🔌
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 FuriosaAI RNGD run DDPM-IP (CelebA)?
Yes — here's how fast and up to what context, with real measured numbers.
DDPM-IP (CelebA) on the FuriosaAI RNGD
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 642k | 1652.5-2008.5 t/s | ~44.1-53.6 (slow) |
| Q8 | yes | 646k | 3305.1-4016.9 t/s | ~88.1-107.1 (slow) |
| Q4 | yes | 648k | 5901.9-7173.1 t/s | ~157.4-191.3 (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 FuriosaAI RNGD run DDPM-IP (CelebA)?▶
Yes. At Q4 it generates ~5901.9-7173.1 tok/s and fits up to 648k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~157.4-191.3 tok/s.
How much context fits?▶
At Q4, up to about 648k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.