🔌
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 MI455X run iGPT-L?
Yes — here's how fast and up to what context, with real measured numbers.
iGPT-L on the Instinct MI455X
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 6000k | 4533-5540.4 t/s | ~9.3-11.3 (slow) |
| Q8 | yes | 6021k | 9066.1-11080.8 t/s | ~18.5-22.6 (slow) |
| Q4 | yes | 6030k | 16189.4-19787.1 t/s | ~33-40.4 (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 iGPT-L?▶
Yes. At Q4 it generates ~16189.4-19787.1 tok/s and fits up to 6030k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~33-40.4 tok/s.
How much context fits?▶
At Q4, up to about 6030k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.