|
977 | 977 | {"id":"bd-2g7oyh.396","title":"[perf][memcmp] Close current memcmp_4096 equal-buffer gap with codegen-backed safe-Rust primitive","description":"Focused RCH gate on vmi1227854 at HEAD 7f1fdb0ed: glibc_baseline_memcmp_4096 FrankenLibC p50/mean 51.049/53.745 ns vs host glibc 37.247/39.734 ns (1.37x p50, 1.35x mean). Prior no-ship families include folded-panel widening, exact-size certificates, rank/select, broadword extraction, cross-crate inline, and chunk-cursor retuning; any source change must be one materially different codegen/disassembly-backed safe-Rust primitive with memcmp ordering/golden proof and same-worker post.","status":"closed","priority":1,"issue_type":"task","assignee":"BoldFalcon","created_at":"2026-06-14T02:36:32.074861421Z","created_by":"ubuntu","updated_at":"2026-06-14T02:43:09.305648364Z","closed_at":"2026-06-14T02:43:09.305544998Z","close_reason":"NO-CODE REJECTED: focused RCH vmi1227854 memcmp_4096 gap reproduced (FL p50/mean 51.049/53.745 ns vs host 37.247/39.734 ns), but codegen diagnostic showed the only immediate source lever would repeat documented rejected XOR/folded-loop families. Artifact: tests/artifacts/perf/bd-2g7oyh.396-memcmp-4096-codegen.md. Route next to allocator slab/hot-list primitive after fresh focused profile.","source_repo":"frankenlibc","source_repo_path":"/data/projects/frankenlibc","compaction_level":0,"original_size":0,"labels":["memcmp","no-gaps","perf","string"],"dependencies":[{"issue_id":"bd-2g7oyh.396","depends_on_id":"bd-2g7oyh","type":"parent-child","created_at":"2026-06-14T02:36:32.074861421Z","created_by":"ubuntu","metadata":"{}","thread_id":""}]} |
978 | 978 | {"id":"bd-2g7oyh.397","title":"[perf][malloc] Focus allocator slab/hot-list residual after memcmp no-ship","description":"Continuation after bd-2g7oyh.396 memcmp no-code rejection. Current route is allocator only if a fresh RCH focused gate reproduces a material malloc_free_64/256 median+mean gap. Prior no-ship families include hot-slot metadata, exact-size/cache shortcuts, Trace lifecycle gates, fixed magazine/plain storage swaps, certificate/log micro-specialization, lazy-accounting retunes, and bitmap-only hot-slot representation. Candidate family if focused gap reproduces: one deeper safe-Rust segregated slab or intrusive hot-list primitive preserving allocation/free ordering, hot-slot LIFO, central-bin/elimination tie-breaking, active_count/total_allocated, lifecycle golden SHA, FP/RNG N/A. Close no-code if gap collapses.","status":"closed","priority":1,"issue_type":"task","assignee":"BoldFalcon","created_at":"2026-06-14T02:44:56.411912712Z","created_by":"ubuntu","updated_at":"2026-06-14T02:55:49.290076589Z","closed_at":"2026-06-14T02:55:49.289991949Z","close_reason":"NO-CODE REJECTED: focused RCH vmi1227854 allocator gate reproduced malloc_free_256 FL p50/mean 5.041/6.476 ns vs host 3.689/4.614 ns and malloc_free_64 FL p50/mean 4.279/5.977 ns vs host 3.422/5.253 ns, with large guard faster than host. No source edit: the one-live benchmark only exercises the already-mined hot-slot cycle after warm-up; deeper slab/list storage is not in the measured path, and exact hot-slot families are documented no-ships. Artifact tests/artifacts/perf/bd-2g7oyh.397-malloc-slab-hotlist-focused.md. Route next to a multi-object allocator gate if allocator returns, otherwise next reproduced non-allocator residual.","source_repo":"frankenlibc","source_repo_path":"/data/projects/frankenlibc","compaction_level":0,"original_size":0,"labels":["malloc","no-gaps","perf"],"dependencies":[{"issue_id":"bd-2g7oyh.397","depends_on_id":"bd-2g7oyh","type":"parent-child","created_at":"2026-06-14T02:44:56.411912712Z","created_by":"ubuntu","metadata":"{}","thread_id":""}]} |
