Nvidia greenboost: transparently extend GPU VRAM using system RAM/NVMe

Posted by mmastrac 3 days ago

353 points | 75 commentspage 3

pabs3 3 days ago|

Would be great to get this into mainline Linux.

aplomb1026 13 hours ago||

[dead]

ajaimk 12 hours ago||

[dead]

holoduke 14 hours ago||

The is extremely slow and not useful in my opinion.

daneel_w 14 hours ago||

It makes the difference between being able to run a lot of machine learning tasks, and not being able at all. Pretty useful.

majorchord 14 hours ago|||

I would say it depends entirely on your usecase. I don't think there can be a simple "not useful" generalization that applies to everyone.

jauntywundrkind 14 hours ago||

Man I wish that was a canned response that could be deployed on demand! Well said.

I really appreciate thriftful & resourceful points of view. Exploring what if, looking for use is such a great virtue.

bigwheels 14 hours ago|||

Can you elaborate beyond the shallow/superficial dismissal?

whywhywhywhy 2 hours ago||

If it takes seconds in VRAM it can take tens of minutes running the same thing offloaded to RAM if it hasn't been designed to do it.

ozgrakkurt 6 hours ago|||

It is about as useful as rtx

sayYayToLife 14 hours ago||

[dead]

NooneAtAll3 8 hours ago|

nvidia failed to provide gpu with actually meaningful amount of vram

and instead of improving the actual product, it decided to "solve the problem in software"

I expect this greenboost to fall and burn, honestly...

cma 8 hours ago|

> it decided to "solve the problem in software"

This isn't made by nvidia

shmeeed 1 hour ago||

Still kinda true, though. As other commenters have pointed out, their Windows drivers do similar stuff.