diff --git a/recipes/qwen3-8-flash-next-nvfp4-1x6000-jj-mtp3/recipe.yaml b/recipes/qwen3-8-flash-next-nvfp4-1x6000-jj-mtp3/recipe.yaml index db2031dbe..1e37b0f97 100644 --- a/recipes/qwen3-8-flash-next-nvfp4-1x6000-jj-mtp3/recipe.yaml +++ b/recipes/qwen3-8-flash-next-nvfp4-1x6000-jj-mtp3/recipe.yaml @@ -106,13 +106,28 @@ performance: tok_s: 117.0 context_len: 262144 notes: 'Observed on the live production server (vllm-fnxt-6000-jj, 2026-09-27/28, 230 engine samples at Running=1): single-stream decode p50 ~117 tok/s, p75 ~144, p90 ~181 (MTP3 mean acceptance length 2.61 of 3, draft accept 45-49%). Prompt chunks p50 ~1,150 tok/s, p90 ~2,900 tok/s (peak 20k on cache-warm tiny prompts). KV pool 619,559 tokens = 2.36x admission at 262K (log: GPU KV cache size 619,559, Maximum concurrency 2.36x). T0 log-derived, not a harness run.' - points: - - - context_len: 262144 - concurrency: 1 - decode_tok_s: 116.8 - prefill_tok_s: 1150.5 - notes: 'p50 single-stream from 228 engine samples (Running=1), mixed workloads; decode p75 144.2 / p90 181.2; prefill p90 ~2950 (12K+ chunks); MTP3 acceptance 2.61/3' + points: + - + context_len: 1 + concurrency: 1 + decode_tok_s: 157.6 + notes: 'r35 community image (consolidated): Saved JJ R35, 5-window median, ctx-0 decode, temp 1/top-p .95/top-k 20 (verifier 76.1 steps/s)' + - + context_len: 1 + concurrency: 8 + decode_tok_s: 674.2 + notes: 'r35 community image (consolidated): Saved JJ R35, aggregate C8 (verifier 330.0 steps/s)' + - + context_len: 32768 + concurrency: 1 + prefill_tok_s: 12104.0 + notes: 'r35 community image (consolidated): Saved JJ R35, uncached nominal-32K prefill, client TTFT' + - + context_len: 262144 + concurrency: 1 + decode_tok_s: 116.8 + prefill_tok_s: 1150.5 + notes: 'p50 single-stream from 228 engine samples (Running=1), mixed workloads; decode p75 144.2 / p90 181.2; prefill p90 ~2950 (12K+ chunks); MTP3 acceptance 2.61/3' artifacts: - kind: model_weights