🔌
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 MI440X run Qwen-VL?
Yes — here's how fast and up to what context, with real measured numbers.
Qwen-VL on the Instinct MI440X
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 2874k | 643.1-786 t/s | ~1.3-1.6 (slow) |
| Q8 | yes | 2948k | 1286.3-1572.1 t/s | ~2.6-3.2 (slow) |
| Q4 | yes | 2980k | 2296.9-2807.3 t/s | ~4.7-5.7 (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 MI440X run Qwen-VL?▶
Yes. At Q4 it generates ~2296.9-2807.3 tok/s and fits up to 2980k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~4.7-5.7 tok/s.
How much context fits?▶
At Q4, up to about 2980k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.