Location: Brendan Iribe Center, Room 1116
This workshop is targeted at users of the Zaratan and/or Juggernaut clusters who are interested in applying or developing multithreaded or multi-process applications. This course will cover:
- techniques to parallelize code on a single node and across nodes,
- openMP for shared-memory parallel programming support on a single node,
- MPI (openMPI, IntelMPI) for distributed-memory parallel programming on single and multiple nodes,
- Optimization of SBATCH directives for parallelized code, esp. hybrid openMP-MPI code.
- Directives to control the behavior of multi-threaded & multi-process programs (environment variables, compiler directives, etc) as well as code layout for parallelism in C, C++, and Fortran programs will be discussed and worked-out examples presented.
The morning session will cover techniques to parallelize code on a single node (multithreading), the afternoon session will cover MPI and optimization of SBATCH directives.
Prerequisites
Basic knowledge of Linux/Unix and the use of our HPC clusters is assumed; we suggest to attend the workshop "Introduction to Unix", and ideally "Introduction to Compiling Software on HPC", before taking this course.
System Requirements
This is a bring-your-own-laptop event. Make sure that a terminal application (e.g. ssh or PuTTY) is installed. A modern web browser (on Windows preferably Firefox) is required for use of the OnDemand web portal.
While the registration deadline is 6 August 2023, we are currently limiting this workshop to 55 participants on a first-come, first-served basis, and registration will close once the workshop is full. A limited number of participants will allow for more detailed discussions of questions.
NOTE: DIT reserves the right to cancel the workshop for any reason with little notice.
If you have questions or need more information about the workshop, please feel free to contact us.