Skip to content

Commit ca78bfb

Browse files
arsenmarsenm
committed
TTI: Check legalization cost of abs nodes
1 parent 330c0e2 commit ca78bfb

File tree

2 files changed

+38
-34
lines changed

2 files changed

+38
-34
lines changed

‎llvm/include/llvm/CodeGen/BasicTTIImpl.h

Lines changed: 18 additions & 14 deletions
Original file line numberDiff line numberDiff line change
@@ -2116,20 +2116,9 @@ class BasicTTIImplBase : public TargetTransformInfoImplCRTPBase<T> {
21162116
case Intrinsic::vector_reduce_fminimum:
21172117
return thisT()->getMinMaxReductionCost(getMinMaxReductionIntrinsicOp(IID),
21182118
VecOpTy, ICA.getFlags(), CostKind);
2119-
case Intrinsic::abs: {
2120-
// abs(X) = select(icmp(X,0),X,sub(0,X))
2121-
Type *CondTy = RetTy->getWithNewBitWidth(1);
2122-
CmpInst::Predicate Pred = CmpInst::ICMP_SGT;
2123-
InstructionCost Cost = 0;
2124-
Cost += thisT()->getCmpSelInstrCost(BinaryOperator::ICmp, RetTy, CondTy,
2125-
Pred, CostKind);
2126-
Cost += thisT()->getCmpSelInstrCost(BinaryOperator::Select, RetTy, CondTy,
2127-
Pred, CostKind);
2128-
// TODO: Should we add an OperandValueProperties::OP_Zero property?
2129-
Cost += thisT()->getArithmeticInstrCost(
2130-
BinaryOperator::Sub, RetTy, CostKind, {TTI::OK_UniformConstantValue, TTI::OP_None});
2131-
return Cost;
2132-
}
2119+
case Intrinsic::abs:
2120+
ISD = ISD::ABS;
2121+
break;
21332122
case Intrinsic::smax:
21342123
ISD = ISD::SMAX;
21352124
break;
@@ -2398,6 +2387,21 @@ class BasicTTIImplBase : public TargetTransformInfoImplCRTPBase<T> {
23982387
Cost += thisT()->getArithmeticInstrCost(Instruction::Or, RetTy, CostKind);
23992388
return Cost;
24002389
}
2390+
case Intrinsic::abs: {
2391+
// abs(X) = select(icmp(X,0),X,sub(0,X))
2392+
Type *CondTy = RetTy->getWithNewBitWidth(1);
2393+
CmpInst::Predicate Pred = CmpInst::ICMP_SGT;
2394+
InstructionCost Cost = 0;
2395+
Cost += thisT()->getCmpSelInstrCost(BinaryOperator::ICmp, RetTy, CondTy,
2396+
Pred, CostKind);
2397+
Cost += thisT()->getCmpSelInstrCost(BinaryOperator::Select, RetTy, CondTy,
2398+
Pred, CostKind);
2399+
// TODO: Should we add an OperandValueProperties::OP_Zero property?
2400+
Cost += thisT()->getArithmeticInstrCost(
2401+
BinaryOperator::Sub, RetTy, CostKind,
2402+
{TTI::OK_UniformConstantValue, TTI::OP_None});
2403+
return Cost;
2404+
}
24012405
case Intrinsic::fptosi_sat:
24022406
case Intrinsic::fptoui_sat: {
24032407
if (Tys.empty())

