New: connect Claude & other AIs to GenAIList over MCP — research the catalog and contribute to the shared knowledge base. Learn how →
Can the Instinct MI455X run Retrieval-Augmented Generator?
Yes — here's how fast and up to what context, with real measured numbers.
Retrieval-Augmented Generator on the Instinct MI455X
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 6022k | 9862.6-12054.3 t/s | ~20.1-24.6 (slow) |
| Q8 | yes | 6032k | 19725.2-24108.6 t/s | ~40.3-49.2 (slow) |
| Q4 | yes | 6036k | 35223.6-43051.1 t/s | ~71.9-87.9 (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 Instinct MI455X run Retrieval-Augmented Generator?▶
Yes. At Q4 it generates ~35223.6-43051.1 tok/s and fits up to 6036k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~71.9-87.9 tok/s.
How much context fits?▶
At Q4, up to about 6036k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.