🔌
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 OpenELM-450M?
Yes — here's how fast and up to what context, with real measured numbers.
OpenELM-450M on the NVIDIA DGX Spark
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 1721k | 221.4-263.9 t/s | ~8.1-9.7 (slow) |
| Q8 | yes | 1728k | 442.9-527.8 t/s | ~16.2-19.3 (slow) |
| Q4 | yes | 1731k | 790.8-942.5 t/s | ~29-34.5 (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 OpenELM-450M?▶
Yes. At Q4 it generates ~790.8-942.5 tok/s and fits up to 1731k context; beyond that it spills to SSD swap and slows to ~29-34.5 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.