🔌
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 H200 run iGPT-L?
Yes — here's how fast and up to what context, with real measured numbers.
iGPT-L on the NVIDIA H200
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 1915k | 1286.3-1533 t/s | ~10.7-12.8 (slow) |
| Q8 | yes | 1936k | 2572.7-3066.1 t/s | ~21.4-25.6 (slow) |
| Q4 | yes | 1945k | 4594.1-5475.1 t/s | ~38.3-45.6 (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 iGPT-L?▶
Yes. At Q4 it generates ~4594.1-5475.1 tok/s and fits up to 1945k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~38.3-45.6 tok/s.
How much context fits?▶
At Q4, up to about 1945k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.