🔌
New: connect Claude & other AIs to GenAIList over MCP — research the catalog and contribute to the shared knowledge base. Learn how →
// RUN — FITS
Can the AMD Instinct MI455X run Me Llama 70B?
Yes — here's how fast and up to what context, with real measured numbers.
Me Llama 70B on the AMD Instinct MI455X
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 781k | 88.2-107.8 t/s | ~0.2-0.2 (slow) |
| Q8 | yes | 995k | 176.4-215.6 t/s | ~0.4-0.4 (slow) |
| Q4 | yes | 1089k | 315-385 t/s | ~0.6-0.8 (slow) |
Estimate (memory-bound). Beyond "max context" the KV cache spills to system RAM offload (depends on your system RAM) and speed drops to the "beyond" figure.
FAQ
Can the AMD Instinct MI455X run Me Llama 70B?▶
Yes. At Q4 it generates ~315-385 tok/s and fits up to 1089k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~0.6-0.8 tok/s.
How much context fits?▶
At Q4, up to about 1089k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.