BOLT YAML profile reading time gets marginally faster (14.1572->13.9207 s) for
a large YAML profile (121MB/31K functions). Not claiming stat significance
though.
Details
Details
Diff Detail
Diff Detail
- Repository
- rG LLVM Github Monorepo
Event Timeline
Comment Actions
@njames93, @silvas, @grimar, @thomasfinch, @Bigcheese, @hintonda – folks, I checked blame on this function and picked you as most recent authors and reviewers.
I'm also looking for suggestions on how to speed up this function and parsing YAML in general. First thing that stands out in the profile is that we do allocate a lot of Nodes. But since they're allocated using BumpPtrAllocator I don't see much room for improvement here:
https://github.com/llvm/llvm-project/blob/49dfbc6efc64bcd9031bc80fd20f4be080ebf56e/llvm/include/llvm/Support/YAMLParser.h#L141