Changing Machine configuration including GPU-affinity

[Gotz_Bongartz]

Hi there,

we have two different renderers in our production: Octane and Redshift. From a pipeline-TDs point of view their only common attribute is that they are both GPU renderers.

We always habe the same render-nodes, it’s ubuntu machines with 8 GPUs.
The issue is that for the redshift-jobs I want to have 8 workers doing 8 Rendersjobs in parallel on one machine, thus I fired up 8 deadline-workers and assigned GPU-affinity to one for this.
For the Octane rendering however I need to use two machines with each 4 GPUs assigned via GPU-affinity.

Is there a procedure in Deadline for that? Scripting this seems to be a difficult task actually, I’d love is that is something that can be solved “by design”.

Thanks!

Goetz

[anthonygelatka]

+1 for this, see so many places using a mix of GPU renderers, usually having the same requirement as yourself, but with no simple solution.

Would be nice to have a much more flexible method to allocate resources dependent on job, without having to recode everything yourself

[Gotz_Bongartz]

One of the issues connected to this is that you’d have to wait for a job to start while it blocks the GPUs. Idk, maybe deadline is even doing this without complicated scripting.
It’s not that I’m not able or willing to script that but I just want to avoid any kind of custom scripts for my pipeline.