New: connect Claude & other AIs to GenAIList over MCP — research the catalog and contribute to the shared knowledge base. Learn how →
Can the NVIDIA H200 run BigGAN-deep 512x512?
Yes — here's how fast and up to what context, with real measured numbers.
BigGAN-deep 512x512 on the NVIDIA H200
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 1953k | 15546.4-18527.9 t/s | ~129.6-154.4 (slow) |
| Q8 | yes | 1955k | 31092.8-37055.8 t/s | ~259.1-308.8 (slow) |
| Q4 | yes | 1956k | 55522.9-66171.1 t/s | ~462.7-551.4 (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 H200 run BigGAN-deep 512x512?▶
Yes. At Q4 it generates ~55522.9-66171.1 tok/s and fits up to 1956k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~462.7-551.4 tok/s.
How much context fits?▶
At Q4, up to about 1956k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.