UBC Home Page -
UBC Home Page -
UBC Home Page UBC Home Page -
-
-
News Events Directories Search UBC myUBC Login
-
- -
Computational Cluster  

driftwood

Hardware:
  • Front-end node driftwood + 8 computational nodes ash1, ..., ash8

  • Each node equipped with Intel Server Board SAI2 with dual Intel Pentium 3/1133 MHz CPU and 2 GB RAM


  • Storage:
  • /home – 84 GB of main storage with user quotas, backed up using HP SDLT320 tape drive
  • /data – 37 GB of overflow and temporary file storage, not backed up
  • /tmp and /global_mpi/tmp are also accessible for writing locally on each node

  • How to Use the Cluster:
    1. Use the front-end node (driftwood) to compile your application.
    2. Use the Portable Batch System (PBS) to submit your job. PBS-based Torque ver. 2.2.0 is currently installed.
    There are 2 available queues:
  • normal – no time limit on time, 16 CPUs maximum, execution hosts ash1, ..., ash8
  • test – 1 hour computational time limit, 2 CPUs maximum, execution host driftwood


  • For more information on job management, see:
  • Wiki Torque Documentation: Job Submission
  • Wiki Torque Documentation: Monitoring Jobs
  • Wiki Torque Documentation: Canceling Jobs


  • See also the PBS User Guide (pdf).

    TOP
    Last Update: 17 Nov 2010 UBC Institute of Applied Mathematics © 2009 All rights reserved.