MI325X FP8: MTP vs Off Speculative Decoding
Speculative decoding comparison of MTP versus Off on MI325X FP8 (AMD CDNA 3) running Qwen 3.5 397B-A17B. Throughput, cost, and interactivity differences across LLM workloads. Use the chart controls below to switch sequences and metrics — same interactions as the main inference chart.
MTP acceptance-rate implementations differ across inference engines. Points from different engines are not directly comparable on the same curve — throughput and cost at matched interactivity may reflect engine-level differences rather than pure speculative decoding gains. Interpret cross-engine comparisons with caution.
At 46 tok/s/user on Qwen 3.5 397B-A17B (MI325X FP8), MTP delivers 402 tok/s/GPU at $0.87 per million tokens; Off delivers 400 tok/s/GPU at $0.87. Cost per token is essentially tied; throughput per GPU is essentially tied. Speculative decoding accepts draft tokens to reduce per-token latency — gains vary by workload and prompt distribution.
MTP posts 230 tok/s/GPU for $1.55 per million tokens at 55 tok/s/user on Qwen 3.5 397B-A17B (MI325X FP8); Off posts 229 tok/s/GPU for $1.56. Cost per token is essentially tied; throughput per GPU is essentially tied. Draft-token acceptance rates determine whether speculative decoding helps or hurts at a given concurrency level.
Throughput at 64 tok/s/user on Qwen 3.5 397B-A17B (MI325X FP8): MTP hits 130 tok/s/GPU, Off hits 132. Per-million costs land at $2.72 and $2.65 respectively. Off is 3% cheaper per token; Off delivers 2% more tok/s/GPU. Speculative decoding trades extra compute on draft tokens for fewer decoding steps — the payoff depends on sequence length and batch size. (Numbers reflect this URL's pinned 1k/1k · fp8 workload — changing sequence or model updates both the table and chart; the table stays pinned to this page's precision, so precision toggles in the controls affect the chart only.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | MTP:402.4Off:400.1 | MTP:230.1Off:229.1 | MTP:129.6Off:132.4 |
| Cost ($/M tok) | MTP:$0.866Off:$0.870 | MTP:$1.552Off:$1.559 | MTP:$2.725Off:$2.651 |
| tok/s/MW | MTP:184598Off:183546 | MTP:105537Off:105102 | MTP:59440Off:60756 |
| Concurrency | MTP:~37Off:~37 | MTP:~18Off:~18 | MTP:~9Off:~9 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.