11. Hunting Down Heap Allocations [HPC in Julia]

In this video we will take a deep dive into optimising away heap allocations from our code. In previous videos we discussed garbage collection, which can have a huge impact on the performance of our algorithms, especially if the number of heap allocations is excessive. Today, we'll look at a practical optimisation example and try to improve a N-body physics simulation that aims to render in real time. We'll be able to clearly see how garbage collection impacts the performance of our simulation and then work to remove all allocations from our physics code - resulting in a much smoother animation. If you are interested in learning more about this topic, below are a few links that you may find interesting: https://www.youtube.com/watch?v=BFvpwC8hEWQ https://docs.julialang.org/en/v1/manual/performance-tips/ You can find the code in the video in the GitHub repository linked below: https://github.com/MPAGS-HPC-in-Julia/reducing-allocations-example Timestamps: 00:00:00 Introduction 00:01:14 Case Study - Physics Simulator 00:03:20 Physics Simulation Code 00:04:28 Slicing Notation 00:04:57 Adjoint Operator 00:05:27 Broadcasting ("Dot") Notation 00:06:49 Benchmarking 00:09:56 Profiling Memory Allocations 00:16:05 Views 00:19:04 Preallocation 00:22:20 Broadcast ("Dot") Fusion 00:25:53 Optimising "acceleration" function 00:28:17 StaticArrays.jl 00:32:03 Simulation (No Optimisations) 00:32:37 Simulation (with Optimisations) 00:33:10 Outro This module was designed as an MPAGS (Midlands Physics Alliance Graduate School) module and aimed at postgraduates and early career researchers. Useful links: Git/GitHub for Researchers: https://www.youtube.com/watch?v=CuOmaUS1FnM Course Information Website: https://jamiemair.github.io/mpags-high-performance-computing/overview/ Julia Documentation: https://docs.julialang.org/en/v1/ MPAGS Registration: https://warwick.ac.uk/fac/sci/physics/mpags/registration/ Course Notes: https://jamiemair.co.uk/courses/hpc