First and foremost, I'm new to PDG and setting up a hqueue render farm so hopefully there's an easy solution to my issue and I'm just unaware of it.
I've set up a hqueue server/client on a machine (the machine has 40 proc cores as opposed to my workstation machine which has only 12) and I've set up a TOPS network on my workstation computer with an hqueuescheduler node pointing at my render farm machine. I also have another TOPS network set up within my HIP file that's running the same operations but the scheduler is local because I wanted to compare speeds.
The operations that I'm doing inside of the TOPS networks are simple, I'm importing geo and then doing a 40 count wedging operation on it to warp the geo randomly and then export the warped version (a ropgeometry node). I'm also doing a ropmantra render from a camera to render out 40 screen shots essentially.
The issue I have seems to be task distribution over multiple CPUs, in my localscheduler TOPS network, I set Maximum CPUs to Use to -1, and when I cook my ropgeometry node, it will begin to cook 11 jobs out of the 40 at a time and complete the cook relatively fast. When I cook the same node in my hqueuescheduler TOPS network, the node displays only 1 active cook out of 40… despite having 40 CPUs on my render farm machine. When I set my hqueuescheduler Task Environment to either 0 or -1… or even when I manually set CPUs per Jobs, it doesn't seem to speed up the process at all. Am I missing something? Is this a permissions or licensing issue potentially?
Differences between Local cook and render farm cook tasking.
1320 1 1- tradewind2k
- Member
- 1 posts
- Joined: 3月 2019
- Offline
- chrisgreb
- Member
- 603 posts
- Joined: 9月 2016
- Offline
-
- Quick Links