llvm/test/Analysis/CostModel/AArch64/arith-overflow.ll
+++ b/llvm/test/Analysis/CostModel/AArch64/arith-overflow.ll
@@ -27,20 +27,20 @@ define i32 @sadd(i32 %arg) {
; CHECK-LABEL: 'sadd'
; CHECK-NEXT: Cost Model: Found costs of 1 for: %I64 = call { i64, i1 } @llvm.sadd.with.overflow.i64(i64 undef, i64 undef)
; CHECK-NEXT: Cost Model: Found costs of 4 for: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.sadd.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:7 CodeSize:4 Lat:4 SizeLat:4 for: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.sadd.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:13 CodeSize:4 Lat:4 SizeLat:4 for: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.sadd.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:7 CodeSize:6 Lat:6 SizeLat:6 for: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.sadd.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:13 CodeSize:10 Lat:10 SizeLat:10 for: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.sadd.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
; CHECK-NEXT: Cost Model: Found costs of 1 for: %I32 = call { i32, i1 } @llvm.sadd.with.overflow.i32(i32 undef, i32 undef)
; CHECK-NEXT: Cost Model: Found costs of 4 for: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.sadd.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:7 CodeSize:4 Lat:4 SizeLat:4 for: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.sadd.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:13 CodeSize:4 Lat:4 SizeLat:4 for: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.sadd.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:7 CodeSize:6 Lat:6 SizeLat:6 for: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.sadd.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:13 CodeSize:10 Lat:10 SizeLat:10 for: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.sadd.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
; CHECK-NEXT: Cost Model: Found costs of 3 for: %I16 = call { i16, i1 } @llvm.sadd.with.overflow.i16(i16 undef, i16 undef)
; CHECK-NEXT: Cost Model: Found costs of 4 for: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.sadd.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:7 CodeSize:4 Lat:4 SizeLat:4 for: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.sadd.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:14 CodeSize:4 Lat:4 SizeLat:4 for: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.sadd.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:7 CodeSize:6 Lat:6 SizeLat:6 for: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.sadd.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:14 CodeSize:10 Lat:10 SizeLat:10 for: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.sadd.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
; CHECK-NEXT: Cost Model: Found costs of 3 for: %I8 = call { i8, i1 } @llvm.sadd.with.overflow.i8(i8 undef, i8 undef)
; CHECK-NEXT: Cost Model: Found costs of 4 for: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.sadd.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:4 Lat:4 SizeLat:4 for: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.sadd.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:16 CodeSize:4 Lat:4 SizeLat:4 for: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.sadd.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:6 Lat:6 SizeLat:6 for: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.sadd.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:16 CodeSize:10 Lat:10 SizeLat:10 for: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.sadd.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
; CHECK-NEXT: Cost Model: Found costs of RThru:0 CodeSize:1 Lat:1 SizeLat:1 for: ret i32 undef
;
%I64 = call {i64, i1} @llvm.sadd.with.overflow.i64(i64 undef, i64 undef)
@@ -90,20 +90,20 @@ define i32 @uadd(i32 %arg) {
; CHECK-LABEL: 'uadd'
; CHECK-NEXT: Cost Model: Found costs of 1 for: %I64 = call { i64, i1 } @llvm.uadd.with.overflow.i64(i64 undef, i64 undef)
; CHECK-NEXT: Cost Model: Found costs of 2 for: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.uadd.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:2 SizeLat:2 for: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.uadd.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:2 Lat:2 SizeLat:2 for: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.uadd.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:3 Lat:3 SizeLat:3 for: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.uadd.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:5 Lat:5 SizeLat:5 for: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.uadd.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
; CHECK-NEXT: Cost Model: Found costs of 1 for: %I32 = call { i32, i1 } @llvm.uadd.with.overflow.i32(i32 undef, i32 undef)
; CHECK-NEXT: Cost Model: Found costs of 2 for: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.uadd.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:2 SizeLat:2 for: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.uadd.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:2 Lat:2 SizeLat:2 for: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.uadd.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:3 Lat:3 SizeLat:3 for: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.uadd.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:5 Lat:5 SizeLat:5 for: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.uadd.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
