AWS Thinkbox Discussion Forums

Clustering 8 GPUs on Maya (4 different computers)

Hello to the forum members.

is it possible for maya to be clustered between 4 different computers by cluster i mean this : i have 4 computers consisting of 8,Geforce RTX 4090 GPUs and I want maya to detect all GPUs (8 of them) at the same time and able to render a scene with all 8 GPUs all at once !

Is that possible ? and if it is, how?
thanks

depends what you’re rendering with. (Redshift, V-Ray, Arnold GPU, Renderman XPU etc)

clustered to one job would be slow as the vram would be shared over an ethernet cable!

but some applications like V-Ray allow for ‘Distributed Rendering’. You can do this directly from Maya and run DR on the Workers so they all feed back into your scene, or set this as an ‘offload’ job, where Deadline opens Maya on 1 machine and distributes the job out.

I find it often more efficient to tile the job (if single frame) across an amount of machines, or to just let them all render a frame out. Diminishing returns when scaling across cards, so if you have 4x machines with 2 cards each then just let each render a quarter of the frame(s).

If you’re trying to cut down on licenses this won’t work as they’re bound by host.

Ant

Privacy | Site terms | Cookie preferences