Guide To Guillimin - bregord/emcee-on-calcul-quebec GitHub Wiki

Guillimin is McGill's own high performance computing cluster. This page will outline particularities about the cluster, as well as provide links to resources.

##Running Jobs

###Queues Available There are several queues available on the Guillimin cluster. However, the only one a user will typically interact with is the default metaq. If the user specifies metaq, or doesn't specify a queue to use at all, Guillimin's task manager will automatically infer which queue to put the job in depending on the memory requirement and requested number of nodes and processes.

For testing, the debug queue should be used. It runs jobs on a maximum of 4 cores for up to 30 minutes.

The following queues are also available:

  • Serial Workload (SW) - for serial jobs and "light" parallel jobs, memory: 3GB/core

  • High Bandwidth (HB) - for massively parallel jobs, memory: 2GB/core

  • Large Memory (LM) - for jobs, requiring large memory footprint, memory: 6GB/core

  • Serial Workload (SW2) - for serial jobs and "light" parallel jobs, memory: 4GB/core.

  • Large Memory (LM2) - for jobs, requiring large memory footprint, memory: 8GB/core.

  • Extra Large Memory (XLM2) - for job, requiring a very large memory footprint, memory: 12, 16 or 32GB/core

  • Accelerated Workload (AW2) - for jobs that use GPUs or Intel Xeon Phi accelerators (4 or 8 GB/core)

###Job Submission One other quirk of guillimin is that scipts utilizing MPI should be run with the "mpiexec" command in the submission script. For example:

  mpiexec -n 36 python exampleProgram.py

Where the -n argument is used to specify the number of cores to distribute the computation across. This should never be more than the number requested in the top of the submission script. If no -n argument is used, the task manager will use the number of cores requested by the submission script.

##More Information More information regarding usage policies, hardware and software specs, as well as tutorials and workshop schedules can be found on the McGill University's Center for High Performance Computing website