Llama 3.3 70B — H100 vs MI355X
Head-to-head AI inference benchmark comparison of H100 (NVIDIA Hopper) and MI355X (AMD CDNA 4) on Llama 3.3 70B. Latency, throughput, and cost across LLM workloads. Use the chart controls below to switch sequences, precisions, and metrics — same interactions as the main inference chart.
H100 / MI355X on Llama 3.3 70B at 53 tok/s/user: 1465 / 2248 tok/s/GPU, $0.25 / $0.18 per million tokens. MI355X is 39% cheaper per token; MI355X delivers 53% more tok/s/GPU.
Around the middle of the 35–109 tok/s/user interactivity band, at 72 tok/s/user on Llama 3.3 70B: H100 runs 773 tok/s/GPU at $0.45/M tokens, MI355X runs 1632 at $0.25/M. MI355X is 84% cheaper per token; MI355X delivers 111% more tok/s/GPU.
Setting 91 tok/s/user as the target on Llama 3.3 70B, H100 produces 304 tok/s/GPU ($1.19 per million tokens) and MI355X produces 952 ($0.42). MI355X is 182% cheaper per token; MI355X delivers 214% more tok/s/GPU. (Numbers reflect the default 1k/1k · fp8 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | H100:1465.2MI355X:2248.4 | H100:772.9MI355X:1631.6 | H100:303.7MI355X:952.3 |
| Cost ($/M tok) | H100:$0.249MI355X:$0.179 | H100:$0.454MI355X:$0.247 | H100:$1.194MI355X:$0.424 |
| tok/s/MW | H100:846932MI355X:848453 | H100:446737MI355X:615682 | H100:175535MI355X:359351 |
| Concurrency | H100:~64MI355X:~47 | H100:~50MI355X:~53 | H100:~14MI355X:~22 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.