🔌
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 BiomedGPT (182M)?
Yes — here's how fast and up to what context, with real measured numbers.
BiomedGPT (182M) on the NVIDIA DGX Spark
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 1729k | 547.5-652.5 t/s | ~20.1-23.9 (slow) |
| Q8 | yes | 1732k | 1095-1305 t/s | ~40.1-47.8 (slow) |
| Q4 | yes | 1733k | 1955.4-2330.4 t/s | ~71.6-85.4 (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 BiomedGPT (182M)?▶
Yes. At Q4 it generates ~1955.4-2330.4 tok/s and fits up to 1733k context; beyond that it spills to SSD swap and slows to ~71.6-85.4 tok/s.
How much context fits?▶
At Q4, up to about 1733k tokens stay in fast memory; longer context spills to SSD swap and throughput collapses.