; CHECK-NEXT: Cost Model: Found costs of 3 for: %I16 = call { i16, i1 } @llvm.uadd.with.overflow.i16(i16 undef, i16 undef)
; CHECK-NEXT: Cost Model: Found costs of 2 for: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.uadd.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:2 SizeLat:2 for: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.uadd.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:2 Lat:2 SizeLat:2 for: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.uadd.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:3 Lat:3 SizeLat:3 for: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.uadd.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:5 Lat:5 SizeLat:5 for: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.uadd.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
; CHECK-NEXT: Cost Model: Found costs of 3 for: %I8 = call { i8, i1 } @llvm.uadd.with.overflow.i8(i8 undef, i8 undef)
; CHECK-NEXT: Cost Model: Found costs of 2 for: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.uadd.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:2 SizeLat:2 for: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.uadd.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:2 Lat:2 SizeLat:2 for: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.uadd.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:3 Lat:3 SizeLat:3 for: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.uadd.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:5 Lat:5 SizeLat:5 for: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.uadd.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
; CHECK-NEXT: Cost Model: Found costs of RThru:0 CodeSize:1 Lat:1 SizeLat:1 for: ret i32 undef
;
%I64 = call {i64, i1} @llvm.uadd.with.overflow.i64(i64 undef, i64 undef)
@@ -153,20 +153,20 @@ define i32 @ssub(i32 %arg) {
; CHECK-LABEL: 'ssub'
; CHECK-NEXT: Cost Model: Found costs of 1 for: %I64 = call { i64, i1 } @llvm.ssub.with.overflow.i64(i64 undef, i64 undef)
; CHECK-NEXT: Cost Model: Found costs of 4 for: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.ssub.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:7 CodeSize:4 Lat:4 SizeLat:4 for: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.ssub.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:13 CodeSize:4 Lat:4 SizeLat:4 for: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.ssub.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:7 CodeSize:6 Lat:6 SizeLat:6 for: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.ssub.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:13 CodeSize:10 Lat:10 SizeLat:10 for: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.ssub.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
; CHECK-NEXT: Cost Model: Found costs of 1 for: %I32 = call { i32, i1 } @llvm.ssub.with.overflow.i32(i32 undef, i32 undef)
; CHECK-NEXT: Cost Model: Found costs of 4 for: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.ssub.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:7 CodeSize:4 Lat:4 SizeLat:4 for: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.ssub.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:13 CodeSize:4 Lat:4 SizeLat:4 for: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.ssub.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:7 CodeSize:6 Lat:6 SizeLat:6 for: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.ssub.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:13 CodeSize:10 Lat:10 SizeLat:10 for: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.ssub.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
; CHECK-NEXT: Cost Model: Found costs of 3 for: %I16 = call { i16, i1 } @llvm.ssub.with.overflow.i16(i16 undef, i16 undef)
; CHECK-NEXT: Cost Model: Found costs of 4 for: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.ssub.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:7 CodeSize:4 Lat:4 SizeLat:4 for: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.ssub.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:14 CodeSize:4 Lat:4 SizeLat:4 for: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.ssub.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:7 CodeSize:6 Lat:6 SizeLat:6 for: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.ssub.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:14 CodeSize:10 Lat:10 SizeLat:10 for: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.ssub.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
; CHECK-NEXT: Cost Model: Found costs of 3 for: %I8 = call { i8, i1 } @llvm.ssub.with.overflow.i8(i8 undef, i8 undef)
; CHECK-NEXT: Cost Model: Found costs of 4 for: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.ssub.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:4 Lat:4 SizeLat:4 for: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.ssub.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:16 CodeSize:4 Lat:4 SizeLat:4 for: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.ssub.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:6 Lat:6 SizeLat:6 for: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.ssub.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:16 CodeSize:10 Lat:10 SizeLat:10 for: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.ssub.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
