// RUN โ€” FITS

Can the NVIDIA GeForce RTX 2050 run NASA SMD?

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

NASA SMD on the NVIDIA GeForce RTX 2050

QuantFits?Max contextSpeedBeyond context
FP16 yes 72k 327-389.8 t/s ~29.2-34.8 (slow)
Q8 yes 74k 654.1-779.5 t/s ~58.4-69.6 (slow)
Q4 yes 75k 1168-1392 t/s ~104.3-124.3 (slow)

Estimate (memory-bound). Beyond "max context" the KV cache spills to SSD swap and speed drops to the "beyond" figure.

FAQ

Can the NVIDIA GeForce RTX 2050 run NASA SMD?โ–ถ

Yes. At Q4 it generates ~1168-1392 tok/s and fits up to 75k context; beyond that it spills to SSD swap and slows to ~104.3-124.3 tok/s.

How much context fits?โ–ถ

At Q4, up to about 75k tokens stay in fast memory; longer context spills to SSD swap and throughput collapses.