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 GPT-2 (355M)?
Yes — here's how fast and up to what context, with real measured numbers.
GPT-2 (355M) on the Intel Gaudi 3
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 1763k | 3022.5-3752.1 t/s | ~32.7-40.6 (slow) |
| Q8 | yes | 1769k | 6045.1-7504.2 t/s | ~65.4-81.1 (slow) |
| Q4 | yes | 1771k | 10794.8-13400.4 t/s | ~116.7-144.9 (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.
FAQ
Can the Intel Gaudi 3 run GPT-2 (355M)?▶
Yes. At Q4 it generates ~10794.8-13400.4 tok/s and fits up to 1771k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~116.7-144.9 tok/s.
How much context fits?▶
At Q4, up to about 1771k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.