// RUN โ€” FITS

Can the Instinct MI455X run ESM2-8M?

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

ESM2-8M on the Instinct MI455X

QuantFits?Max contextSpeedBeyond context
FP16 yes 6041k 771750-943250 t/s ~1575-1925 (slow)
Q8 yes 6041k 1543500-1886500 t/s ~3150-3850 (slow)
Q4 yes 6041k 2756250-3368750 t/s ~5625-6875 (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 ESM2-8M?โ–ถ

Yes. At Q4 it generates ~2756250-3368750 tok/s and fits up to 6041k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~5625-6875 tok/s.

How much context fits?โ–ถ

At Q4, up to about 6041k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.