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 AWD-LSTM-DOC (fin) (37M)?
Yes — here's how fast and up to what context, with real measured numbers.
AWD-LSTM-DOC (fin) (37M) on the Instinct MI455X
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 6040k | 166864.9-203945.9 t/s | ~340.5-416.2 (slow) |
| Q8 | yes | 6041k | 333729.7-407891.9 t/s | ~681.1-832.4 (slow) |
| Q4 | yes | 6041k | 595945.9-728378.4 t/s | ~1216.2-1486.5 (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 AWD-LSTM-DOC (fin) (37M)?▶
Yes. At Q4 it generates ~595945.9-728378.4 tok/s and fits up to 6041k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~1216.2-1486.5 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.