perf(pm): two-phase sibling parse + 4× channel buffer (p1 attack)#2929

Draft

elrrrrrrr wants to merge 40 commits into

nextfrom

perf/p1-sibling-two-phase

Commits on May 8, 2026

perf(pm): bump manifests-concurrency-limit 64 → 256 + add fetch breakdown

elrrrrrrr
and
claude
committed
chore(p1): revert concurrency 256 → 64 + restore manifest-bench

elrrrrrrr
and
claude
committed
ci(p1): wire manifest-bench standalone HTTP sweep into bench-phases-linux

elrrrrrrr
and
claude
committed
perf(ruborist): inline JSON parse, drop rayon::spawn dispatch

elrrrrrrr
and
claude
committed
perf(ruborist): switch JSON parse to tokio spawn_blocking

elrrrrrrr
and
claude
committed
perf(ruborist): switch extract_core_version to spawn_blocking too

elrrrrrrr
and
claude
committed
revert + instrument(ruborist): post-build phase timing

elrrrrrrr
and
claude
committed
instrument(ruborist): preload main loop dispatch + result split

elrrrrrrr
and
claude
committed
perf(pm): grow rayon pool to max(num_cpus, 8) to drain p1 extract queue

elrrrrrrr
and
claude
committed
perf(pm): skip preload for p1 path; BFS does per-level parallel prefetch

elrrrrrrr
and
claude
committed
perf(pm): fast_preload bypasses UnifiedRegistry for utoo deps path

elrrrrrrr
and
claude
committed
perf(pm): dispatch fast_preload settle to rayon to free tokio runtime

elrrrrrrr
and
claude
committed
perf(pm): bump manifests-concurrency-limit 64 → 96 (manifest-bench best)

elrrrrrrr
and
claude
committed
perf(pm): fast_preload populates (name, spec) cache slot for BFS fast path

elrrrrrrr
and
claude
committed
perf(pm): fuse primary settle into fetch task to drop dispatch RTT

elrrrrrrr
and
claude
committed
perf(pm): combined-parse fetch path eliminates per-fetch double simd_json

elrrrrrrr
and
claude
committed
perf(pm): bump manifests-concurrency-limit 96 → 128

elrrrrrrr
and
claude
committed
revert(pm): manifests-concurrency-limit back to 96

elrrrrrrr
and
claude
committed

Commits on May 9, 2026

Commits on May 11, 2026