Md5 hashing is expansive. Using a hash map to look up already computed GUID for dwarf names. Saw a 2% build time improvement on an internal large application.
Details
Details
Diff Detail
Diff Detail
- Repository
- rG LLVM Github Monorepo
Unit Tests
Unit Tests
Event Timeline
Comment Actions
lgtm assuming the test failure is fixed, thanks.
looks like there's indeed room for improvement by replacing md5 with xxhash etc.
llvm/lib/CodeGen/AsmPrinter/PseudoProbePrinter.cpp | ||
---|---|---|
38–40 | A comment explaining the caching/memoization as a build speed optimization would be helpful. |
Comment Actions
Updating D108722: [CSSPGO] Avoid repeatedly computing md5 hash code for pseudo probe inline contexts.
A comment explaining the caching/memoization as a build speed optimization would be helpful.