maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity

[CostModel] remove cost-kind predicate for memcpy cost 

The default implementation base returns TCC_Expensive (currently
set to '4'), so that explains the test diff. This probably does
not make sense for most callers, but at least now the costs will
be consistently wrong instead of mysteriously wrong.

The ARM target has an override that tries to model codegen expansion,
and that should likely be adapted for general usage.

This probably does not affect anything because the vectorizers are
the primary users of the throughput cost, but memcpy is not listed
as a trivially vectorizable intrinsic.
This commit is contained in:
Sanjay Patel 2020-10-21 08:25:09 -04:00
parent 1e46d1aa3f
commit 01ea93d85d
2 changed files with 2 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
llvm/include/llvm/CodeGen/BasicTTIImpl.h
View File

@ -1166,10 +1166,7 @@ public:
break;
case Intrinsic::memcpy:
// FIXME: all cost kinds should default to the same thing?
if (CostKind != TTI::TCK_RecipThroughput)
return thisT()->getMemcpyCost(ICA.getInst());
return BaseT::getIntrinsicInstrCost(ICA, CostKind);
case Intrinsic::masked_scatter: {
// FIXME: all cost kinds should default to the same thing?

2
llvm/test/Analysis/CostModel/X86/intrinsic-cost-kinds.ll
View File

@ -226,7 +226,7 @@ define void @reduce_fmax(<16 x float> %va) {
define void @memcpy(i8* %a, i8* %b, i32 %c) {
; THRU-LABEL: 'memcpy'
; THRU-NEXT: Cost Model: Found an estimated cost of 1 for instruction: call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 1 %a, i8* align 1 %b, i32 32, i1 false)
; THRU-NEXT: Cost Model: Found an estimated cost of 4 for instruction: call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 1 %a, i8* align 1 %b, i32 32, i1 false)
; THRU-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; LATE-LABEL: 'memcpy'