engineering/arborist
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions
arborist/docs/tickets
History
russell@unturf.com 2ad2dceada
ticket #000039: record embed-throughput measurement + ingest-stays-fast rationale (§14.6) 
New §14.6 captures the 2026-05-11 findings:

- Measured embed rate: ~4.3 chunks/s on the contended dev box (load
  ~11 on 8 cores; the embed process got ~28% of one core). One wiki
  shard (~1.56M chunks) at that rate ≈ 100h ≈ 4+ days; all four ≈
  ~16 days. The full int8 backfill was abandoned as not feasible to
  brute-force there.
- The 54K-chunk partial on 000.db confirmed ~409 B/chunk apparent
  → 384 B amortized → ~2.4 GB for the full 6.24M-chunk corpus at
  int8 — the deterministic number a full backfill would only
  re-confirm, so finishing it bought nothing.
- On an idle healthy box (batching + all cores) bge-small does
  ~50-200 chunks/s → full corpus ≈ ~9-35h (the ticket's earlier
  "~17h" is the optimistic end).
- Per-chunk cost breakdown: bge-small ONNX inference ~5-30 ms/chunk
  dominates; the existing arborist ingest steps (chunker + SHA-256
  leaf + Merkle commit + sqlite INSERTs + zstd + FTS5) are well
  under 1 ms total. So adding vec multiplies ingest by ~10-100×,
  entirely in the ONNX matmuls — the non-vec ingest path is
  unchanged and still runs at hundreds of chunks/s.
- Implication for §14.2: this is *why* lazy-out-of-band is the
  default and `arborist ingest --embed` is the opt-in. Production
  guidance: a corpus-wide backfill is a one-time batch job (hours
  on idle / days on contended), best run off-peak or on a dedicated
  box; it does not slow ongoing ingest (which never embeds unless
  --embed is passed); a GPU/accelerated embedder is a drop-in via
  the pluggable Embedder callable if backfill latency matters.

Status line updated to point at §14.6. Doc-only.
2026-05-12 07:14:12 -04:00
..
ticket-000001-retrieval-keywords-audit-gap.md aborist/arborist 2026-05-07 09:31:49 -04:00
ticket-000002-reference-frame-polarity-contract.md aborist/arborist 2026-05-07 09:31:49 -04:00
ticket-000003-anchor-class-warrant.md aborist/arborist 2026-05-07 09:31:49 -04:00
ticket-000004-directive-coverage-in-bench.md docs: tickets/ subdir + delete naming-deferral, ref-frame, coverage-audit 2026-05-02 12:51:09 -04:00
ticket-000005-label-ladder-migration.md aborist/arborist 2026-05-07 09:31:49 -04:00
ticket-000006-bench-emergent-findings.md ticket #000035: flip φ_PRG counter big-endian → little-endian to match v7 TLV 2026-05-11 07:47:35 -04:00
ticket-000007-query-layer-hyphen-fold.md aborist/arborist 2026-05-07 09:31:49 -04:00
ticket-000008-broad-quantifier-preflight-guard.md aborist/arborist 2026-05-07 09:31:49 -04:00
ticket-000009-quantifier-preflight-dag-binding.md aborist/arborist 2026-05-07 09:31:49 -04:00
ticket-000010-metacognition-preflight-guard.md docs/tickets: stale-map sweep — 3 deferred markers refreshed 2026-05-10 07:00:56 -04:00
ticket-000011-soft-preflight-hint-sidecar.md aborist/arborist 2026-05-07 09:31:49 -04:00
ticket-000012-selection-consensus-protocol.md #000048 — fold the step-2.2 finding through the ticket + the cross-refs 2026-05-11 17:34:18 -04:00
ticket-000013-spatial-temporal-substrate.md #000013 closed: v7-W spatial-temporal substrate paper + namespace 2026-05-09 15:00:05 -04:00
ticket-000014-self-model.md selfmodel: land ticket #000014 (identity record + falsification) 2026-05-07 16:36:34 -04:00
ticket-000015-pi-star-domain-library.md pi_star: land ticket #000015 (π* domain library + composition algebra) 2026-05-07 16:51:33 -04:00
ticket-000016-zk-frontier-proof.md #000016 parked: ZK frontier-proof bench plan + wire protocol 2026-05-09 15:05:08 -04:00
ticket-000017-memory-root-lifelong-learning.md memory: land ticket #000017 (memory-root lifelong learning summary) 2026-05-07 16:46:41 -04:00
ticket-000018-soft-hash-covert-channel.md docs/tickets: open #000034 / #000035 / #000036 — #000018 follow-ups 2026-05-09 14:50:15 -04:00
ticket-000019-specification-methodology.md docs: land ticket #000019 (spec methodology for π*, V, policy fields) 2026-05-07 16:53:28 -04:00
