MI355X FP4: MTP vs Off Speculative Decoding
Speculative decoding comparison of MTP versus Off on MI355X FP4 (AMD CDNA 4) running DeepSeek V4 Pro 1.6T. 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.
Near the low end of the 3–76 tok/s/user interactivity band, at 21 tok/s/user on DeepSeek V4 Pro 1.6T (MI355X FP4): MTP runs 2232 tok/s/GPU at $0.18/M tokens, Off runs 4343 at $0.09/M. Off is 94% cheaper per token; Off delivers 95% more tok/s/GPU. Gains from speculative decoding vary by workload; short-output prompts tend to benefit less.
At 39 tok/s/user on DeepSeek V4 Pro 1.6T (MI355X FP4), MTP delivers 1636 tok/s/GPU at $0.25 per million tokens; Off delivers 1330 tok/s/GPU at $0.31. MTP is 23% cheaper per token; MTP delivers 23% more tok/s/GPU. Speculative decoding accepts draft tokens to reduce per-token latency — gains vary by workload and prompt distribution.
MTP posts 1155 tok/s/GPU for $0.35 per million tokens at 58 tok/s/user on DeepSeek V4 Pro 1.6T (MI355X FP4); Off posts 555 tok/s/GPU for $0.73. MTP is 105% cheaper per token; MTP delivers 108% more tok/s/GPU. Draft-token acceptance rates determine whether speculative decoding helps or hurts at a given concurrency level. (Numbers reflect this URL's pinned 8k/1k · fp4 workload — changing sequence or model updates both the table and chart; the table stays pinned to this page's precision, so precision toggles in the controls affect the chart only.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | MTP:2232.0Off:4342.7 | MTP:1636.2Off:1330.3 | MTP:1155.4Off:554.9 |
| Cost ($/M tok) | MTP:$0.184Off:$0.095 | MTP:$0.251Off:$0.309 | MTP:$0.355Off:$0.726 |
| tok/s/MW | MTP:842281Off:1638742 | MTP:617443Off:501999 | MTP:435983Off:209413 |
| Concurrency | MTP:~95Off:~901 | MTP:~39Off:~16 | MTP:~19Off:~5 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.