bench: parallel customization scaling ladder + conclusions - #4
Merged
Conversation
Add an opt-in large-grid customize bench (CCH_BENCH_SIDE selects sizes; builds each CCH once and times customize under a 1-thread vs all-cores rayon pool). Measured on 18 cores: ~1.0x at 16k nodes, 1.85x at 65k, 2.79x at 656k — parallel customization is a large-graph win, sub-linear in cores, no small-graph regression. Full method/numbers in docs/customize-parallel-scaling.md; README performance section updated with the measured scaling.
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Keeps and extends the customize benchmark with a large-grid parallel-scaling ladder, and records the conclusions.
Measured (Apple Silicon, 18 cores; nested-dissection-ordered grids;
customizetimed at 1 thread vs all cores):6.6M runs but is memory-bound (~12.5 GB, swapping) on this box; 66M (~125 GB) is infeasible here — a true continental figure needs a bigger-memory host or the real road corpus.
Conclusions (full writeup in
docs/customize-parallel-scaling.md): parallel customization is a real but modest, large-graph win — negligible below ~tens of thousands of nodes, ~2–3× at hundreds of thousands, sub-linear in cores (level-synchronized barrier + sparse top of hierarchy), and no small-graph regression. Grids are a pessimistic proxy for road networks.The bench is opt-in (
CCH_BENCH_SIDE=810 cargo bench --bench cch -- customize_large) so plaincargo benchstays fast. README performance section updated with the measured numbers.