HPC Tutorial No1

From HPC users
Jump to navigationJump to search

Material related to the first tutorial on "High Performance Computing" (HPC) which took place at the University of Oldenburg from 19.11.2013 to 21.11.2013. The tutorial was split into three sessions. The first two sessions featured the parts I and II (listed below) and were held at:

Seminar-Room: W04 1-162:
Tue, 19.11.13 - 14-16 Uhr
Wed, 20.11.13 - 16-18 Uhr

The third session (part III) comprised practical exercises which were meant to illustrate some of the content presented in the earlier parts and was held at:

Computer-Lab: W01 0-008:
Thu, 21.11.13 - 09-12 Uhr

The target audience of this 1st HPC tutorial were new Users of the local HPC system, for whom, in order to benefit from the tutorial, the skills of reading and writing C-programs were of avail.

The full program of this 1st HPC tutorial reads:

Part I

1. Cluster Overview:

  1. System Overview
  2. Available compilers
  3. Modification of user environments via "module"
  4. Available parallel environments
  5. Available Libraries


  • The slides related to the presentation can be found here.

Part II

2. Introduction to the usage of SGE:

  1. General Job submission (specifying job requirements)
  2. Single Slot jobs (how to compile submit and monitor status)
  3. Parallel Jobs (openMPI, impi, smp)
  4. Monitoring and Controlling jobs (qstat, qrsh, qacct)

3. Debugging:

  1. Compiling programs for debugging
  2. Tracking memory issues
  3. Profiling

4. Misc:

  1. Logging in from outside the university
  2. Mounting the HPC home directory
  3. Parallel environment memory issue
  4. Importance of allocating proper resources


Part III

5. Exercises (Computer-Lab):

  1. Simple example programs
  2. Practical aspects not covered in part II


  • The example programs can be downloaded here.
  • A brief documentation of some of the examples (i.e. those which are notdiscussed in the Brief introduction to HPC Wiki page) can be found here.
  • In part III we also discussed how to monitor the current resource usage for a running job using an interactive session. A brief writeup of the whole procedure can now be found here as part of the Brief introduction to HPC Wiki page.