Code profiling and performance optimization on the RCC clusters – an overview
Abstract:
High performance computing (HPC) plays a crucial role in a wide range of fields, from scientific research and engineering to finance. It involves complex simulations, large-scale data analysis , and intensive CPU or GPU computations.
This workshop will focus on techniques for analyzing and optimizing code performance, with an emphasis on the underlying hardware. Through code profiling, participants will learn how to identify bottlenecks, optimize resource usage, and improve the overall application efficiency.
-
Do you want to optimize the performance of your code?
-
Curious about identifying bottlenecks in your code?
-
Never heard of “profilers”, or not sure how to use profiling tools, or what options are available?
-
Have you heard of or used tools like “Tau” or “Intel VTune” but want to know more?
-
Do you work with code written in R,Python, C++, or Fortran and want to language-specific profiling or optimization strategies ?
If the answer to any of these questions is “yes” – this workshop is for you.
We will introduce code profiling on the Midway HPC system, covering both basic and advanced profiling techniques. Participants will use hands-on exercises with TAU and cProfile. Through guided exercises, you will learn to interpret profiling results and develop strategies for performance improvement. .
.
Objectives:
-
Have an understanding of profiling tools available on Midway and their differences.
-
Gain hands-on experience with code profiling.
-
Leave the workshop with skills to profile your own code based on demonstrated examples.
Level: Intermediate
Duration: 2 hours
Prerequisites: All participants are encouraged to bring a laptop with a Mac, Linux, or Windows operating system. Having an RCC account will be helpful to perform the exercises on Midway3.