Record the Phase 27 current-stack llama.cpp n128 serving profile captured with CUDA graph node tracing and gated before and after the run.
Assisted-by: Codex:gpt-5
Record the Phase 26 current-stack paged-vs-vLLM serving snapshot with hardware classification and compact pre/post inference gates.
Assisted-by: Codex:gpt-5
Emit a compact gate_summary.tsv from current serving snapshots so each artifact records the pre/post MoE md5, dense md5, and backend op checks. Add a summary-only mode for auditing existing artifacts and document the Phase 25 backfill on the Phase 20 snapshot.
Assisted-by: Codex:gpt-5
Add a hardware report to the current serving snapshot harness so every paged-vs-vLLM artifact records the GPU identity and conservative hardware class before any server starts. Document the Phase 24 dry run and the GB10 classification for future parity comparisons.
Assisted-by: Codex:gpt-5
Update the paged parity handoff to the current fork head, patch count, mirror invariant, current serving harness, and LocalAI AI-attribution policy after Phases 20-22.
Assisted-by: Codex:gpt-5
Record the Phase 22 strict git-apply mirror check proving the LocalAI paged patch series reconstructs the canonical llama.cpp fork tree after patch 0055.
Assisted-by: Codex:gpt-5
Add a reusable current-stack paged-vs-vLLM serving snapshot harness that targets the clean DGX mirror, enforces idle/lock preflight, runs pre/post inference gates, and records ratio summaries.
Assisted-by: Codex:gpt-5
Record the Phase 20 same-session MoE paged-vs-vLLM serving snapshot on the current clean DGX mirror, including pre/post inference gates and the resulting GB10 parity decision.
Assisted-by: Codex:gpt-5
Record the Phase 19 trace-only serving entropy run and close the group/defer-by-draft follow-up based on measured shape distribution, throughput regression, and green inference gates.
Assisted-by: Codex:gpt-5
Add the next incremental llama.cpp patch for default-off speculative batch-shape tracing and record the Phase 18 red/green and inference-gate results.
Assisted-by: Codex:gpt-5
Record Phase 17 source inspection: MTP verification rows change hard graph dimensions, padding is not a safe shortcut, and any future work should start with shape instrumentation before an opt-in scheduler experiment.
Assisted-by: Codex:gpt-5
Mark the rejected-branch disposition as not taken because Phase 4 was kept as patch 0050 with recorded md5, op, perf, and mirror gates.
Assisted-by: Codex:gpt-5
Record the Phase 8 safety rerun, canonical transcript md5 gates, full and ragged MUL_MAT_ID op gates, and the no-production-patch decision for metadata-only fused dispatch work.
Assisted-by: Codex:gpt-5
Record the default-off Global-Ai32 implementation, exact md5 gates, GB10 A/B regression, rejected diff artifact, and the resulting stop decision for GDN kernel work on GB10.
Assisted-by: Codex:gpt-5
Record the shared-A/Ai GB10 cost model, the GO decision for one default-off f32 Ai prototype, and the Phase 13 implementation plan.
Assisted-by: Codex:gpt-5
Record the Phase 11 default-off QS-early GDN experiment, its canonical md5 gates, the same-session GB10 A/B regression, and the rejected diff artifact.
Assisted-by: Codex:gpt-5
Record the default-off C32 slab experiment, its md5 gates, the dense tail-row fix, and the performance regression that rejects the source patch.
Assisted-by: Codex:gpt-5
Record the Phase 10 current-M5 prefill baseline and the source inspection finding that C32 M5 needs a real U-staging implementation rather than a launcher-only shortcut.
Assisted-by: Codex:gpt-5
Record the Phase 9 MTP smoke gate, mirror the fork patch that disables backend sampling for MTP drafts, and scope the follow-up C32 slab GDN prefill phase.
Assisted-by: Codex:gpt-5
Mark the Phase 6 serving classifier complete, preserve the old parity final as historical, and scope Phase 7 source candidates with explicit md5 and op gates.
Assisted-by: Codex:gpt-5
Record both-engine serving nsys buckets, rejected sampler short-circuit, and rejected GDN/MMQ env grids for the GB10 parity work.
