🔌
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 RX 7900 XTX run Llama-NV-Embed-Reasoning-3B?
Yes — here's how fast and up to what context, with real measured numbers.
Llama-NV-Embed-Reasoning-3B on the AMD Radeon RX 7900 XTX
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 216k | 94.5-115.5 t/s | ~3.9-4.8 (slow) |
| Q8 | yes | 265k | 189-231 t/s | ~7.9-9.6 (slow) |
| Q4 | yes | 287k | 337.5-412.5 t/s | ~14.1-17.2 (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 AMD Radeon RX 7900 XTX run Llama-NV-Embed-Reasoning-3B?▶
Yes. At Q4 it generates ~337.5-412.5 tok/s and fits up to 287k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~14.1-17.2 tok/s.
How much context fits?▶
At Q4, up to about 287k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.