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 CodeT5-base?
Yes — here's how fast and up to what context, with real measured numbers.
CodeT5-base on the Instinct MI430X
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 6035k | 33361.4-40775 t/s | ~57.3-70 (slow) |
| Q8 | yes | 6038k | 66722.7-81550 t/s | ~114.5-140 (slow) |
| Q4 | yes | 6040k | 119147.7-145625 t/s | ~204.5-250 (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 CodeT5-base?▶
Yes. At Q4 it generates ~119147.7-145625 tok/s and fits up to 6040k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~204.5-250 tok/s.
How much context fits?▶
At Q4, up to about 6040k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.