Assisted-by: Codex:gpt-5
Record the Phase 5 Wq shared-memory padding experiment, its gates, sub-threshold benchmark gain, and the decision to ship no 0051 patch.
Assisted-by: Codex:gpt-5
Mirror fork commit d9b9be0be as patch 0050 and record the Phase 4 W4A16 shared-memory padding gates, benchmarks, and mirror verification.
Assisted-by: Codex:gpt-5
Record the Phase 3 scale-broadcast experiment, its md5 and MUL_MAT_ID gates, the prefill regression, and the decision to ship no 0050 patch.
Assisted-by: Codex:gpt-5
Mirror fork commit 7dfa0e175 as patch 0049 and record the Phase 2 GB10 W4A16 shape sweep, md5 gates, MUL_MAT_ID checks, and mirror verification.
Assisted-by: Codex:gpt-5
Mirror the fork-first W4A16 packed tile metadata commit into the LocalAI paged patch series, record the Phase 1 benchmark result, and keep the implementation plan checked off.
Assisted-by: Codex:gpt-5
Record the clean forced W4A16 baseline, default comparison, selected metadata target, and completed plan checkpoint for the GB10 parity reopen.
Assisted-by: Codex:gpt-5
Record the clean DGX build retry, binary provenance, canonical greedy md5 gates, and completed plan steps for the GB10 parity reopen.
Assisted-by: Codex:gpt-5
Add the Superpowers implementation plan for the GB10 parity reopen, including Phase 0 provenance, decode repro, W4A16 kill gates, and later kernel workstream entry criteria.
Assisted-by: Codex:gpt-5
The opt-in hybrid per-head bf16 SSM-state lever (ssm_bf16_tau, patch 0026) is
removed from the llama-cpp-localai-paged patch series. Clean re-measurement after
the decode fusions (0028 recurrent-state gather-fusion + 0029 block-table cache)
landed shows it buys nothing: forcing ALL gated-DeltaNet heads to bf16
(tau=100000, the most aggressive setting) gives flat decode throughput, 780.6 vs
780.0 t/s. The mode engages but adds zero speed because it is subsumed by the
fusions. The earlier "+12%" was measured before the fusions completed. bf16-tau
was a precision trade (not bit-exact, ~91% same-top-p) plus extra bug surface and
extra CUDA template-instantiation compile cost with no offsetting benefit.
Dependency check: no later patch (0028/0029/0030) depends on 0026. 0030's only
mention is a description comment; its code keys off fused_gdn_ar/ch/auto_fgdn,
which originate in 0018/0019/0021 (before 0026). The remaining series (0001-0025,
0028-0030) applies clean with git apply --check against the pin
0ed235ea2c17a19fc8238668653946721ed136fd. The Makefile applies the series by glob
(patches/paged/0*.patch); the resulting gap at 0026 is tolerated (0005/0027 are
already absent).
Removed:
- patches/paged/0026-qwen35-hybrid-perhead-ssm-state.patch
- the dead ssm_bf16_tau / ssm_hybrid_tau option handler in the shared
grpc-server.cpp (it only set LLAMA_SSM_BF16_TAU, now a no-op the library no
longer reads)
- the patched+bf16-tau benchmark columns and llama-patched-bf16tau rows
(README + final_benchmark.csv), the ssm_bf16_tau option text in backend
index.yaml, the gallery NOTE block, and the docs/features/backends.md mention.
The rejected-lever lesson is kept (why it was dropped: subsumed, tau=100000 flat)
in the backend README section 5, the paged-backend agent guide, and the
vLLM-parity methodology, so it is not re-tried.
Assisted-by: Claude:opus-4.8 [Claude Code]
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
master auto-bumped the stock llama-cpp pin 9d5d882d -> 0ed235ea and updated the
shared grpc-server.cpp. The paged backend's pin must track the stock pin (the
grpc-server.cpp is shared), so bump its LLAMA_VERSION to match. All 28 paged
patches apply clean on 0ed235ea (verified against a fresh upstream clone). The
bf16-tau state-serialization fix (patch 0026) is included. Bit-exact gate + full
grpc-server build verify on GPU/CI to follow.
Assisted-by: Claude:opus-4.8 [Claude Code]
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