‎llvm/test/Analysis/CostModel/AMDGPU/abs.ll

Lines changed: 20 additions & 20 deletions
Original file line numberDiff line numberDiff line change
@@ -54,11 +54,11 @@ define i32 @abs_nonpoison(i32 %arg) {
5454
; FAST-NEXT: Cost Model: Found an estimated cost of 80 for instruction: %V16I32 = call <16 x i32> @llvm.abs.v16i32(<16 x i32> undef, i1 false)
5555
; FAST-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I16 = call i16 @llvm.abs.i16(i16 undef, i1 false)
5656
; FAST-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %V2I16 = call <2 x i16> @llvm.abs.v2i16(<2 x i16> undef, i1 false)
57-
; FAST-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V4I16 = call <4 x i16> @llvm.abs.v4i16(<4 x i16> undef, i1 false)
58-
; FAST-NEXT: Cost Model: Found an estimated cost of 34 for instruction: %V8I16 = call <8 x i16> @llvm.abs.v8i16(<8 x i16> undef, i1 false)
59-
; FAST-NEXT: Cost Model: Found an estimated cost of 70 for instruction: %V16I16 = call <16 x i16> @llvm.abs.v16i16(<16 x i16> undef, i1 false)
60-
; FAST-NEXT: Cost Model: Found an estimated cost of 114 for instruction: %V17I16 = call <17 x i16> @llvm.abs.v17i16(<17 x i16> undef, i1 false)
61-
; FAST-NEXT: Cost Model: Found an estimated cost of 174 for instruction: %V32I16 = call <32 x i16> @llvm.abs.v32i16(<32 x i16> undef, i1 false)
57+
; FAST-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4I16 = call <4 x i16> @llvm.abs.v4i16(<4 x i16> undef, i1 false)
58+
; FAST-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8I16 = call <8 x i16> @llvm.abs.v8i16(<8 x i16> undef, i1 false)
59+
; FAST-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = call <16 x i16> @llvm.abs.v16i16(<16 x i16> undef, i1 false)
60+
; FAST-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V17I16 = call <17 x i16> @llvm.abs.v17i16(<17 x i16> undef, i1 false)
61+
; FAST-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V32I16 = call <32 x i16> @llvm.abs.v32i16(<32 x i16> undef, i1 false)
6262
; FAST-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I8 = call i8 @llvm.abs.i8(i8 undef, i1 false)
6363
; FAST-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V2I8 = call <2 x i8> @llvm.abs.v2i8(<2 x i8> undef, i1 false)
6464
; FAST-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %V4I8 = call <4 x i8> @llvm.abs.v4i8(<4 x i8> undef, i1 false)
@@ -112,11 +112,11 @@ define i32 @abs_nonpoison(i32 %arg) {
112112
; FAST-SIZE-NEXT: Cost Model: Found an estimated cost of 50 for instruction: %V16I32 = call <16 x i32> @llvm.abs.v16i32(<16 x i32> undef, i1 false)
113113
; FAST-SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I16 = call i16 @llvm.abs.i16(i16 undef, i1 false)
114114
; FAST-SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V2I16 = call <2 x i16> @llvm.abs.v2i16(<2 x i16> undef, i1 false)
115-
; FAST-SIZE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V4I16 = call <4 x i16> @llvm.abs.v4i16(<4 x i16> undef, i1 false)
116-
; FAST-SIZE-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V8I16 = call <8 x i16> @llvm.abs.v8i16(<8 x i16> undef, i1 false)
117-
; FAST-SIZE-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V16I16 = call <16 x i16> @llvm.abs.v16i16(<16 x i16> undef, i1 false)
118-
; FAST-SIZE-NEXT: Cost Model: Found an estimated cost of 50 for instruction: %V17I16 = call <17 x i16> @llvm.abs.v17i16(<17 x i16> undef, i1 false)
119-
; FAST-SIZE-NEXT: Cost Model: Found an estimated cost of 50 for instruction: %V32I16 = call <32 x i16> @llvm.abs.v32i16(<32 x i16> undef, i1 false)
115+
; FAST-SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4I16 = call <4 x i16> @llvm.abs.v4i16(<4 x i16> undef, i1 false)
116+
; FAST-SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8I16 = call <8 x i16> @llvm.abs.v8i16(<8 x i16> undef, i1 false)
117+
; FAST-SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = call <16 x i16> @llvm.abs.v16i16(<16 x i16> undef, i1 false)
118+
; FAST-SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V17I16 = call <17 x i16> @llvm.abs.v17i16(<17 x i16> undef, i1 false)
119+
; FAST-SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V32I16 = call <32 x i16> @llvm.abs.v32i16(<32 x i16> undef, i1 false)
120120
; FAST-SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I8 = call i8 @llvm.abs.i8(i8 undef, i1 false)
121121
; FAST-SIZE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2I8 = call <2 x i8> @llvm.abs.v2i8(<2 x i8> undef, i1 false)
122122
; FAST-SIZE-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V4I8 = call <4 x i8> @llvm.abs.v4i8(<4 x i8> undef, i1 false)
