Author Topic: Optimisation  (Read 4826 times)

Jabar

  • Jr. Member
  • **
  • Posts: 8
Optimisation
« on: April 21, 2008, 07:24:31 PM »
Hi Eric,

I'm finding my renders taking longer with qube than with renderpal. I see that there is talk of optimising the supervisor with the max_threads and max_clients settings but I still don't quite understand them. Is there a link or resource that might better explain them?
I'm finding that frames that should take about 30 seconds are taking up to 15 minutes.
Does qube start a new render for every frame? I am not using batch mode.

Brad

Jabar

  • Jr. Member
  • **
  • Posts: 8
Re: Optimisation
« Reply #1 on: April 22, 2008, 11:23:30 PM »
Thanks for that eric, makes a lot more sense now. For any other noob thats struggling:

5 quad core processors (That would be the actual hardware :-)
set the cpu setting in the submission dialog to 10 to open 10 'slots' for the job
set the host.processors=2,host.memory=1000 (I have 4 gigs on each render node but I want to run two subjobs on a node so i am limiting the subjobs to 1 gig each, with a little extra)
So this means I get to render two frames at a time on each render node!
I still think I'm missing something in terms of mem management. One or two of the hosts seem to hold onto a lot of memory after each render. Can this have an negative accumulative effect in some way? Shouldn't there be some sort of purge between renders?