Description
This 2-day advanced level on-line course will give attendees the skills needed to understand the system architecture of platforms such as ARCHER2 and the new Cirrus system, the programming environment and furthermore explains how application performance can be explored by means of profiling tools and which optimisation strategies are available. The course comprises lectures, examples and follow-along demos. It covers both CPU and GPU usage.
Prerequisites
- Understanding of building and running HPC applications on supercomputers
- Familiarity with Linux command line usage
- It would be highly advantageous to be an existing user of a system such as ARCHER2
Learning objectives
- Short but comprehensive introduction to the Cray EX architecture and programming environment to the point of being able to build and run applications
- Performance tools available to profile applications: includes Cray and AMD performance tools for CPU and GPU applications
- Other performance tools which might be of interest
- Understand how different software and hardware subsystems work and could be optimised, topics to be covered are: application placement, application compilation, Cray MPI on Slingshot, GPU optimisation and I/O.
- Considerations of Python applications and Frameworks with a view to understanding which tools are relevant (but note the focus of the workshop is mainly on traditional HPC applications) The course will use ARCHER2, its GPU development platform and the newer development environment on Cirrus. The workshop will use AMD GPUs but other than use of some specific tools, the approaches are generic to both AMD and NVIDIA platforms.
Requirements:
Participants must bring a laptop with a Mac, Linux, or Windows operating system (not a tablet, Chromebook, etc.) that they have administrative privileges on.
They are also required to abide by the ARCHER2 Code of Conduct.
Timetable:
Details to follow