[Wall time +0.01% vs main; mixed repeat directions] Index macro arms lazily by their first literal token - #157
Draft
xmakro wants to merge 1 commit into
Draft
Conversation
xmakro
changed the title
xmakro
force-pushed
the
perf-review-20260907/43-external-module-negative-cache
branch
from
September 9, 2026 06:21
3ceaae5 to
5e03967
Compare
xmakro
force-pushed
the
perf-review-20260907/44-macro-first-token-rule-index-fixed
branch
from
September 9, 2026 06:21
b9439c5 to
0df282c
Compare
xmakro
changed the title
xmakro
changed the base branch from
perf-review-20260907/43-external-module-negative-cache
to
main
September 9, 2026 06:22
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Extend the first-token filter with an ordered lazy rule index, retaining fallback arms, original precedence and complete diagnostic replay.
Negative percentages mean less elapsed time or fewer resources. The title is an individual local timing estimate; small changes may be measurement noise.
The two recorded repetitions give aggregate timing estimates of +0.077% and -0.053% for the title comparison. Mixed directions are marked in the title. This repeatability check is not a confidence interval.
Both compilers use the same source base, CI LLVM, fixed standard libraries, ThinLTO and jemalloc. Each has separate nine-workload Rust PGO training. The stock collector measures all 20 Primary benchmarks: 249 Check/Debug/Opt and full/incremental cases, two observations per compiler per case, alternating baseline/candidate order by crate. No other campaign builds run during measurement. Percentages are equal-case geometric means. LLVM retraining and BOLT are omitted.
This branch has one commit directly on main. Its exact measured source passed the full UI and incremental suites plus the additional suites recorded in the build provenance.
Required support copied without the other optimizations: Tracker flag and token equality visibility from PR 141; no scalar arm filter.
Fixed main commit:
da47efd27287edb02eed0b5a178a165c8f258e37. Measured source patch SHA-256:3494435518fdcb7c2fc39af7bedb3d47fc14d0cd3cf4d616837e10d29dd46e39.Local raw measurements and per-case CSVs:
independent-pr157-main-v1-comparison.csvin the September 6 investigation directory. Build/test provenance:pgo/independent-pr157-source-timing-v1/provenance.json.