// RUN โ€” FITS

Can the Instinct MI455X run BEIT-3?

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

BEIT-3 on the Instinct MI455X

QuantFits?Max contextSpeedBeyond context
FP16 yes 5984k 3249.5-3971.6 t/s ~6.6-8.1 (slow)
Q8 yes 6013k 6498.9-7943.2 t/s ~13.3-16.2 (slow)
Q4 yes 6025k 11605.3-14184.2 t/s ~23.7-28.9 (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 Instinct MI455X run BEIT-3?โ–ถ

Yes. At Q4 it generates ~11605.3-14184.2 tok/s and fits up to 6025k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~23.7-28.9 tok/s.

How much context fits?โ–ถ

At Q4, up to about 6025k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.