RT and DR CUDA Hardware?

[Streetwise]

I'm building a couple of 12-core Servers to use for DR. Since my client machine is limited to a single 1GB CUDA card, would there be any benefit of outfitting the new servers with CUDA cards? I do large single frame renders, but like to use Active Shade when I can to see what's going on with lighting, etc. I then use CPU's for the final renders at that resolution.

Secondly, I noticed that by default, the ActiveShade Engine is set for CPU. Is that for any particular reason?

Thanks.

[vlado]

There are some caveats in getting the GPU to work on the render servers; first, you need to actually have something plugged into the video output of the card, second, unless they are Tesla cards, you need to be physically logged onto the machine and run the V-Ray RT render server manually (or at least not as a service) for the GPU to be accessible to V-Ray.

Best regards,
Vlado

[Streetwise]

Interesting. So aside from those caveats, is there any real advantage? Or would it be better to just have a better client machine with a couple of beefy CUDA cards?

[vlado]

Having the cards locally is always better, but there is often a limit on how many cards you can plug into one machine. DR works quite well too.

Best regards,
Vlado