🔌
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 Spec-Drafter?
Yes — here's how fast and up to what context, with real measured numbers.
Spec-Drafter on the Instinct MI455X
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 6026k | 12348-15092 t/s | ~25.2-30.8 (slow) |
| Q8 | yes | 6034k | 24696-30184 t/s | ~50.4-61.6 (slow) |
| Q4 | yes | 6037k | 44100-53900 t/s | ~90-110 (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 Spec-Drafter?▶
Yes. At Q4 it generates ~44100-53900 tok/s and fits up to 6037k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~90-110 tok/s.
How much context fits?▶
At Q4, up to about 6037k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.