New: connect Claude & other AIs to GenAIList over MCP — research the catalog and contribute to the shared knowledge base. Learn how →
Can the Instinct MI455X run DeBERTaV3large?
Yes — here's how fast and up to what context, with real measured numbers.
DeBERTaV3large on the Instinct MI455X
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 6029k | 14770.3-18052.6 t/s | ~30.1-36.8 (slow) |
| Q8 | yes | 6035k | 29540.7-36105.3 t/s | ~60.3-73.7 (slow) |
| Q4 | yes | 6038k | 52751.2-64473.7 t/s | ~107.7-131.6 (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 MI455X run DeBERTaV3large?▶
Yes. At Q4 it generates ~52751.2-64473.7 tok/s and fits up to 6038k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~107.7-131.6 tok/s.
How much context fits?▶
At Q4, up to about 6038k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.