🔌
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 Unified-IO (XL)?
Yes — here's how fast and up to what context, with real measured numbers.
Unified-IO (XL) on the NVIDIA DGX Spark
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 1646k | 34.1-40.6 t/s | ~1.2-1.5 (slow) |
| Q8 | yes | 1690k | 68.1-81.2 t/s | ~2.5-3 (slow) |
| Q4 | yes | 1710k | 121.7-145 t/s | ~4.5-5.3 (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 Unified-IO (XL)?▶
Yes. At Q4 it generates ~121.7-145 tok/s and fits up to 1710k context; beyond that it spills to SSD swap and slows to ~4.5-5.3 tok/s.
How much context fits?▶
At Q4, up to about 1710k tokens stay in fast memory; longer context spills to SSD swap and throughput collapses.