Key Information

Tutor: Dr Katerina Michalickova
Course Level: Level 2
Course Credit: 1 credit
Prerequisites:
 Basic knowledge of the Linux command line. You need to know the file management commands (pwd, mkdir, cd, cp, mv, rm, cat, head, tail) and the nano text editor.
Course Duration: 2 x 2 hour sessions
Format:  Remote or Microsoft Teams session with slides and plenty of hands-on practice

Course Resources

Pre-course setup:

To be able to join the hands-on work please complete the following steps. It is important that you start a few days before in case you need to troubleshoot

  • Gain access to the Imperial HPC resource - staff can self-register, students have to ask their supervisor to register them online (please see instructions).  If you have problems registering, write to the RCS support.  It may take one day from the registration to getting access.
  • To access the HPC resource from the campus, use Imperial-WPA. To access from off-campus, follow the Unified Access instructions

If you come to the class unable to log into the HPC cluster, you will not be able to follow in an effective manner. The instructor will not be able to give you access on a short notice.

This workshop is for students who wish to use the high performance computing resource at the College (the HPC cluster). The pace of the course is suitable for anyone who knows basics of the command line and has no prior experience with HPC. Students who have used the HPC system may also benefit by formalising their knowledge and expanding their overview of parallel methods. Although, the course contains material specific to the local resource, it provides generic skills portable to resources at other universities. 

The Imperial HPC cluster is a central resource that covers a large part of the College’s scientific computing needs.  You will learn how to set up a project and then proceed to deploying simple computations. Next, we will consider data parallelism that is especially useful if you find yourself processing a large number of files or repeating the same task many times over with varying inputs.  Finally, we will move onto the fundamentals of parallel computing and demonstrate how to deploy parallel programs on the resource.  The class is entirely hands-on, you will follow the instructor and will have time for independent exercises. 

Syllabus

  • Systems at Imperial
  • Log in
  • File management
  • Software management
  • Resource manager
  • Computation parameters
  • Deployment scripts
  • Serial jobs
  • Data parallelism
  • Quick introduction to Open Multi-Processing
  • OpenMP deployment on the cluster
  • Quick introduction to Message Passing Interface
  • MPI deployment on the cluster
  • Parallel Python examples

Learning Outcomes


On completion of this workshop you will be able to:

  • Set up a project on the HPC resource
  • Combine knowledge of bash scripting and queue system to write deployment scripts
  • Use Imperial HPC systems for serial, multiple serial and parallel jobs
  • Ask for help in an informed way

 

Dates & Booking Information

There are no further sessions taking place this academic year. Course dates for 2024-25 will be available to book from late September.

To book your place, please follow the booking process advertised on the main programme page