bench: MtCompressBenchmark - NB_WORKERS scaling on multi-job payloads#84
Merged
Conversation
Separate suite rather than an nbWorkers axis on CompressBenchmark: two of its four sizes sit below zstd's 512 KiB MT engagement minimum where MT is a guaranteed no-op, and adding an axis would break the longitudinal comparability of the single-threaded numbers (ADR 0012). Matrix: 1 MiB / 64 MiB x nbWorkers 0/2/4 at level 3, zero-copy segment path. The 1 MiB legs pin MT overhead on a single-job input (at level 3 the default job size is ~8 MiB, so no speedup is expected by design); the 64 MiB legs span multiple jobs and show the actual scaling. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Follow-up to #83; produces the numbers #80 asked for.
Separate suite rather than an
nbWorkersaxis onCompressBenchmark: two of its four sizes sit below zstd's 512 KiB MT engagement minimum where MT is a guaranteed no-op, and adding an axis would break the longitudinal comparability of the single-threaded numbers (ADR 0012).Matrix: 1 MiB / 64 MiB ×
nbWorkers0/2/4 at level 3, zero-copy segment path, context configured once in@Setup.Results (osx-aarch64, M-series, JMH 5×5s after 3×10s warmup)
Reads exactly as designed: at level 3 the default job size is ~8 MiB, so a 1 MiB input is a single job — enabling workers there buys no parallelism but pays the MT dispatch/buffering overhead (15–30% slower, and the workers=2 leg is visibly noisy). At 64 MiB the input spans multiple jobs and workers scale for real.
NB_WORKERSis a large-payload feature; the javadoc and ADR 0015 already say so, and these legs pin it with numbers.🤖 Generated with Claude Code