// RUN โ€” FITS

Can the Instinct MI440X run ESM2-35M?

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

ESM2-35M on the Instinct MI440X

QuantFits?Max contextSpeedBeyond context
FP16 yes 6040k 176400-215600 t/s ~360-440 (slow)
Q8 yes 6041k 352800-431200 t/s ~720-880 (slow)
Q4 yes 6041k 630000-770000 t/s ~1285.7-1571.4 (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 MI440X run ESM2-35M?โ–ถ

Yes. At Q4 it generates ~630000-770000 tok/s and fits up to 6041k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~1285.7-1571.4 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.