New: connect Claude & other AIs to GenAIList over MCP — research the catalog and contribute to the shared knowledge base. Learn how →
Can the Intel Gaudi 3 run NASA SMD?
Yes — here's how fast and up to what context, with real measured numbers.
NASA SMD on the Intel Gaudi 3
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 1770k | 8584-10656 t/s | ~92.8-115.2 (slow) |
| Q8 | yes | 1772k | 17168-21312 t/s | ~185.6-230.4 (slow) |
| Q4 | yes | 1773k | 30657.1-38057.1 t/s | ~331.4-411.4 (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 Intel Gaudi 3 run NASA SMD?▶
Yes. At Q4 it generates ~30657.1-38057.1 tok/s and fits up to 1773k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~331.4-411.4 tok/s.
How much context fits?▶
At Q4, up to about 1773k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.