979 | 979 | {"id":"bd-2g7oyh.398","title":"[perf][math] Focus exp10f residual after allocator no-code","description":"Continuation after bd-2g7oyh.397 allocator no-code closeout. Current broad RCH routing table showed exp10f as the largest reproduced math residual on current head (FrankenLibC roughly 681/761 ns p50/mean vs host roughly 476/496 ns), but broad profiles are routing evidence only. First action: run a fresh focused RCH crate-scoped criterion gate for glibc_baseline_math/exp10f before any source edit. If the focused gap reproduces, use one fundamentally different safe-Rust f32-native generated minimax/table/range-reduction primitive, not the rejected exp2f(x*log2(10)) correction family and not a nearby all-domain f64 widening. Preserve integer exactness, special-case ordering, fallback behavior, <=4 ULP ABI differential, golden SHA, FP ordering, RNG N/A. Keep only Score>=2.0; otherwise close no-code and reprofile.","status":"closed","priority":1,"issue_type":"task","assignee":"BoldFalcon","created_at":"2026-06-14T02:58:36.446968015Z","created_by":"ubuntu","updated_at":"2026-06-14T03:37:12.780807289Z","closed_at":"2026-06-14T03:37:12.780721937Z","close_reason":"NO-CODE REJECTED: focused RCH vmi1227854 exp10f gate at 0655fbda9 showed FrankenLibC already faster than host glibc: FL p50/mean 283.178/287.507 ns vs host 321.977/326.649 ns. The broad routing signal collapsed/reversed, so no source lever was kept; temporary exact-grid candidate was restored. Artifact tests/artifacts/perf/bd-2g7oyh.398-exp10f-focused-no-code.md. Route next to the current strcpy_4096 focused gate bd-2g7oyh.399.","source_repo":"frankenlibc","source_repo_path":"/data/projects/frankenlibc","compaction_level":0,"original_size":0,"labels":["exp10f","math","no-gaps","perf"],"dependencies":[{"issue_id":"bd-2g7oyh.398","depends_on_id":"bd-2g7oyh","type":"parent-child","created_at":"2026-06-14T02:58:36.446968015Z","created_by":"ubuntu","metadata":"{}","thread_id":""}]} |
980 | | -{"id":"bd-2g7oyh.399","title":"[perf][strcpy] Focus current strcpy_4096 residual with generated-store gate","description":"Current-head broad RCH profile on vmi1227854 after bd-2g7oyh.397 showed glibc_baseline_strcpy_4096 as the strongest plausible residual: FrankenLibC p50/mean 58.389/74.316 ns vs host 35.047/39.395 ns. Prior no-retry families include final-block rank-select, per-certified-block copy_from_slice lowering, whole-string scan-then-bulk-copy, generic fused scan/store retunes, and the already-kept exact-size unrolled path. First action: focused same-worker RCH baseline for strcpy_4096. Only attempt source if a material p50+mean gap reproduces and the lever is a materially different generated-code/bounded-store primitive preserving first-NUL ordering, tail no-overwrite, panic/no-NUL contract, golden SHA, FP/RNG N/A. Close no-code if focused gate collapses.","status":"in_progress","priority":1,"issue_type":"task","assignee":"BoldFalcon","created_at":"2026-06-14T03:12:04.708683739Z","created_by":"ubuntu","updated_at":"2026-06-14T03:12:04.708683739Z","source_repo":"frankenlibc","source_repo_path":"/data/projects/frankenlibc","compaction_level":0,"original_size":0,"labels":["no-gaps","perf","strcpy","string"],"dependencies":[{"issue_id":"bd-2g7oyh.399","depends_on_id":"bd-2g7oyh","type":"parent-child","created_at":"2026-06-14T03:12:04.708683739Z","created_by":"ubuntu","metadata":"{}","thread_id":""}]} |
