Adding a 'CUDA' videocard for cycles.

Hi all,

I know I should / could be patient for cycles to work with an AMD card (6950)…
But could it be done/ usefull to add an Nvidia card untill then?
Can the two brands sit side by side in one machine?
Can they both be used at the same time?
Nvidia for cycles and the other for desktop/ gaming in the mean time?

Kind regards,

Erik

It will be good, especially because current GPUs have very little preemtive capacity, with single GPU that used for GUI redraw and Cycles calculation you get laggy GUI. Get one extra NVIDIA card, select it as Computing device in User Property, and you get fast smooth GUI (on AMD by default as main device) and NVIDIA that work only as Cycles acceleration. One note, try to guess what is your your typical scene RAM requirement, Cycles cannot render scene that internal data not fit in GPU RAM. As common sense, 3GB onboard RAM is some minimal requirement, but sure there are scenes that cannot fin in that too.

Best case, get card with opportunity to return, and try to render scenes with high production subdivision quality.

Edit:

There is one exception, on windows there is OpenGL ICD feature that allow to coexist different vendor OPenGL drivers, In Linux side it different, you probably need some unusual modifications to environment variables to force one card as main OpenGL device, or setup and start another Xserver just for compute device.

Hi crins, bit late, but usually it is no problem using 2 Cards in one machine. In Blender you can select the Cuda card then.

EDIT: probably answered already…

For using Cycles it’s an excellent idéa which I’m also interested in - I actually tested this a few days back, though using dual Nvidia cards… BUT…

1. Be sure that your PSU can handle the strain. Just think about that before putting in a new card. An example: My GTX285 is recommended a 500w PSU, seeing the 560Ti peaks at 170w I needed at least a 670w PSU to be shure… (though I tested this on 500w being careful not to put to much strain on the card)
2. A Nvidia card will need two 6-pin connectors from the PSU to the card. If you don’t have that (say your Radeon card uses those) you’ll need four molex connections free from your PSU as the molex->6-pin adapters use two molex-connectors each. If you have free SATA power connections, not molex, you’ll also need SATA->molex adapters. An example: In my case I had two molex and two sata connections free thus I needed two molex->6-pin adapters as well as two SATA->molex adapters to get power for my secondary card. And then you’re all set. :slight_smile:

Hi,

thanks for the replies!
Now holding my breath for the new Nvidia 680. Hoping it’s not overpriced and that my power-supply can still handle things…
But good to hear that two cards can probably sit side by side…

If you have the space, it looks like it is possible to chain multiple power supplies together. Tether them with a $20 interface card, then use one to power your mainboard, and a second to power your video cards. Do a google search on “multiple power supplies” to see what I mean. When you start pushing multicard systems, you also have to be aware that you are starting to approach the upper limit of what a standard wall socket can handle (about 1650 watts in North America, IIRC).
Note that I have no personal experience with running multiple power supplies, yet.

Hehe, as my box doesn’t heat this room enough with one PSU… ;D

Well, the best solution is a bigger PSU but it’s a while until I need to descide fully if I’m to skip Maya/MR for Blender, until then I wanna see how Cycles develops over the next few months and also I want to test Cycles with a GTX680… So I’m not in a panic… :slight_smile: