// RUN โ€” FITS

Can the AMD Instinct MI325X run ByT5-XXL?

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

ByT5-XXL on the AMD Instinct MI325X

QuantFits?Max contextSpeedBeyond context
FP16 yes 1271k 146.5-179.1 t/s ~1-1.2 (slow)
Q8 yes 1350k 293-358.1 t/s ~2-2.4 (slow)
Q4 yes 1384k 523.3-639.5 t/s ~3.5-4.3 (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 AMD Instinct MI325X run ByT5-XXL?โ–ถ

Yes. At Q4 it generates ~523.3-639.5 tok/s and fits up to 1384k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~3.5-4.3 tok/s.

How much context fits?โ–ถ

At Q4, up to about 1384k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.