🔌
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 P-LLama3?
Yes — here's how fast and up to what context, with real measured numbers.
P-LLama3 on the Huawei Ascend 950DT
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 877k | 162.5-197.5 t/s | ~1.6-2 (slow) |
| Q8 | yes | 938k | 325-395 t/s | ~3.3-4 (slow) |
| Q4 | yes | 965k | 580.4-705.4 t/s | ~5.8-7.1 (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 Huawei Ascend 950DT run P-LLama3?▶
Yes. At Q4 it generates ~580.4-705.4 tok/s and fits up to 965k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~5.8-7.1 tok/s.
How much context fits?▶
At Q4, up to about 965k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.