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 Qwen2.5-Coder (1.5B)?
Yes — here's how fast and up to what context, with real measured numbers.
Qwen2.5-Coder (1.5B) on the Instinct MI430X
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 5995k | 4765.9-5825 t/s | ~8.2-10 (slow) |
| Q8 | yes | 6018k | 9531.8-11650 t/s | ~16.4-20 (slow) |
| Q4 | yes | 6028k | 17021.1-20803.6 t/s | ~29.2-35.7 (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 Qwen2.5-Coder (1.5B)?▶
Yes. At Q4 it generates ~17021.1-20803.6 tok/s and fits up to 6028k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~29.2-35.7 tok/s.
How much context fits?▶
At Q4, up to about 6028k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.