// RUN โ€” FITS

Can the NVIDIA L40S 48GB run SANA 1.6B?

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

SANA 1.6B on the NVIDIA L40S 48GB

QuantFits?Max contextSpeedBeyond context
FP16 yes 601k 191.4-228.1 t/s ~8.9-10.6 (slow)
Q8 yes 626k 382.7-456.1 t/s ~17.7-21.1 (slow)
Q4 yes 637k 683.4-814.5 t/s ~31.6-37.7 (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 NVIDIA L40S 48GB run SANA 1.6B?โ–ถ

Yes. At Q4 it generates ~683.4-814.5 tok/s and fits up to 637k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~31.6-37.7 tok/s.

How much context fits?โ–ถ

At Q4, up to about 637k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.