| 980 | +{"id":"bd-2g7oyh.399","title":"[perf][strcpy] Focus current strcpy_4096 residual with generated-store gate","description":"Current-head broad RCH profile on vmi1227854 after bd-2g7oyh.397 showed glibc_baseline_strcpy_4096 as the strongest plausible residual: FrankenLibC p50/mean 58.389/74.316 ns vs host 35.047/39.395 ns. Prior no-retry families include final-block rank-select, per-certified-block copy_from_slice lowering, whole-string scan-then-bulk-copy, generic fused scan/store retunes, and the already-kept exact-size unrolled path. First action: focused same-worker RCH baseline for strcpy_4096. Only attempt source if a material p50+mean gap reproduces and the lever is a materially different generated-code/bounded-store primitive preserving first-NUL ordering, tail no-overwrite, panic/no-NUL contract, golden SHA, FP/RNG N/A. Close no-code if focused gate collapses.","status":"closed","priority":1,"issue_type":"task","assignee":"BoldFalcon","created_at":"2026-06-14T03:12:04.708683739Z","created_by":"ubuntu","updated_at":"2026-06-14T03:54:34.690875869Z","closed_at":"2026-06-14T03:54:34.690713882Z","close_reason":"REJECTED-RESTORED: focused RCH vmi1227854 baseline reproduced strcpy_4096 gap (FL p50/mean 52.608/70.364 ns vs host 45.830/47.842 ns). Tested one uniform-source certificate/store-shape lever; proof passed 8/8 filtered strcpy tests with golden SHA fe05ef410f204902cd5f53586645647b8ce5db87e49b840752b24d2b11995401, but post regressed p50 52.608->55.539 ns and widened vs-host gap. Source restored; artifact tests/artifacts/perf/bd-2g7oyh.399-strcpy-4096-focused.md.","source_repo":"frankenlibc","source_repo_path":"/data/projects/frankenlibc","compaction_level":0,"original_size":0,"labels":["no-gaps","perf","strcpy","string"],"dependencies":[{"issue_id":"bd-2g7oyh.399","depends_on_id":"bd-2g7oyh","type":"parent-child","created_at":"2026-06-14T03:12:04.708683739Z","created_by":"ubuntu","metadata":"{}","thread_id":""}]} |
981 | 981 | {"id":"bd-2g7oyh.4","title":"[perf] strlen safe-SIMD 32B NUL scan","description":"PROFILE-BACKED TARGET from RCH string reprofile on vmi1156319 after commit 5c13048b (2026-06-02): strlen p50 raw 16=4.907ns, 64=12.828ns, 256=37.688ns, 1024=155.556ns, 4096=542.733ns. strlen still scans NUL with 8-byte SWAR words; strrchr now spends a large share in strlen before the reverse SIMD pass.\\n\\nLever: one code lever only - add a safe std::simd Simd<u8,32> NUL probe inside strlen before the existing SWAR/scalar exact-resolution path. Candidate panels break to the existing word/scalar resolver, preserving first-NUL ordering and unterminated-slice behavior.\\n\\nAlien primitive recommendation card: contiguous SIMD panel scanning / vectorized execution over byte control planes. EV estimate: Impact 4 x Confidence 4 x Reuse 2 / (Effort 1 x Friction 1) = 32. Fallback trigger: revert if golden string-test sha256 changes, if first-NUL/unterminated behavior changes, or if RCH post p50 does not improve at 1024/4096.\\n\\nArtifact proof obligations: RCH behavior baseline first, deterministic golden-output sha256 for string::str tests, isomorphism proof for first NUL index and strlen-fed strrchr ordering, no FP/RNG impact, RCH rebench/check/test/clippy before close.","status":"closed","priority":2,"issue_type":"task","assignee":"BlackThrush","created_at":"2026-06-02T19:24:06.510955626Z","created_by":"ubuntu","updated_at":"2026-06-02T20:05:22.626645607Z","closed_at":"2026-06-02T20:05:22.626547953Z","close_reason":"Superseded duplicate target. The same profile-backed strlen safe-SIMD lever shipped and closed as bd-2g7oyh.5 in commit 29c316a8 with tracker close a6131e18: same-worker RCH p50 64B 11.911->5.528ns, 256B 41.121->11.063ns, 1024B 164.438->37.864ns, 4096B 515.187->157.133ns; existing string-test sha256 unchanged 235e4da93b185ebe3f01964dd290d33f97b492e083f8bd5e4fbc39208023a507.","source_repo":"frankenlibc","source_repo_path":"/data/projects/frankenlibc","compaction_level":0,"original_size":0,"labels":["perf","simd","string"],"dependencies":[{"issue_id":"bd-2g7oyh.4","depends_on_id":"bd-2g7oyh","type":"parent-child","created_at":"2026-06-02T19:24:06.510955626Z","created_by":"ubuntu","metadata":"{}","thread_id":""}]} |
