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 GBERT-Large?
Yes — here's how fast and up to what context, with real measured numbers.
GBERT-Large on the NVIDIA B200
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 2662k | 8716.4-10388.1 t/s | ~43.6-51.9 (slow) |
| Q8 | yes | 2667k | 17432.8-20776.1 t/s | ~87.2-103.9 (slow) |
| Q4 | yes | 2670k | 31130.1-37100.2 t/s | ~155.7-185.5 (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 GBERT-Large?▶
Yes. At Q4 it generates ~31130.1-37100.2 tok/s and fits up to 2670k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~155.7-185.5 tok/s.
How much context fits?▶
At Q4, up to about 2670k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.