; CHECK-NEXT: Cost Model: Found costs of RThru:0 CodeSize:1 Lat:1 SizeLat:1 for: ret i32 undef
;
%I64 = call {i64, i1} @llvm.ssub.with.overflow.i64(i64 undef, i64 undef)
@@ -216,20 +216,20 @@ define i32 @usub(i32 %arg) {
; CHECK-LABEL: 'usub'
; CHECK-NEXT: Cost Model: Found costs of 1 for: %I64 = call { i64, i1 } @llvm.usub.with.overflow.i64(i64 undef, i64 undef)
; CHECK-NEXT: Cost Model: Found costs of 2 for: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.usub.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:2 SizeLat:2 for: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.usub.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:2 Lat:2 SizeLat:2 for: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.usub.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:3 Lat:3 SizeLat:3 for: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.usub.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:5 Lat:5 SizeLat:5 for: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.usub.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
; CHECK-NEXT: Cost Model: Found costs of 1 for: %I32 = call { i32, i1 } @llvm.usub.with.overflow.i32(i32 undef, i32 undef)
; CHECK-NEXT: Cost Model: Found costs of 2 for: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.usub.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:2 SizeLat:2 for: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.usub.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:2 Lat:2 SizeLat:2 for: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.usub.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:3 Lat:3 SizeLat:3 for: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.usub.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:5 Lat:5 SizeLat:5 for: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.usub.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
; CHECK-NEXT: Cost Model: Found costs of 3 for: %I16 = call { i16, i1 } @llvm.usub.with.overflow.i16(i16 undef, i16 undef)
; CHECK-NEXT: Cost Model: Found costs of 2 for: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.usub.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:2 SizeLat:2 for: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.usub.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:2 Lat:2 SizeLat:2 for: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.usub.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:3 Lat:3 SizeLat:3 for: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.usub.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:5 Lat:5 SizeLat:5 for: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.usub.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
; CHECK-NEXT: Cost Model: Found costs of 3 for: %I8 = call { i8, i1 } @llvm.usub.with.overflow.i8(i8 undef, i8 undef)
; CHECK-NEXT: Cost Model: Found costs of 2 for: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.usub.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:2 SizeLat:2 for: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.usub.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:2 Lat:2 SizeLat:2 for: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.usub.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:3 Lat:3 SizeLat:3 for: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.usub.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:5 Lat:5 SizeLat:5 for: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.usub.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
; CHECK-NEXT: Cost Model: Found costs of RThru:0 CodeSize:1 Lat:1 SizeLat:1 for: ret i32 undef
;
%I64 = call {i64, i1} @llvm.usub.with.overflow.i64(i64 undef, i64 undef)
@@ -279,20 +279,20 @@ define i32 @smul(i32 %arg) {
; CHECK-LABEL: 'smul'
; CHECK-NEXT: Cost Model: Found costs of 3 for: %I64 = call { i64, i1 } @llvm.smul.with.overflow.i64(i64 undef, i64 undef)
; CHECK-NEXT: Cost Model: Found costs of RThru:34 CodeSize:8 Lat:8 SizeLat:8 for: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.smul.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:68 CodeSize:8 Lat:8 SizeLat:8 for: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.smul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:136 CodeSize:8 Lat:8 SizeLat:8 for: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.smul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:68 CodeSize:9 Lat:9 SizeLat:9 for: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.smul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:136 CodeSize:11 Lat:11 SizeLat:11 for: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.smul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
; CHECK-NEXT: Cost Model: Found costs of 2 for: %I32 = call { i32, i1 } @llvm.smul.with.overflow.i32(i32 undef, i32 undef)
; CHECK-NEXT: Cost Model: Found costs of RThru:38 CodeSize:8 Lat:8 SizeLat:8 for: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.smul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:76 CodeSize:8 Lat:8 SizeLat:8 for: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.smul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:152 CodeSize:8 Lat:8 SizeLat:8 for: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.smul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:76 CodeSize:9 Lat:9 SizeLat:9 for: %V8I32 = call { <8 x i32>, <8 x...