982 | 982 | {"id":"bd-2g7oyh.40","title":"[perf] memset 4096 safe-Rust fill kernel","description":"Target: profile-backed memset_4096 gap outside BlackThrush str.rs claim. Fresh broad RCH profile after bd-2g7oyh.37 showed FrankenLibC memset_4096 p50 19.526ns vs host glibc 17.094ns (mean 20.996 vs 18.493).\\n\\nPrimitive: /alien-graveyard symptom match to vectorized hot operators/cache-local panel work; /alien-artifact-coding proof contract is exact prefix-fill isomorphism with no ordering, FP, RNG, errno, or tie-break effects.\\n\\nMethod: focused RCH baseline first, one lever only in crates/frankenlibc-core/src/string/mem.rs if Score>=2.0 is plausible, golden SHA behavior proof, RCH post bench, close rejected or completed with before/after numbers.","status":"closed","priority":1,"issue_type":"task","assignee":"MossyFern","created_at":"2026-06-03T12:29:57.244556269Z","created_by":"MossyFern","updated_at":"2026-06-03T12:36:11.691427653Z","closed_at":"2026-06-03T12:36:11.691330509Z","close_reason":"Rejected/no-op: focused RCH memset_4096 baseline on vmi1264463 showed FrankenLibC faster than host (p50 58.740ns vs 75.130ns; mean 102.433ns vs 131.449ns), so no profile-backed vs-upstream gap remained and no source lever was applied.","source_repo":"frankenlibc","source_repo_path":"/data/projects/frankenlibc","compaction_level":0,"original_size":0,"labels":["no-gaps","perf"],"dependencies":[{"issue_id":"bd-2g7oyh.40","depends_on_id":"bd-2g7oyh","type":"parent-child","created_at":"2026-06-03T12:29:57.244556269Z","created_by":"MossyFern","metadata":"{}","thread_id":""}]} |
983 | 983 | {"id":"bd-2g7oyh.41","title":"[perf] malloc free external elimination handle gate","description":"Target: fresh RCH broad profile after 27e332a7 on vmi1227854 showed malloc_free_64 FrankenLibC p50 163.665ns p95 221.719ns p99 328.539ns mean 174.630ns vs host glibc p50 4.810ns p95 10.000ns p99 40.000ns mean 7.049ns.\\n\\nOne lever: in non-test builds, avoid the per-free Arc::strong_count(&self.elimination) check because the elimination Arc is private and no external handle can exist outside cfg(test); keep the exact strong_count gate under cfg(test) so elimination handoff tests and observable test behavior are unchanged. No edits to peer-dirty malloc/elimination.rs.\\n\\nAlien primitive: hot/cold capability separation from vectorized/control-plane optimization discipline: move an impossible coordination check out of the production hot path while preserving proof-carrying test surface.\\n\\nRequired proof: focused RCH malloc baseline before edit; golden allocator lifecycle SHA before/post; ordering/tie-breaking proof for allocator path selection, thread-cache LIFO, central-bin order, elimination handoff in tests, FP/RNG unaffected; RCH post bench and validation; keep only Score>=2.0 else restore.","status":"closed","priority":1,"issue_type":"task","assignee":"MossyFern","created_at":"2026-06-03T12:41:31.288703150Z","created_by":"MossyFern","updated_at":"2026-06-03T12:55:01.692315708Z","closed_at":"2026-06-03T12:55:01.692219165Z","close_reason":"Rejected: production elimination handle gate failed real-win keep gate. Pre RCH malloc_free_64 FL p50 156.635ns mean 154.100 vs host p50 4.139ns mean 5.890; post RCH FL p50 156.085ns mean 157.231 vs host p50 3.661ns mean 5.036. Golden lifecycle SHA unchanged: 01df8806e2bfd0fda041e153ec61ec4737ad2d3cb1ce22050a2e35bab1688455. No source change retained; artifact tests/artifacts/perf/bd-2g7oyh.41-malloc-elimination-handle-gate.md.","source_repo":"frankenlibc","source_repo_path":"/data/projects/frankenlibc","compaction_level":0,"original_size":0,"labels":["malloc","no-gaps","perf"],"dependencies":[{"issue_id":"bd-2g7oyh.41","depends_on_id":"bd-2g7oyh","type":"parent-child","created_at":"2026-06-03T12:41:31.288703150Z","created_by":"MossyFern","metadata":"{}","thread_id":""}]} |
|
0 commit comments