// RUN โ€” FITS

Can the Intel Gaudi 3 run HiFi - NN?

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

HiFi - NN on the Intel Gaudi 3

QuantFits?Max contextSpeedBeyond context
FP16 yes 1774k 357666.7-444000 t/s ~3866.7-4800 (slow)
Q8 yes 1774k 715333.3-888000 t/s ~7733.3-9600 (slow)
Q4 yes 1774k 1277381-1585714.3 t/s ~13809.5-17142.9 (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 HiFi - NN?โ–ถ

Yes. At Q4 it generates ~1277381-1585714.3 tok/s and fits up to 1774k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~13809.5-17142.9 tok/s.

How much context fits?โ–ถ

At Q4, up to about 1774k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.