NOW
August 17, 2026 · View on GitHub
Snapshot, not log. History is git; evidence: parity ledger, and benchmarks. Budget: 100 lines / 6,000 characters.
Live claims
Rendered on demand: scripts/now.py. It assembles every SPIKE/ACTIVE
row, its claim and PR, and the row's own next step from the matrices,
.agents/claims/, and each row spec's ## Now.
They are NOT listed here any more (ENG-NOW-DERIVED, #374). A per-row table in
this file made it a surface every row-advancing PR had to write, which is a lock
under AGENTS.md §Records; it conflicted in 5 of the 16 conflicting open PRs
measured at d928e2c3. What remains below is authored at operator cadence, so
no per-row change needs to touch this file at all.
Current gate
Token-exact (or ratified distributional) vs pinned vLLM; ≥ throughput and ≤
latency/memory on every axis, both gate models, reproduced 2–3x idle. See
verification. Pin: vLLM 555967922 (0.26.0.dev0).
Next actions
- 35B mid-band: first lever LANDED (+1.31% c8, +1.38% c4). The fused
shared gate_up sink still took the MoE-marlin route (20320 launches = 5.4%
GPU);
VT_MARLIN_DENSE_PAIRON. Second lever LANDED: shared down-proj emits bf16, +2.05% BIT-IDENTICAL. SiLU spec NEGATIVE: the 9.2x was a MEAN over a bimodal kernel (min 1.34/max 979us); decode SiLU already beats vLLM's. ~5% UNATTRIBUTED; needs decode-only, 1 tool. - 27B NVFP4 0.72x -> 0.85x (FP8 tower native). Next: NVFP4 MLP marlin, 68% of roof. Dense-marlin +0.5%; Triton-AOT GDN a WASH.
- Spike the Parakeet encoder row (vLLM:
nano_nemotron_vl.py; the transducer half is NOT in vLLM: separate call). - Qwen3.5-4B #206: +2.83%
PENDING; latency/VRAM open. - Invocation-parity prevention: CI guard + checklist; build-verify
kGemvHeuristicAlgoson dgx. - Restore
local-ai-workeron dgx at campaign end (--restart=always). - Protocol substrate — partly done. Triage/audit +
STATUS.mdratchet +AGENTS.mdtiering DONE. REMAINING: anchor backfill (6 model rows need a DECISION); record-era rollover BLOCKED onDONErows bound toparity-ledger.mdLINE anchors (re-anchor by ROW ID).
Operator/helper protocol (spec): roles are a coordinator
record or worktree+PR; helpers claim row/<ROW-ID> with a DRAFT PR. Role gates
ENFORCE agent-start.py → claim → preflight. Review FAIL loops to a fresh
implementer until PASS. Queue: 10 rows; backfill 79, 30 anchored.
Upstream inventory (spec):
SM060/061/070 below vLLM's floor = OUT-OF-SCOPE; COMP-/DISTRIBUTED- are REAL
unported work; every arch in the pinned registry has a row (the count lives in
model-matrix.md, not here — #622); llama.cpp's 11 extra
devices IN SCOPE (ROAD-V1-D6).
Protocol invariants that bite most often
- Every commit carries
FOLLOWING_AGENTS_PROTOCOL+Assisted-by:; neverCo-Authored-ByorSigned-off-byfrom AI. - Three MUST-route seams: fusion, merged-GEMM, born-on-the-runner decode. Not routing is drift; allowlist consciously or fold.
- Mirror vLLM; never ask how a feature should behave.
nsysBOTH sides, SAME tool, before any perf claim; cross-tool comparisons never establish invocation parity; whole-run sums mix prefill.- GPU: claim a lease with
rc runorrc hold. Neversshto a box, because that makes the fleet report it free while you are on it. The flock now lives INSIDE the lease (environment). Single-load steady-state, never reload per rep, named tmux. - Never weaken a checker to pass; repair the record.
- Work happens in its own worktree on a task branch; the shared checkout stays
clean on
main, never a work surface. Land viarow/*PR or authorized local merge; remove the worktree after.