GPU on the network

[Mats_O]

Hi all,

I have a setup where I work on a HP mobile workstation with a Quadro card in it, and then a couple of HP workstations (Z600) as renderslaves. For VRay RT, is it possible to put one or more GPUs in my rendering slaves and use them for RT? If it would work, what kind of card should I look at, Tesla´s or the cheaper GeForce cards? I looked at Nvidias Maximus solution where one card takes care of all the Max stuff and another all the RT stuff, and I am looking at something similar but based on my setup.

[vlado]

It is possible to use GPUs with distributed rendering; however keep in mind that if you use GeForce cards, you need to have something plugged into the video output of the cards, and also you need to be physically logged onto the machine and run the V-Ray render server from an interactive user session, so that it can access the GPUs. These restrictions are not present if you use Tesla cards.

Best regards,
Vlado

[Mats_O]

Thanks Vlado!! Continuing on the same thread I am wondering if there is any usable solution with an external GPU. I´ve heard the are external boxes for laptops that can hold a graphics card. But is it a solution that would work with Max/VRay? An external box equipped with a Tesla card for example.

[vlado]

Yes, it will work and I know that many people are using such a configuration. However, I would advise you to test the performance of such a solution vs the card plugged directly into a machine - I've heard that there may be differences (though I have not tested myself).

Best regards,
Vlado

[Mats_O]

Interesting Vlado, do you even have suggestions on such a solution, I mean what others are using?

[vlado]

Interesting Vlado, do you even have suggestions on such a solution, I mean what others are using?

No, unfortunately, I haven't used any of these myself. Hopefully someone else will be able to share their experiences.

Best regards,
Vlado