🔌
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 L40S 48GB run (ensemble): AWD-LSTM-DOC (fin) × 5 (WT2)?
Yes — here's how fast and up to what context, with real measured numbers.
(ensemble): AWD-LSTM-DOC (fin) × 5 (WT2) on the NVIDIA L40S 48GB
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 645k | 1704.6-2031.6 t/s | ~78.9-94.1 (slow) |
| Q8 | yes | 648k | 3409.3-4063.1 t/s | ~157.8-188.1 (slow) |
| Q4 | yes | 649k | 6088-7255.6 t/s | ~281.9-335.9 (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.
→ All models on the NVIDIA L40S 48GB
→ (ensemble): AWD-LSTM-DOC (fin) × 5 (WT2) specs
→ Other hardware
FAQ
Can the NVIDIA L40S 48GB run (ensemble): AWD-LSTM-DOC (fin) × 5 (WT2)?▶
Yes. At Q4 it generates ~6088-7255.6 tok/s and fits up to 649k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~281.9-335.9 tok/s.
How much context fits?▶
At Q4, up to about 649k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.