🔌
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 Intel Gaudi 3 run BART-large?
Yes — here's how fast and up to what context, with real measured numbers.
BART-large on the Intel Gaudi 3
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 1762k | 2641-3278.4 t/s | ~28.6-35.4 (slow) |
| Q8 | yes | 1768k | 5281.9-6556.9 t/s | ~57.1-70.9 (slow) |
| Q4 | yes | 1771k | 9432-11708.7 t/s | ~102-126.6 (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 BART-large?▶
Yes. At Q4 it generates ~9432-11708.7 tok/s and fits up to 1771k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~102-126.6 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.