[Paul Carter]

Paul Carter

29th April 2013 at 4:10 am #14504

I’m trying out Muster 7.07 and Maya MentalRay 2013 and trying to limit the number of cpu cores being used. In the behavior section of the particular instance, I’ve created an instance in the window and set the cpu affinity mask to be half the number of cores I have available (6 cores selected). Hyper threading is on, so I have 12 cores available on a 6 core machine. It still renders with all cores. I’ve had a look at the command line it sends, it says -rt 0. Is there something in the template that needs changing to take note of the thread affinity?

I’ve been having a play with the behavior window and creating more than 1 instance. I’m not sure how this is supposed to be used. With a multi core machine, can you have 2 different instances with different thread affinities, that can be in different pools?

[Mike Thomson]

Mike Thomson

21st May 2013 at 4:39 am #15220

We’re having similar issues. When we manually limit the thread count to half the hyper threaded core count (12 of 24) , the computer will render at 50% capacity, but the render is scattered across all the cores, not 1-12 (of the 24 available).

Our render computer setup:
– 2×6 cores (12 physical cores)
– Hyperthreading turned on
– 24 HT cores
– 64 GB ram
– two instances of muster client running
– instance 1 set to cores 1-12
– instance 2 set to cores 13-24
– both instance templates limited to 12 cores

[Emanuele Zambrano]

Emanuele Zambrano

1st June 2013 at 3:29 pm #15224

I believe that you should use the additional parameter “-rt” to specify how many threads you want to use for a job.

You can define this parameter at template level for every instance (instance->configure->template->maya mr->maximum threads) or with the additional flags of the job (from the submission panel->additional flags).

The instance affinity mask is a platform specific parameter (only work in windows and linux) and the maya mr process can override this setting by itself.