MI300X FP8: M3 EAGLE vs Off Speculative Decoding
Speculative decoding comparison of M3 EAGLE versus Off on MI300X FP8 (AMD CDNA 3) running MiniMax M3 428B. 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.
M3 EAGLE posts 1132 tok/s/chip for $0.23 per million tokens at 30 tok/s/user on MiniMax M3 428B (MI300X FP8); Off posts 1013 tok/s/chip for $0.26. M3 EAGLE is 12% cheaper per token; M3 EAGLE delivers 12% more tok/s/chip. Draft-token acceptance rates determine whether speculative decoding helps or hurts at a given concurrency level.
Throughput at 53 tok/s/user on MiniMax M3 428B (MI300X FP8): M3 EAGLE hits 676 tok/s/chip, Off hits 627. Per-million costs land at $0.39 and $0.42 respectively. M3 EAGLE is 8% cheaper per token; M3 EAGLE delivers 8% more tok/s/chip. Speculative decoding trades extra compute on draft tokens for fewer decoding steps — the payoff depends on sequence length and batch size.
Toward the upper edge of the 8–97 tok/s/user interactivity band, at 75 tok/s/user on MiniMax M3 428B (MI300X FP8): M3 EAGLE runs 631 tok/s/chip at $0.42/M tokens, Off runs 407 at $0.65/M. M3 EAGLE is 55% cheaper per token; M3 EAGLE delivers 55% more tok/s/chip. Gains from speculative decoding vary by workload; short-output prompts tend to benefit less. (Numbers reflect this URL's pinned 8k/1k · fp8 workload — changing sequence or model updates both the table and chart; the table stays pinned to this page's precision, so precision toggles in the controls affect the chart only.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | M3 EAGLE:1131.7Off:1012.7 | M3 EAGLE:675.6Off:626.9 | M3 EAGLE:631.4Off:406.6 |
| Cost ($/M tok) | M3 EAGLE:$0.233Off:$0.261 | M3 EAGLE:$0.391Off:$0.421 | M3 EAGLE:$0.418Off:$0.649 |
| tok/s/MW | M3 EAGLE:814199Off:728533 | M3 EAGLE:486038Off:451023 | M3 EAGLE:454263Off:292539 |
| Concurrency | M3 EAGLE:~41Off:~31 | M3 EAGLE:~13Off:~11 | M3 EAGLE:~9Off:~5 |
Inference Performance
Agentic inference metrics from the AgentX scenario and fixed-sequence inference metrics across models, hardware configurations, and serving parameters.