// RUN โ€” FITS

Can the NVIDIA RTX 6000 Ada 48GB run AntiFormer?

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

AntiFormer on the NVIDIA RTX 6000 Ada 48GB

QuantFits?Max contextSpeedBeyond context
FP16 yes 650k 14203.1-16927 t/s ~591.8-705.3 (slow)
Q8 yes 651k 28406.3-33854.1 t/s ~1183.6-1410.6 (slow)
Q4 yes 651k 50725.5-60453.7 t/s ~2113.6-2518.9 (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 RTX 6000 Ada 48GB run AntiFormer?โ–ถ

Yes. At Q4 it generates ~50725.5-60453.7 tok/s and fits up to 651k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~2113.6-2518.9 tok/s.

How much context fits?โ–ถ

At Q4, up to about 651k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.