Will network rendering be available in the up coming SP1?
Our all-new Network Rendering workflow is planned to become available with Visualize 2017, shipping this October. It's quite powerful!
Brian, will the new networking solution allow networking between multiple PCs or a sever based GPU cluster, or both? We are possibly adding a few seats of Visualize in the near future and are debating if we are continuing with powerful, multi GPU PCs or if we will move to a server GPU solution that can be used by everyone. Any info you can give on the networking solution would be great.
BOTH! We are really excited about this newly revamped Network Rendering solution. It doesn't have a minimum set of hardware requirements...so it can be installed on anything from a laptop, to an extra machine lying around the office, to a dedicated render farm of several machines, custom GPU server, or even the NVIDIA Quadro VCA(s). Even though GPUs are way more powerful than CPUs, our network rendering solution will run on both. So if a company already has an old existing CPU render farm, our network rendering will use that as well.
More info will be "teased" leading up to the SW launch event in September, on all the new 2017 Visualize stuffs.
For best performance, it's recommended to install on a dedicated machine, rather than a co-worker's laptop To help answer your question, albeit with more ambiguity, it depends on the size of the Visualize userbase and the type of content generated. If animations, 360 spins, interactive VRs are something that your Visualize userbase normally generate, then completely offloading them to a dedicated render farm
ALSO on the network rendering topic, we have a little present coming for all the Visualize Professional users out there...
Thanks Brian, I look forward to hearing more about the new network solution.
"For best performance, it's recommended to install on a dedicated machine, rather than a co-worker's laptop"
I assume this means the network solution has no control over the amount of resources it uses from each networked machine? It would simply use all available resources on systems setup with the Visualize network solution. In one possible scenario, I would share my triple GPU system with some lower end machines for rendering but I wouldn't want them bogging me down while doing other non-rendering work, as its my daily PC. Would they only have the CPU/GPU/Hybrid render options or would there be anything more advanced where I could allocate 2 GPUs for network rendering and nothing else?
I understand if you cannot talk directly to this, at this time.
As of now, our network rendering solution will utilize all the available hardware in the machine where it is installed. I will check if users can control whether it's CPU/GPU/Hybrid mode, or if it just uses all hardware regardless.
We are pushing to get this network rendering solution available in Beta program over this summer! And real-world feedback like yours would help us make our network rendering solution better.
Neil - you can control which resources are available to rendering in the Nvidia control panel providing you are running professional cards (quadro or tesla). You can say that only 2 of the 3 cards may be used for rendering, or shut off all cards and force the CPU to be used if you want.
say we did have pro cards, turning off cards for rendering in NCP would control the resources available to the network but would also limit the number available for me to locally render. I don't have experience controlling these via NCP but I assume I would be turning the control on and off frequently unless you can set all GPUs locally available for rendering and a subset of GPUs for network rendering separately.
Regardless of all that, we run GTX based GPUs here for Solidworks and Visualize without issue. Since there are no minimum requirements, as Brian said, hopefully future versions of the networking solution could allow resource control for non pro card machines and even lower end machines. I could see cases were you would want to allocate 100% of dedicated network PCs, 20% of a bunch of daily use low end PCs, or 50% or more of high end daily use PCs. I think there are a number of scenarios where partial allocation could make the network solution really useful. Just my 2 cents I guess.
Retrieving data ...