Qwen3.6-27B · vLLM · NVFP4 + MTP · conc 2
status: done
Configuration
| Model | Qwen/Qwen3.6-27B |
|---|---|
| Company | Alibaba |
| Family | Qwen |
| Parameters | 27B (dense) |
| Engine | vLLM + MTP (num_speculative_tokens=3) (speculative decoding) |
| Quant / precision | NVFP4 |
| Why this quant | conc-2 MTP point of the Qwen3.6-27B NVFP4 sweep — the base+MTP speedup-decay curve (EXPERIMENTS.md |
| Download | unsloth/Qwen3.6-27B-NVFP4 |
| Context window | 65536 |
| Input modalities | text, image, video (served text-only here) |
Measured results
| Prefill tok/s | 56.13 |
|---|---|
| Decode tok/s | 35.68 |
| Peak memory (GB) | 109.42 (system MemAvailable delta (10s sampling) — vLLM static KV reservation (util 0.85) + MTP head) |
| Completed | 2026-07-01 23:14 +0800 |
Full run command
# conc-2 MTP. Same recipe as -mtp-c8 — only --max-num-seqs differs.
scripts/bench-vllm-serving.sh unsloth/Qwen3.6-27B-NVFP4 65536 2 1000 600 256 \
--trust-remote-code --dtype bfloat16 \
--speculative-config '{"method":"mtp","num_speculative_tokens":3}'
Notes
conc-2 MTP point of the Qwen3.6-27B NVFP4 sweep — paired with the matched base -c2
for the base-vs-MTP speedup-decay curve (EXPERIMENTS.md #14).
- Result (conc 2): prefill 56.13 / decode 35.68 tok/s aggregate; 0 errors; peak mem 109.4 GB.
- MTP speedup at conc-2: +97.0% (35.68 vs base 18.11) — nearly a 2× single-stream-ish win on the dense 27B. The dense MTP advantage is largest at low batch: +80.6% (c1) → +97.0% (c2) → +62.6% (c8), far above the 35B-A3B MoE’s ~+25–40%, because a dense target has essentially no spare compute for the batch to reclaim.
- Acceptance ~68%, accept-len ~3.0-of-3 — the MTP draft is highly efficient (per-position 0.85/0.69/0.55), matching the -c8 row’s ~71%. Acceptance is workload-driven and batch-stable here (unlike the gpt-oss EAGLE3 low-batch pathology).
- Sweep: base
-c2· MTP-c1·-c4·-c8·c32.