🔌
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 DeBERTaV3large?
Yes — here's how fast and up to what context, with real measured numbers.
DeBERTaV3large on the NVIDIA DGX Spark
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 1722k | 238.4-284.1 t/s | ~8.7-10.4 (slow) |
| Q8 | yes | 1729k | 476.8-568.2 t/s | ~17.5-20.8 (slow) |
| Q4 | yes | 1731k | 851.4-1014.7 t/s | ~31.2-37.2 (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 DeBERTaV3large?▶
Yes. At Q4 it generates ~851.4-1014.7 tok/s and fits up to 1731k context; beyond that it spills to SSD swap and slows to ~31.2-37.2 tok/s.
How much context fits?▶
At Q4, up to about 1731k tokens stay in fast memory; longer context spills to SSD swap and throughput collapses.