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 UDSMProt?
Yes — here's how fast and up to what context, with real measured numbers.
UDSMProt on the Instinct MI430X
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 6041k | 259311.5-316936.2 t/s | ~445.2-544.1 (slow) |
| Q8 | yes | 6041k | 518622.9-633872.5 t/s | ~890.3-1088.2 (slow) |
| Q4 | yes | 6041k | 926112.4-1131915.1 t/s | ~1589.9-1943.2 (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 UDSMProt?▶
Yes. At Q4 it generates ~926112.4-1131915.1 tok/s and fits up to 6041k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~1589.9-1943.2 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.