// RUN โ€” FITS

Can the NVIDIA L40S 48GB run Baichuan 1-13B?

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

Baichuan 1-13B on the NVIDIA L40S 48GB

QuantFits?Max contextSpeedBeyond context
FP16 yes 98k 23.8-28.3 t/s ~1.1-1.3 (slow)
Q8 yes 179k 47.5-56.7 t/s ~2.2-2.6 (slow)
Q4 yes 215k 84.9-101.2 t/s ~3.9-4.7 (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 Baichuan 1-13B?โ–ถ

Yes. At Q4 it generates ~84.9-101.2 tok/s and fits up to 215k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~3.9-4.7 tok/s.

How much context fits?โ–ถ

At Q4, up to about 215k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.