🔌
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 Huawei Ascend 950DT 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 Huawei Ascend 950DT
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 1993k | 7027-8540.5 t/s | ~70.3-85.4 (slow) |
| Q8 | yes | 1996k | 14054.1-17081.1 t/s | ~140.5-170.8 (slow) |
| Q4 | yes | 1997k | 25096.5-30501.9 t/s | ~251-305 (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 Huawei Ascend 950DT
→ (ensemble): AWD-LSTM-DOC (fin) × 5 (WT2) specs
→ Other hardware
FAQ
Can the Huawei Ascend 950DT run (ensemble): AWD-LSTM-DOC (fin) × 5 (WT2)?▶
Yes. At Q4 it generates ~25096.5-30501.9 tok/s and fits up to 1997k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~251-305 tok/s.
How much context fits?▶
At Q4, up to about 1997k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.