[truncated]
|
|
@llvm/pr-subscribers-backend-x86 Author: David Green (davemgreen) ChangesCurrently we always produce a cost of 1 for all CostKinds that are not RecipThroughput, which can underestimate the cost if the type has a higher legalization cost (like larger vectors). This relaxes it to cover all cost kinds. Patch is 333.01 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/148233.diff 24 Files Affected:
diff --git a/llvm/include/llvm/CodeGen/BasicTTIImpl.h b/llvm/include/llvm/CodeGen/BasicTTIImpl.h
index fefd36ec54ae2..bff9979493125 100644
--- a/llvm/include/llvm/CodeGen/BasicTTIImpl.h
+++ b/llvm/include/llvm/CodeGen/BasicTTIImpl.h
@@ -1369,8 +1369,7 @@ class BasicTTIImplBase : public TargetTransformInfoImplCRTPBase<T> {
int ISD = TLI->InstructionOpcodeToISD(Opcode);
assert(ISD && "Invalid opcode");
- // TODO: Handle other cost kinds.
- if (CostKind != TTI::TCK_RecipThroughput)
+ if (getTLI()->getValueType(DL, ValTy, true) == MVT::Other)
return BaseT::getCmpSelInstrCost(Opcode, ValTy, CondTy, VecPred, CostKind,
Op1Info, Op2Info, I);
diff --git a/llvm/test/Analysis/CostModel/AArch64/arith-overflow.ll b/llvm/test/Analysis/CostModel/AArch64/arith-overflow.ll
index 8b52b123d6d25..3fbd12ac8813c 100644
--- a/llvm/test/Analysis/CostModel/AArch64/arith-overflow.ll
+++ b/llvm/test/Analysis/CostModel/AArch64/arith-overflow.ll
@@ -27,20 +27,20 @@ define i32 @sadd(i32 %arg) {
; CHECK-LABEL: 'sadd'
; CHECK-NEXT: Cost Model: Found costs of 1 for: %I64 = call { i64, i1 } @llvm.sadd.with.overflow.i64(i64 undef, i64 undef)
; CHECK-NEXT: Cost Model: Found costs of 4 for: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.sadd.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:7 CodeSize:4 Lat:4 SizeLat:4 for: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.sadd.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:13 CodeSize:4 Lat:4 SizeLat:4 for: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.sadd.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:7 CodeSize:6 Lat:6 SizeLat:6 for: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.sadd.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:13 CodeSize:10 Lat:10 SizeLat:10 for: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.sadd.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
; CHECK-NEXT: Cost Model: Found costs of 1 for: %I32 = call { i32, i1 } @llvm.sadd.with.overflow.i32(i32 undef, i32 undef)
; CHECK-NEXT: Cost Model: Found costs of 4 for: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.sadd.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:7 CodeSize:4 Lat:4 SizeLat:4 for: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.sadd.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:13 CodeSize:4 Lat:4 SizeLat:4 for: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.sadd.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:7 CodeSize:6 Lat:6 SizeLat:6 for: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.sadd.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:13 CodeSize:10 Lat:10 SizeLat:10 for: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.sadd.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
; CHECK-NEXT: Cost Model: Found costs of 3 for: %I16 = call { i16, i1 } @llvm.sadd.with.overflow.i16(i16 undef, i16 undef)
; CHECK-NEXT: Cost Model: Found costs of 4 for: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.sadd.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:7 CodeSize:4 Lat:4 SizeLat:4 for: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.sadd.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:14 CodeSize:4 Lat:4 SizeLat:4 for: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.sadd.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:7 CodeSize:6 Lat:6 SizeLat:6 for: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.sadd.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:14 CodeSize:10 Lat:10 SizeLat:10 for: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.sadd.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
; CHECK-NEXT: Cost Model: Found costs of 3 for: %I8 = call { i8, i1 } @llvm.sadd.with.overflow.i8(i8 undef, i8 undef)
; CHECK-NEXT: Cost Model: Found costs of 4 for: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.sadd.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:4 Lat:4 SizeLat:4 for: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.sadd.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:16 CodeSize:4 Lat:4 SizeLat:4 for: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.sadd.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:6 Lat:6 SizeLat:6 for: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.sadd.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:16 CodeSize:10 Lat:10 SizeLat:10 for: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.sadd.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
