Repository navigation
Conversation
Replace the recursive std::variant/Cycle representation of WeakTopologicalOrdering with a single contiguous std::vector<Entry> where cycle ends store the header block and jump target index. This avoids per-cycle vector allocations and recursive evaluation in WTOWorklist::run. Benchmark results across 16 WebAssembly modules (3 iterations, interleaved): - --constraint-analysis: - Geomean: 1.606s -> 1.592s (-0.9%) - Total time: 64.05s -> 63.67s (-0.6%; faster on 11/16 modules) - --rse: - Geomean: 0.871s -> 0.852s (-2.2%) - Total time: 26.03s -> 25.39s (-2.5%; faster on 12/16 modules)
This branch has not been deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Replace the recursive std::variant/Cycle representation of
WeakTopologicalOrdering with a single contiguous std::vector
where cycle ends store the header block and jump target index. This
avoids per-cycle vector allocations and recursive evaluation in
WTOWorklist::run.
Benchmark results across 16 WebAssembly modules (3 iterations,
interleaved):