Julia for research on the HPC

A brief lightning talk on using Julia for research, allowing rapid prototyping of ideas alongside high performance parallelism at any scale.

Resources

This repository contains the raw source code (along with the raw benchmark timings and graphs) showing how Julia can be used in parallel applications. If you are coming here from the blog, checkout the blog subdirectory!

How to run the code

You will need to install Julia either directly from the website or install via JuliaUp. After that, clone/download this repository and lauch a terminal in this folder. From the terminal, run:

julia --project

This will open the Julia REPL (Read-Evaluate-Print-Loop), where you can interactively run code. To install all the packages needed you can access the package manager by pressing the ] key. You can write instantiate to download the packages:

] instantiate

or, alternatively:

using Pkg;
Pkg.instantiate();

You can press backspace to get out of the package manager in the REPL. If you want to use CUDA, you can download the CUDA binaries by running:

using CUDA
CUDA.versioninfo()

If you are following the blog, you have to move into the blog subfolder by running:

cd("blog")

Then you can get all of the functions in blog/main.jl by typing:

include("main.jl")

Explore this source code in your text editor (I recommend VS Code with the Julia extension) and you can load the timings and run the code, and experiment by changing things.