🔌
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 CodeT5-large?
Yes — here's how fast and up to what context, with real measured numbers.
CodeT5-large on the NVIDIA DGX Spark
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 1711k | 129.4-154.2 t/s | ~4.7-5.6 (slow) |
| Q8 | yes | 1723k | 258.8-308.5 t/s | ~9.5-11.3 (slow) |
| Q4 | yes | 1728k | 462.2-550.8 t/s | ~16.9-20.2 (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 CodeT5-large?▶
Yes. At Q4 it generates ~462.2-550.8 tok/s and fits up to 1728k context; beyond that it spills to SSD swap and slows to ~16.9-20.2 tok/s.
How much context fits?▶
At Q4, up to about 1728k tokens stay in fast memory; longer context spills to SSD swap and throughput collapses.