// RUN โ€” FITS

Can the Instinct MI430X run Tri-21B?

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

Tri-21B on the Instinct MI430X

QuantFits?Max contextSpeedBeyond context
FP16 yes 1352k 354.1-432.7 t/s ~0.6-0.7 (slow)
Q8 yes 1431k 708.1-865.5 t/s ~1.2-1.5 (slow)
Q4 yes 1466k 1264.5-1545.5 t/s ~2.2-2.7 (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 MI430X run Tri-21B?โ–ถ

Yes. At Q4 it generates ~1264.5-1545.5 tok/s and fits up to 1466k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~2.2-2.7 tok/s.

How much context fits?โ–ถ

At Q4, up to about 1466k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.