// RUN โ€” FITS

Can the Intel Arc B580 12GB run AWD-LSTM-DOC (fin) (37M)?

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

AWD-LSTM-DOC (fin) (37M) on the Intel Arc B580 12GB

QuantFits?Max contextSpeedBeyond context
FP16 yes 144k 3574.1-4436.8 t/s ~313.5-389.2 (slow)
Q8 yes 145k 7148.1-8873.5 t/s ~627-778.4 (slow)
Q4 yes 145k 12764.5-15845.6 t/s ~1119.7-1390 (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 Intel Arc B580 12GB run AWD-LSTM-DOC (fin) (37M)?โ–ถ

Yes. At Q4 it generates ~12764.5-15845.6 tok/s and fits up to 145k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~1119.7-1390 tok/s.

How much context fits?โ–ถ

At Q4, up to about 145k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.