; CHECK-NEXT: Cost Model: Found costs of RThru:0 CodeSize:1 Lat:1 SizeLat:1 for: ret i32 undef
;
%I64 = call {i64, i1} @llvm.sadd.with.overflow.i64(i64 undef, i64 undef)
@@ -90,20 +90,20 @@ define i32 @uadd(i32 %arg) {
; CHECK-LABEL: 'uadd'
; CHECK-NEXT: Cost Model: Found costs of 1 for: %I64 = call { i64, i1 } @llvm.uadd.with.overflow.i64(i64 undef, i64 undef)
; CHECK-NEXT: Cost Model: Found costs of 2 for: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.uadd.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:2 SizeLat:2 for: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.uadd.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:2 Lat:2 SizeLat:2 for: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.uadd.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:3 Lat:3 SizeLat:3 for: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.uadd.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:5 Lat:5 SizeLat:5 for: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.uadd.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
; CHECK-NEXT: Cost Model: Found costs of 1 for: %I32 = call { i32, i1 } @llvm.uadd.with.overflow.i32(i32 undef, i32 undef)
; CHECK-NEXT: Cost Model: Found costs of 2 for: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.uadd.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:2 SizeLat:2 for: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.uadd.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:2 Lat:2 SizeLat:2 for: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.uadd.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:3 Lat:3 SizeLat:3 for: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.uadd.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:5 Lat:5 SizeLat:5 for: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.uadd.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
; CHECK-NEXT: Cost Model: Found costs of 3 for: %I16 = call { i16, i1 } @llvm.uadd.with.overflow.i16(i16 undef, i16 undef)
; CHECK-NEXT: Cost Model: Found costs of 2 for: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.uadd.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:2 SizeLat:2 for: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.uadd.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:2 Lat:2 SizeLat:2 for: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.uadd.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:3 Lat:3 SizeLat:3 for: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.uadd.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:5 Lat:5 SizeLat:5 for: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.uadd.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
; CHECK-NEXT: Cost Model: Found costs of 3 for: %I8 = call { i8, i1 } @llvm.uadd.with.overflow.i8(i8 undef, i8 undef)
; CHECK-NEXT: Cost Model: Found costs of 2 for: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.uadd.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:2 SizeLat:2 for: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.uadd.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:2 Lat:2 SizeLat:2 for: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.uadd.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:3 Lat:3 SizeLat:3 for: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.uadd.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:5 Lat:5 SizeLat:5 for: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.uadd.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
; CHECK-NEXT: Cost Model: Found costs of RThru:0 CodeSize:1 Lat:1 SizeLat:1 for: ret i32 undef
;
%I64 = call {i64, i1} @llvm.uadd.with.overflow.i64(i64 undef, i64 undef)
@@ -153,20 +153,20 @@ define i32 @ssub(i32 %arg) {
; CHECK-LABEL: 'ssub'
; CHECK-NEXT: Cost Model: Found costs of 1 for: %I64 = call { i64, i1 } @llvm.ssub.with.overflow.i64(i64 undef, i64 undef)
; CHECK-NEXT: Cost Model: Found costs of 4 for: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.ssub.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:7 CodeSize:4 Lat:4 SizeLat:4 for: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.ssub.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:13 CodeSize:4 Lat:4 SizeLat:4 for: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.ssub.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:7 CodeSize:6 Lat:6 SizeLat:6 for: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.ssub.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:13 CodeSize:10 Lat:10 SizeLat:10 for: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.ssub.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
; CHECK-NEXT: Cost Model: Found costs of 1 for: %I32 = call { i32, i1 } @llvm.ssub.with.overflow.i32(i32 undef, i32 undef)
