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

[SamplePGO] Add switch to honor zero count on block level as accurate 

Add a new LLVM switch `-profile-sample-block-accurate` to trust zero block counts for branches. Currently we leave out such zero counts when annotating branch weight metadata, which would lead to weights being considered as unknown.

Differential Revision: https://reviews.llvm.org/D110117
This commit is contained in:
Wenlei He 2021-09-20 17:35:36 -07:00
parent f8b1cc3657
commit 5f187f0afa
1 changed files with 7 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
llvm/lib/Transforms/IPO/SampleProfile.cpp
View File

@ -143,6 +143,12 @@ static cl::opt<bool> ProfileSampleAccurate(
"callsite and function as having 0 samples. Otherwise, treat "
"un-sampled callsites and functions conservatively as unknown. "));
static cl::opt<bool> ProfileSampleBlockAccurate(
"profile-sample-block-accurate", cl::Hidden, cl::init(false),
cl::desc("If the sample profile is accurate, we will mark all un-sampled "
"branches and calls as having 0 samples. Otherwise, treat "
"them conservatively as unknown. "));
static cl::opt<bool> ProfileAccurateForSymsInList(
"profile-accurate-for-symsinlist", cl::Hidden, cl::ZeroOrMore,
cl::init(true),
@ -1529,7 +1535,7 @@ void SampleProfileLoader::generateMDProfMetadata(Function &F) {
{static_cast<uint32_t>(BlockWeights[BB])}));
}
}
} else if (OverwriteExistingWeights) {
} else if (OverwriteExistingWeights || ProfileSampleBlockAccurate) {
// Set profile metadata (possibly annotated by LTO prelink) to zero or
// clear it for cold code.
for (auto &I : BB->getInstList()) {