2022-07-28 –, Purple
Have you ever wondered how many FLOPS your CPU or GPU actually performs when executing (parts of) your Julia code? Or how much data it has read from main memory or a certain cache? Then this talk is for you! I will present LIKWID.jl (Like I Knew What I'm Doing), a Julia wrapper around the same-named performance benchmarkig suite, that allows you to analyse the performance of your Julia code by monitoring various hardware performance counters sitting inside of your CPU or GPU.
In my talk I will first lay the ground by telling you everything you need to know about hardware performance counters and will then introduce you to LIKWID.jl. Specifically, I will explain how to install LIKWID, how to use LIKWID.jl's Marker API, i.e. how to mark certain regions in your Julia code for performance monitoring, and how to properly run your Julia code under LIKWID. We will then use these techniques to analyse a few illustrative Julia examples running on CPUs and an NVIDIA GPU. Finally, I will discuss potential pitfalls (e.g. when benchmarking multithreaded Julia code) and future plans.
Disclaimer: LIKWID.jl works on Linux :) but not on Windows or macOS :(
Theoretical Physicist - Scientific HPC Advisor - National High Performance Computing (NHR) - Paderborn Center for Parallel Computing (PC²)