🔌
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 Ovis1.6-Gemma2-9B?
Yes — here's how fast and up to what context, with real measured numbers.
Ovis1.6-Gemma2-9B on the Instinct MI455X
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 2292k | 605.3-739.8 t/s | ~1.2-1.5 (slow) |
| Q8 | yes | 2354k | 1210.6-1479.6 t/s | ~2.5-3 (slow) |
| Q4 | yes | 2382k | 2161.8-2642.2 t/s | ~4.4-5.4 (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 Ovis1.6-Gemma2-9B?▶
Yes. At Q4 it generates ~2161.8-2642.2 tok/s and fits up to 2382k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~4.4-5.4 tok/s.
How much context fits?▶
At Q4, up to about 2382k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.