Still image distributed across the farm using Qube!

[dball]

Currently we render still images via the command line in OS X using Mental Ray standalone . The command would look something like /Applications/Autodesk/mentalray3.5.6/bin/mentalrayrender -v 5 /mr_files/render_this.mi.
This will take my file 'render_this.mi' then read the IP address in the .rayhosts file and use all those machines to render the single image.
I would like this same behavior when I submit the job to Qube!, is this possible? I'm aware you can use MR satellite, but that's limited to 8 cpu's I need to use our 25 standalone licenses.
Thanks for your help.

Dave Ball

[shinya]

Hi Dave,

The mentalray jobtype does indeed support standalone satellite renders, using arbitrarily many nodes.  It will automatically allocate nodes and use the "-hosts" option to the "mentalrayrender" command. 

For example, if you want to use 25 nodes, then on submission, you will want to set the "CPUs" field to, 25, and then also set the "reservations" field to "host.processors=1+" to make sure that each node will only take one subjob.

[dball]

Shinya,

Thanks for the reply. Can you just clarify, in your example those 25 nodes will all be contributing towards rendering a single still image. I'm not looking to render 25 frames, just one very large print image using 25cpus.
If that's correct then fantastic I can push forward with Qube!

Dave