THE PIN MOVED UNDER TWO OF THESE ROWS. The parity pin
advanced to a7c23ac96d on 2026-09-26
(#2817). The MTP row and
the DFlash row carry a vLLM denominator measured against the PREVIOUS pin
555967922 with FlashInfer 0.6.15.post1, which is the oracle's attention
backend on that path and moves to 0.6.18 at the new pin. Both owe a
re-measurement (#2818), and the
DFlash re-take must record the oracle's SELECTED backend rather than assume it
carries over. The advance preceded the re-measurement by a developer ruling that
inverts the sync cycle's step order; a red there requires reverting the pin.
Nothing here is withdrawn today.
| Speculator | Model | Result | Status |
|---|---|---|---|
| MTP | Qwen3.6-27B NVFP4 | token-identical to vLLM MTP, ~4% faster at c1; on-par at c2-c8 | DONE |
| DFlash | Qwen3.6-27B NVFP4 | 2.9x over spec-off (10.16 → 29.32 tok/s), at/above vLLM DFlash-on (1.003x, non-overlapping bands) | DONE |
| n-gram | Qwen3.6-27B NVFP4 | draft-free (SPEC-NGRAM); 27B 5/5 STRICT our-ngram-ON == vLLM-ngram-ON, 180/180 drafts accepted (correctness only, no speed row yet) |
DONE |
| DSpark | 27B NVFP4 dense k=15; 35B-A3B MoE k=8 | MoE 35B-A3B: 0.835x paired on kairos-17dd (matched 89 tokens, warm oracle cache). Prior 0.957-0.989 came from a different machine with a cold oracle (#442) | ACTIVE |
| DSpark block floor | Qwen3.8-27B + RadixArk/Qwen3.8-27B-DSpark @ 85ef153b |
a k below the draft's block is refused instead of drafted; the run gate that exhibits the garbling is owed and needs a GPU lease (#1225) |
ACTIVE |
| DSpark draft routing | Qwen3.8-27B + RadixArk/Qwen3.8-27B-DSpark @ 85ef153b |
PENDING, no number. The token-exact run gate needs the 2.53 GiB draft and GPU time, and neither authority is recorded; only the CPU classification gate has run (.agents/specs/dspark-qwen3-routing.md §6) |
ACTIVE |
| DFlash2 | Qwen3.8-27B + z-lab/Qwen3.8-27B-DFlash2, k=7, GB10, oracle BEYOND-PIN @66e5414c on TRITON_ATTN (#1456, #1562) |
0.8017x decode throughput (13.051 vs 16.279 tok/s, 16 warm legs each): RECORDED, NO floor declared -- a ratio, NOT a pass. 3 of 4 axes NOT MEASURED. Correctness READ: 4/4 token-exact, 45/47 draft blocks identical | Ours is OFF the paged CUDA-graph fast path, the oracle GRAPHS its draft step. 5 draft layers resolved FlashAttentionBackend (#1685); arm clock windows differ (#1673). record |
Drafter chain (SPEC-DRAFTER-CHAIN, #1522) |
n/a, nothing runs a chain yet | PENDING, no number admissible. W1 landed the field and its refusals only. The premise is an ACCEPTANCE claim, invisible to a token gate, so W2/W4 must read first. G1 and G5 PASS on CPU | ACTIVE |
| Breadth (EAGLE1/3, suffix, ngram-gpu, dynamic-k, ...) | n/a | enumerated from vLLM source + INVENTORIED 2026-08-06 (.agents/specs/spec-decode-inventory.md), unmeasured |
INVENTORIED |