New: connect Claude & other AIs to GenAIList over MCP — research the catalog and contribute to the shared knowledge base. Learn how →
Can the Instinct MI430X run Swallow?
Yes — here's how fast and up to what context, with real measured numbers.
Swallow on the Instinct MI430X
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 6039k | 104850-128150 t/s | ~180-220 (slow) |
| Q8 | yes | 6040k | 209700-256300 t/s | ~360-440 (slow) |
| Q4 | yes | 6041k | 374464.3-457678.6 t/s | ~642.9-785.7 (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 MI430X run Swallow?▶
Yes. At Q4 it generates ~374464.3-457678.6 tok/s and fits up to 6041k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~642.9-785.7 tok/s.
How much context fits?▶
At Q4, up to about 6041k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.