ticket-000020-capital-cost-ledger.md capital: land ticket #000020 (8-capital-form cost ledger) 2026-05-07 16:41:48 -04:00
ticket-000021-bench-fixtures-5s-5t-5r.md docs/tickets: stale-map sweep — 3 deferred markers refreshed 2026-05-10 07:00:56 -04:00
ticket-000022-adapter-loss-report.md loss_report: land ticket #000022 (adapter LossReport sidecar) 2026-05-07 17:58:50 -04:00
ticket-000023-5s-phase-1b-syllogism-synthesis-semiotics.md bench: land #000023 + #000024 + #000025 (Phase 1a/1b — Dav1DPrometheus suite) 2026-05-07 20:14:44 -04:00
ticket-000024-5t-phase-1b-and-rename.md bench: land #000023 + #000024 + #000025 (Phase 1a/1b — Dav1DPrometheus suite) 2026-05-07 20:14:44 -04:00
ticket-000025-5f-battery.md #000025 §10.11 + §10.13 + §10.14 — close the 5F battery 2026-05-11 07:41:37 -04:00
ticket-000026-real-shard-workload-baseline.md tickets: close #000026 (Phase 3 landed 60b5748) 2026-05-10 15:54:32 -04:00
ticket-000027-canonical-projections-in-providence-cache.md #000027 + #000028: canonical projections persist; STRICT-WITNESSED reachable 2026-05-09 11:37:06 -04:00
ticket-000028-multi-modality-witness.md ticket #000028 §2.6 sketch: refresh stale TODO post-#000027 2026-05-10 07:52:54 -04:00
ticket-000029-claim-pack-source.md ticket #000029: claim-pack source for axiom/theorem JSON bundles 2026-05-09 11:44:08 -04:00
ticket-000030-math-pi-star-expansion-sympy-substrate.md fan-out: close #000030 · composition fixtures · witness end-to-end 2026-05-09 13:29:59 -04:00
ticket-000031-surface-ingest-cited-textbooks.md docs: refresh stale alias counts (residual 40/54 → 74) 2026-05-10 12:27:23 -04:00
ticket-000032-combinatorics-pi-star.md ticket #000032: combinatorics@v1 π* (pure-integer counting kernel) 2026-05-09 13:20:29 -04:00
ticket-000033-claim-pack-pillar-vii-combinatorics.md ticket #000033: pillar VII (combinatorics) live in shard 000 2026-05-09 15:42:40 -04:00
ticket-000034-hessian-alignment-phi-linear.md docs: ticket §7 status refresh (#000034 + #000036) + CLAUDE.md AUTOCOUNT rule + #000045 ref 2026-05-10 18:39:05 -04:00
ticket-000035-prg-choice-phi-prg.md #000035 closed — v7 substrate spec into the repo; § 9.10 folded in 2026-05-11 17:06:19 -04:00
ticket-000036-t3-per-window-bound.md ticket #000036: add KAT-regen tooling + close 2026-05-11 08:02:25 -04:00
ticket-000037-prometheus-sigma-recursive-falsification-controller.md #000012 Phase 1c follow-through: wire #000037 §12 Trigger 1 probe to fork_score_branches 2026-05-11 06:56:09 -04:00
ticket-000038-phase-4-content-acquisition.md docs: refresh stale citation-alias counts (40 → 54 → 74) across 4 surfaces 2026-05-10 10:41:07 -04:00
ticket-000039-sqlite-vec-optional-backend.md ticket #000039: record embed-throughput measurement + ingest-stays-fast rationale (§14.6) 2026-05-12 07:14:12 -04:00
ticket-000040-phase-5-resolver-content-tokens.md docs/tickets: stale-map sweep — 3 deferred markers refreshed 2026-05-10 07:00:56 -04:00
ticket-000041-citation-aliases-table.md docs: refresh stale citation-alias counts (40 → 54 → 74) across 4 surfaces 2026-05-10 10:41:07 -04:00
ticket-000042-term-aliases-table.md docs: 100% claim-pack warrant resolution recorded across tickets 2026-05-10 08:40:35 -04:00
ticket-000043-t3-bound-constant-tightening.md ticket #000043: open empirical-tightening sub-ticket for T3 bound constants 2026-05-10 11:55:36 -04:00
ticket-000044-autocount-doc-drift-discipline.md ticket #000035: flip φ_PRG counter big-endian → little-endian to match v7 TLV 2026-05-11 07:47:35 -04:00
ticket-000045-prometheus-sigma-phase-3-sleep-sweep.md docs: #000045 §4 sweep-swap closure + cookbook prometheus_audit 22→25 2026-05-10 19:37:20 -04:00
ticket-000046-harder-5sf-fixture-tier.md #000048 — fold the step-2.2 finding through the ticket + the cross-refs 2026-05-11 17:34:18 -04:00
ticket-000047-fork-score-delta-aggregator.md #000047 — close: delta_aggregator knob on ForkScore (Option D) 2026-05-11 08:27:38 -04:00
ticket-000048-verifier-upgrade-recombination-segmentation.md #000048 — fold the step-2.2 finding through the ticket + the cross-refs 2026-05-11 17:34:18 -04:00