// RUN โ€” FITS

Can the Intel Gaudi 3 run Gemma 2 27B?

Yes โ€” here's how fast and up to what context, with real measured numbers.

Gemma 2 27B on the Intel Gaudi 3

QuantFits?Max contextSpeedBeyond context
FP16 yes 238k 39.7-49.3 t/s ~0.4-0.5 (slow)
Q8 yes 340k 79.5-98.7 t/s ~0.9-1.1 (slow)
Q4 yes 386k 141.9-176.2 t/s ~1.5-1.9 (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 Gemma 2 27B?โ–ถ

Yes. At Q4 it generates ~141.9-176.2 tok/s and fits up to 386k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~1.5-1.9 tok/s.

How much context fits?โ–ถ

At Q4, up to about 386k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.