You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Perf: consume batched zig-gc sweep accounting for object churn #114
Parent: #62\nRelated: #111\nDependency: zig-utils/zig-gc#23\n\nObject-churn profiles remain dominated by the nursery sweep loop. Consume zig-gc’s batched heap-counter accounting and validate the direct, warmed 1/2/4/8 independent, cold, and shared comparison modes with exact checksums. Acceptance: material stable improvement with no comparison/shared regression; focused and full correctness plus TSan; refresh raw reports, benchmark docs, and README scores; reference this issue from each commit.
Completed by dependency commit zig-utils/zig-gc@9d4af0d and publication commit 86d0c5b. Gates: zig-gc 37/37; zig-js 766/766 normal; TSan 765 passed, 1 expected skip, 0 failed/leaked; full 1,540-sample comparison validated exact checksums and the 50 ms floor. README, benchmark guide, raw TSV, and dated report now carry the latest scores and battery-state caveat.
Activity
docs(bench): publish batched accounting results (#114)
chrisbbreuer commentedon Jul 15, 2026
Completed by dependency commit zig-utils/zig-gc@9d4af0d and publication commit 86d0c5b. Gates: zig-gc 37/37; zig-js 766/766 normal; TSan 765 passed, 1 expected skip, 0 failed/leaked; full 1,540-sample comparison validated exact checksums and the 50 ms floor. README, benchmark guide, raw TSV, and dated report now carry the latest scores and battery-state caveat.