🔌
New: connect Claude & other AIs to GenAIList over MCP — research the catalog and contribute to the shared knowledge base. Learn how →
// RUN — FITS
Can the Instinct MI440X run Unified-IO (XL)?
Yes — here's how fast and up to what context, with real measured numbers.
Unified-IO (XL) on the Instinct MI440X
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 5952k | 2110.8-2579.8 t/s | ~4.3-5.3 (slow) |
| Q8 | yes | 5997k | 4221.5-5159.7 t/s | ~8.6-10.5 (slow) |
| Q4 | yes | 6017k | 7538.5-9213.7 t/s | ~15.4-18.8 (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 MI440X run Unified-IO (XL)?▶
Yes. At Q4 it generates ~7538.5-9213.7 tok/s and fits up to 6017k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~15.4-18.8 tok/s.
How much context fits?▶
At Q4, up to about 6017k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.