🔌
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 AMD Radeon Pro W7900 48GB run Cosmos-Predict2-2B-Text2Image?
Yes — here's how fast and up to what context, with real measured numbers.
Cosmos-Predict2-2B-Text2Image on the AMD Radeon Pro W7900 48GB
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 590k | 136.1-166.3 t/s | ~6.3-7.7 (slow) |
| Q8 | yes | 620k | 272.2-332.6 t/s | ~12.6-15.4 (slow) |
| Q4 | yes | 634k | 486-594 t/s | ~22.5-27.5 (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.
→ All models on the AMD Radeon Pro W7900 48GB
→ Cosmos-Predict2-2B-Text2Image specs
→ Other hardware
FAQ
Can the AMD Radeon Pro W7900 48GB run Cosmos-Predict2-2B-Text2Image?▶
Yes. At Q4 it generates ~486-594 tok/s and fits up to 634k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~22.5-27.5 tok/s.
How much context fits?▶
At Q4, up to about 634k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.