Does Nvidia really limit OpenGL on GTX 4XX+?

[Faizol]

Hi all,

This issue was highlighted sometime in July/August last year but I didn't really paid attention to it back then. Been checking on the news lately (google, phoronix) but nothing can really confirm if the above is the actual case. Nvidia staffs on their forum didn't reply to the questions posted by developers whom provide benchmarks to support their claims . Anyway, is the GTX 4XX+ still reportedly has poor performance in OpenGL?

Someone mentions before that he returned a bunch of Quadro 4000 cards also because of unexpectedly poor performance. If that's the case, then it might not necessarily driver throttling?

Can anyone who experience shed some light?

Thanks.

cheers.

[pixelrush]

Nvidia have been constructively managing the performance of Quadro vs Geforce versions of their cards for some time under the guise that those and their drivers are tuned for different purposes.
A Geforce typically has about 1/3-1/4x the OpenGL grunt for CAD applications for the same chip configured as a Quadro and for the later you pay even 5-6x as much.
In the case of the 4xx and 5xx series I am inclined to think there is additionally a flaw with the Fermi chip design rather than there being a driver issue. Should someone with technical smarts organise a class action and would it succeed? dunno...
It would seem impractical to recall and replace them all.
Although they do work clearly there is an issue with performance being sub par especially if you bought thinking you were upgrading.
Somewhere in there is a line in the sand that says it is not fit for the purpose, the advertising is misleading, there is a serious design/manufacturing defect, or such, and you should get a full refund if you ask for it by returning the item .
AFAIK the issue is with double sided lighting or at least thats what has been mentioned on the Blender forum lately and people seem to have confirmed a healthy improvement in benchmarks with dsl set off.