// RUN โ€” FITS

Can the NVIDIA GeForce RTX 4070 Super run BART-large?

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

BART-large on the NVIDIA GeForce RTX 4070 Super

QuantFits?Max contextSpeedBeyond context
FP16 yes 133k 452.8-539.6 t/s ~35.9-42.8 (slow)
Q8 yes 139k 905.6-1079.2 t/s ~71.9-85.7 (slow)
Q4 yes 142k 1617.1-1927.2 t/s ~128.3-153 (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 GeForce RTX 4070 Super run BART-large?โ–ถ

Yes. At Q4 it generates ~1617.1-1927.2 tok/s and fits up to 142k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~128.3-153 tok/s.

How much context fits?โ–ถ

At Q4, up to about 142k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.