@@ -204,11 +204,11 @@ define i32 @abs_poison(i32 %arg) {
204204
; FAST-NEXT: Cost Model: Found an estimated cost of 80 for instruction: %V16I32 = call <16 x i32> @llvm.abs.v16i32(<16 x i32> undef, i1 true)
205205
; FAST-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I16 = call i16 @llvm.abs.i16(i16 undef, i1 true)
206206
; FAST-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %V2I16 = call <2 x i16> @llvm.abs.v2i16(<2 x i16> undef, i1 true)
207-
; FAST-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V4I16 = call <4 x i16> @llvm.abs.v4i16(<4 x i16> undef, i1 true)
208-
; FAST-NEXT: Cost Model: Found an estimated cost of 34 for instruction: %V8I16 = call <8 x i16> @llvm.abs.v8i16(<8 x i16> undef, i1 true)
209-
; FAST-NEXT: Cost Model: Found an estimated cost of 70 for instruction: %V16I16 = call <16 x i16> @llvm.abs.v16i16(<16 x i16> undef, i1 true)
210-
; FAST-NEXT: Cost Model: Found an estimated cost of 114 for instruction: %V17I16 = call <17 x i16> @llvm.abs.v17i16(<17 x i16> undef, i1 true)
211-
; FAST-NEXT: Cost Model: Found an estimated cost of 174 for instruction: %V32I16 = call <32 x i16> @llvm.abs.v32i16(<32 x i16> undef, i1 true)
207+
; FAST-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4I16 = call <4 x i16> @llvm.abs.v4i16(<4 x i16> undef, i1 true)
208+
; FAST-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8I16 = call <8 x i16> @llvm.abs.v8i16(<8 x i16> undef, i1 true)
209+
; FAST-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = call <16 x i16> @llvm.abs.v16i16(<16 x i16> undef, i1 true)
210+
; FAST-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V17I16 = call <17 x i16> @llvm.abs.v17i16(<17 x i16> undef, i1 true)
211+
; FAST-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V32I16 = call <32 x i16> @llvm.abs.v32i16(<32 x i16> undef, i1 true)
212212
; FAST-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I8 = call i8 @llvm.abs.i8(i8 undef, i1 true)
213213
; FAST-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V2I8 = call <2 x i8> @llvm.abs.v2i8(<2 x i8> undef, i1 true)
214214
; FAST-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %V4I8 = call <4 x i8> @llvm.abs.v4i8(<4 x i8> undef, i1 true)
@@ -262,11 +262,11 @@ define i32 @abs_poison(i32 %arg) {
262262
; FAST-SIZE-NEXT: Cost Model: Found an estimated cost of 50 for instruction: %V16I32 = call <16 x i32> @llvm.abs.v16i32(<16 x i32> undef, i1 true)
263263
; FAST-SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I16 = call i16 @llvm.abs.i16(i16 undef, i1 true)
264264
; FAST-SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %V2I16 = call <2 x i16> @llvm.abs.v2i16(<2 x i16> undef, i1 true)
265-
; FAST-SIZE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V4I16 = call <4 x i16> @llvm.abs.v4i16(<4 x i16> undef, i1 true)
266-
; FAST-SIZE-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V8I16 = call <8 x i16> @llvm.abs.v8i16(<8 x i16> undef, i1 true)
267-
; FAST-SIZE-NEXT: Cost Model: Found an estimated cost of 10 for instruction: %V16I16 = call <16 x i16> @llvm.abs.v16i16(<16 x i16> undef, i1 true)
268-
; FAST-SIZE-NEXT: Cost Model: Found an estimated cost of 50 for instruction: %V17I16 = call <17 x i16> @llvm.abs.v17i16(<17 x i16> undef, i1 true)
269-
; FAST-SIZE-NEXT: Cost Model: Found an estimated cost of 50 for instruction: %V32I16 = call <32 x i16> @llvm.abs.v32i16(<32 x i16> undef, i1 true)
265+
; FAST-SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4I16 = call <4 x i16> @llvm.abs.v4i16(<4 x i16> undef, i1 true)
266+
; FAST-SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8I16 = call <8 x i16> @llvm.abs.v8i16(<8 x i16> undef, i1 true)
267+
; FAST-SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = call <16 x i16> @llvm.abs.v16i16(<16 x i16> undef, i1 true)
268+
; FAST-SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V17I16 = call <17 x i16> @llvm.abs.v17i16(<17 x i16> undef, i1 true)
269+
; FAST-SIZE-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V32I16 = call <32 x i16> @llvm.abs.v32i16(<32 x i16> undef, i1 true)
270270
; FAST-SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %I8 = call i8 @llvm.abs.i8(i8 undef, i1 true)
271271
; FAST-SIZE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2I8 = call <2 x i8> @llvm.abs.v2i8(<2 x i8> undef, i1 true)
272272
; FAST-SIZE-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %V4I8 = call <4 x i8> @llvm.abs.v4i8(<4 x i8> undef, i1 true)

0 commit comments

Comments
 (0)