Our render machine has four GPUs and I want to use it as two 2GPUs slaves for animations jobs and one 4GPUs slave for big/huge stills.
As for now I created three slaves for each machine ( two with 2GPUs each and one with 4GPUs) and assigned the 2GPUs to the animation pool and the 4GPUs to the still pool so when the artist submit the job can choose which pool based on the type of job to use.
The problem is that when I submit an animation the job uses animation slaves and still slaves…
Can I somehow tell deadline to not use a still slave if it’s an animation job and vice versa?
I really-really-really recommend checking out this blog post! Its a great resource to show you how to configure your workers and jobs.
Short version is you’d submit your still jobs to the ‘still’ group, and have the 4GPU worker in the ‘still’ group. That should cover what you’re looking to do. But really, check out that blog post its worth your time.
Hi!
Just an update. I set up pool and groups as @Justin_B suggested.
Now everything is more clear and linear, we have project_A, project_B, etc etc pools and two groups (animation and still).
Now we are facing another little issue. We have some machines that have four GPUs and three workers running on each one:
worker01 -> 2GPUs animation group
worker02 -> 2GPUs animation group
worker03 -> 4GPUs still group
What we would like is that when worker01 and worker02 are busy rendering one frame each worker03 doesn’t pick up any job and viceversa if worker03 is busy rendering a still image worker01 and worker02 don’t pick up any job.
Now the problem is that if the two animation workers are rendering and someone submit a still job worker03 starts rendering and mess up everything…
Is there a way to set up a rule that says if this machine is already busy working on a animation group job don’t take any other group job and viceversa?
FYI All our projects are Maya Redshift or Cinema4D Redshift
