You guys have more experience than Ihave and I’m just curious.
Is there any benefit to building arender farm for CPU rendering over using cycles for GPU rendering?
I just find a new render farm case here: www.modularcreation.com and I know it is more expensive but maybe it is more useful topopulate one of these than always buying the best available GPU.
There are obvious benefits like morememory and continuous operation for home made render farm, but thereare benefits for GPU rendering as well.
The main advantage of CPU is memory support, as You can get as much as 32GB of ram on single CPU motherboard compared to 2-6GB on video cards that cost $600 or less.
Also, many of Cycles features aren’t implemented on GPU yet, and some may never will, so a CPU setup will allow You to use every features Cycles has to offer!
On the plus side, GPU cost way less then CPU, and on many motherboards You can stack as many as 4 graphics cards, lowering the cost, and GPU rendering is allot faster then CPU.
It all boils down to your needs, if You need to render huge scenes, with tons of mappings and millions of polygons, or if You need features like volumetrics (last I checked these weren’t supported on GPU), then a CPU render farm might be the best solution for You, if not, then just stack a bunch of GTX 780 with 6GB of ram, a good power supply, and You’ll end up with a small, but fast GPU render farm, for allot less, and much more powerful then a 6 CPU render farm!
Echoing CGharb here, it depends on what you are trying to do. If you are doing a lot of animations, then having multiple nodes in a render farm can be advantageous. But for large stills, there isn’t as much benefit.
As far as cpu vs gpu, The speed benefits of a gpu are hard to ignore, especially at a much lower pricepoint than equivalently performing CPUs. It really depends on how complex what you are rendering is and if it will fit in the video ram.
