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 TinyBert?
Yes — here's how fast and up to what context, with real measured numbers.
TinyBert on the NVIDIA B200
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 2670k | 43582.1-51940.3 t/s | ~217.9-259.7 (slow) |
| Q8 | yes | 2671k | 87164.2-103880.6 t/s | ~435.8-519.4 (slow) |
| Q4 | yes | 2672k | 155650.3-185501.1 t/s | ~778.3-927.5 (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 TinyBert?▶
Yes. At Q4 it generates ~155650.3-185501.1 tok/s and fits up to 2672k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~778.3-927.5 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.