// RUN — FITS

Can the NVIDIA GeForce RTX 5090 run PIXART-α?

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

PIXART-α on the NVIDIA GeForce RTX 5090

QuantFits?Max contextSpeedBeyond context
FP16 yes 408k 1090.1-1299.2 t/s ~24.3-29 (slow)
Q8 yes 417k 2180.3-2598.4 t/s ~48.7-58 (slow)
Q4 yes 421k 3893.3-4640 t/s ~86.9-103.6 (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 5090 run PIXART-α?

Yes. At Q4 it generates ~3893.3-4640 tok/s and fits up to 421k context; beyond that it spills to system RAM offload (depends on your system RAM) and slows to ~86.9-103.6 tok/s.

How much context fits?

At Q4, up to about 421k tokens stay in fast memory; longer context spills to system RAM offload (depends on your system RAM) and throughput collapses.