; CHECK-NEXT: Cost Model: Found costs of 4 for: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.ssub.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:7 CodeSize:4 Lat:4 SizeLat:4 for: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.ssub.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:13 CodeSize:4 Lat:4 SizeLat:4 for: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.ssub.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:7 CodeSize:6 Lat:6 SizeLat:6 for: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.ssub.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:13 CodeSize:10 Lat:10 SizeLat:10 for: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.ssub.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
; CHECK-NEXT: Cost Model: Found costs of 3 for: %I16 = call { i16, i1 } @llvm.ssub.with.overflow.i16(i16 undef, i16 undef)
; CHECK-NEXT: Cost Model: Found costs of 4 for: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.ssub.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:7 CodeSize:4 Lat:4 SizeLat:4 for: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.ssub.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:14 CodeSize:4 Lat:4 SizeLat:4 for: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.ssub.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:7 CodeSize:6 Lat:6 SizeLat:6 for: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.ssub.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:14 CodeSize:10 Lat:10 SizeLat:10 for: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.ssub.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
; CHECK-NEXT: Cost Model: Found costs of 3 for: %I8 = call { i8, i1 } @llvm.ssub.with.overflow.i8(i8 undef, i8 undef)
; CHECK-NEXT: Cost Model: Found costs of 4 for: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.ssub.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:4 Lat:4 SizeLat:4 for: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.ssub.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:16 CodeSize:4 Lat:4 SizeLat:4 for: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.ssub.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:6 Lat:6 SizeLat:6 for: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.ssub.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:16 CodeSize:10 Lat:10 SizeLat:10 for: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.ssub.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
; CHECK-NEXT: Cost Model: Found costs of RThru:0 CodeSize:1 Lat:1 SizeLat:1 for: ret i32 undef
;
%I64 = call {i64, i1} @llvm.ssub.with.overflow.i64(i64 undef, i64 undef)
@@ -216,20 +216,20 @@ define i32 @usub(i32 %arg) {
; CHECK-LABEL: 'usub'
; CHECK-NEXT: Cost Model: Found costs of 1 for: %I64 = call { i64, i1 } @llvm.usub.with.overflow.i64(i64 undef, i64 undef)
; CHECK-NEXT: Cost Model: Found costs of 2 for: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.usub.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:2 SizeLat:2 for: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.usub.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:2 Lat:2 SizeLat:2 for: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.usub.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:3 Lat:3 SizeLat:3 for: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.usub.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:5 Lat:5 SizeLat:5 for: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.usub.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
; CHECK-NEXT: Cost Model: Found costs of 1 for: %I32 = call { i32, i1 } @llvm.usub.with.overflow.i32(i32 undef, i32 undef)
; CHECK-NEXT: Cost Model: Found costs of 2 for: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.usub.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:2 SizeLat:2 for: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.usub.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:2 Lat:2 SizeLat:2 for: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.usub.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:3 Lat:3 SizeLat:3 for: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.usub.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:5 Lat:5 SizeLat:5 for: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.usub.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
; CHECK-NEXT: Cost Model: Found costs of 3 for: %I16 = call { i16, i1 } @llvm.usub.with.overflow.i16(i16 undef, i16 undef)
