New: connect Claude & other AIs to GenAIList over MCP — research the catalog and contribute to the shared knowledge base. Learn how →
Can the NVIDIA B200 run DiffDock?
Yes — here's how fast and up to what context, with real measured numbers.
DiffDock on the NVIDIA B200
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 2672k | 144268.8-171936.8 t/s | ~721.3-859.7 (slow) |
| Q8 | yes | 2672k | 288537.5-343873.5 t/s | ~1442.7-1719.4 (slow) |
| Q4 | yes | 2672k | 515245.6-614059.9 t/s | ~2576.2-3070.3 (slow) |
Estimate (memory-bound). Beyond "max context" the KV cache spills to system RAM offload (depends on your system RAM) and speed drops to the "beyond" figure.
FAQ
Can the NVIDIA B200 run DiffDock?▶
Yes. At Q4 it generates ~515245.6-614059.9 tok/s and fits up to 2672k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~2576.2-3070.3 tok/s.
How much context fits?▶
At Q4, up to about 2672k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.