🔌
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 NVIDIA DGX Spark run Retrieval-Augmented Generator?
Yes — here's how fast and up to what context, with real measured numbers.
Retrieval-Augmented Generator on the NVIDIA DGX Spark
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 1716k | 159.2-189.7 t/s | ~5.8-6.9 (slow) |
| Q8 | yes | 1725k | 318.4-379.4 t/s | ~11.7-13.9 (slow) |
| Q4 | yes | 1730k | 568.5-677.5 t/s | ~20.8-24.8 (slow) |
Estimate (memory-bound). Beyond "max context" the KV cache spills to SSD swap and speed drops to the "beyond" figure.
FAQ
Can the NVIDIA DGX Spark run Retrieval-Augmented Generator?▶
Yes. At Q4 it generates ~568.5-677.5 tok/s and fits up to 1730k context; beyond that it spills to SSD swap and slows to ~20.8-24.8 tok/s.
How much context fits?▶
At Q4, up to about 1730k tokens stay in fast memory; longer context spills to SSD swap and throughput collapses.