; CHECK-NEXT: Cost Model: Found costs of 2 for: %V8I16 = call { <8 x i16>, <8 x i1> } @llvm.usub.with.overflow.v8i16(<8 x i16> undef, <8 x i16> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:2 SizeLat:2 for: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.usub.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:2 Lat:2 SizeLat:2 for: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.usub.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:3 Lat:3 SizeLat:3 for: %V16I16 = call { <16 x i16>, <16 x i1> } @llvm.usub.with.overflow.v16i16(<16 x i16> undef, <16 x i16> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:5 Lat:5 SizeLat:5 for: %V32I16 = call { <32 x i16>, <32 x i1> } @llvm.usub.with.overflow.v32i16(<32 x i16> undef, <32 x i16> undef)
; CHECK-NEXT: Cost Model: Found costs of 3 for: %I8 = call { i8, i1 } @llvm.usub.with.overflow.i8(i8 undef, i8 undef)
; CHECK-NEXT: Cost Model: Found costs of 2 for: %V16I8 = call { <16 x i8>, <16 x i1> } @llvm.usub.with.overflow.v16i8(<16 x i8> undef, <16 x i8> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:2 SizeLat:2 for: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.usub.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:2 Lat:2 SizeLat:2 for: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.usub.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:3 Lat:3 SizeLat:3 for: %V32I8 = call { <32 x i8>, <32 x i1> } @llvm.usub.with.overflow.v32i8(<32 x i8> undef, <32 x i8> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:5 Lat:5 SizeLat:5 for: %V64I8 = call { <64 x i8>, <64 x i1> } @llvm.usub.with.overflow.v64i8(<64 x i8> undef, <64 x i8> undef)
; CHECK-NEXT: Cost Model: Found costs of RThru:0 CodeSize:1 Lat:1 SizeLat:1 for: ret i32 undef
;
%I64 = call {i64, i1} @llvm.usub.with.overflow.i64(i64 undef, i64 undef)
@@ -279,20 +279,20 @@ define i32 @smul(i32 %arg) {
; CHECK-LABEL: 'smul'
; CHECK-NEXT: Cost Model: Found costs of 3 for: %I64 = call { i64, i1 } @llvm.smul.with.overflow.i64(i64 undef, i64 undef)
; CHECK-NEXT: Cost Model: Found costs of RThru:34 CodeSize:8 Lat:8 SizeLat:8 for: %V2I64 = call { <2 x i64>, <2 x i1> } @llvm.smul.with.overflow.v2i64(<2 x i64> undef, <2 x i64> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:68 CodeSize:8 Lat:8 SizeLat:8 for: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.smul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:136 CodeSize:8 Lat:8 SizeLat:8 for: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.smul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:68 CodeSize:9 Lat:9 SizeLat:9 for: %V4I64 = call { <4 x i64>, <4 x i1> } @llvm.smul.with.overflow.v4i64(<4 x i64> undef, <4 x i64> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:136 CodeSize:11 Lat:11 SizeLat:11 for: %V8I64 = call { <8 x i64>, <8 x i1> } @llvm.smul.with.overflow.v8i64(<8 x i64> undef, <8 x i64> undef)
; CHECK-NEXT: Cost Model: Found costs of 2 for: %I32 = call { i32, i1 } @llvm.smul.with.overflow.i32(i32 undef, i32 undef)
; CHECK-NEXT: Cost Model: Found costs of RThru:38 CodeSize:8 Lat:8 SizeLat:8 for: %V4I32 = call { <4 x i32>, <4 x i1> } @llvm.smul.with.overflow.v4i32(<4 x i32> undef, <4 x i32> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:76 CodeSize:8 Lat:8 SizeLat:8 for: %V8I32 = call { <8 x i32>, <8 x i1> } @llvm.smul.with.overflow.v8i32(<8 x i32> undef, <8 x i32> undef)
-; CHECK-NEXT: Cost Model: Found costs of RThru:152 CodeSize:8 Lat:8 SizeLat:8 for: %V16I32 = call { <16 x i32>, <16 x i1> } @llvm.smul.with.overflow.v16i32(<16 x i32> undef, <16 x i32> undef)
+; CHECK-NEXT: Cost Model: Found costs of RThru:76 CodeSize:9 Lat:9 SizeLat:9 for: %V8I32 = call { <8 x i32>, <8 x...
[truncated]
|
RKSimon
reviewed
Jul 14, 2025
| @@ -1369,8 +1369,7 @@ class BasicTTIImplBase : public TargetTransformInfoImplCRTPBase<T> { | |||
| int ISD = TLI->InstructionOpcodeToISD(Opcode); | |||
| assert(ISD && "Invalid opcode"); | |||
|
|
|||
| // TODO: Handle other cost kinds. | |||
| if (CostKind != TTI::TCK_RecipThroughput) | |||
| if (getTLI()->getValueType(DL, ValTy, true) == MVT::Other) | |||
There was a problem hiding this comment.
where does the MVT::Other case occur?
There was a problem hiding this comment.
I also don't really understand how this correlates to the todo
There was a problem hiding this comment.
Hi. IIRC something (maybe simplifycfg) was trying to cost [2 x i64] selects. Otherwise this if would have been removed.
|
merge to trunk to see if those CI failures were intermittent? |
Currently we always produce a cost of 1 for all CostKinds that are not RecipThroughput, which can underestimate the cost if the type has a higher legalization cost (like larger vectors). This relaxes it to cover all cost kinds.
davemgreen
force-pushed
the
gh-cm-getCmpSelInstrCostAll
branch
from
386f388 to
70924e1
Compare
Updated. It seemed to run OK locally. |
RKSimon
approved these changes
Jul 14, 2025
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Currently we always produce a cost of 1 for all CostKinds that are not RecipThroughput, which can underestimate the cost if the type has a higher legalization cost (like larger vectors). This relaxes it to cover all cost kinds.