C++17 Structured Binding Tutorial
Summary Structured binding is a feature released in C++17 that easily allows a function to return multiple values. This is a quick summary of the various ways to use it, including in a device func...
Summary Structured binding is a feature released in C++17 that easily allows a function to return multiple values. This is a quick summary of the various ways to use it, including in a device func...
GitHub Action HIP Builds I ended up spending most of this week trying to get the automated HIP/ROCm builds to work on GitHub Actions. The latest release of ROCm (v5.2.1) seems to have broken some ...
DeviceVector I added simple method for assigning values to locations in a DeviceVector array. I also got the DeviceVector class and GPU reduction fixes into my MHD branch. This required quite a f...
Fixes This week our long term development branch, CAAR, got merged into main. Before that could happen I needed to submit a couple of fixes for outstanding issues. GoogleTest Version Our automat...
Constrained Transport Slopes I figured out how to implement the CT slopes function this week. I was thinking about it too hard, instead of passing information like whether or not we’re upwinding, ...
Constrained Transport Slopes I’m pretty sure there’s a bug somewhere in my calculations for the slopes used to calculate the electric fields. I started trying to debug this and quickly realized th...
MHD Debugging I found and fixed the last hydro but in my MHD code this week. There was a yid-1 when it should have been zid-1 in the PCM reconstruction function. After that fix, and thorough testi...
International High Performance Computing Summer School (IHPCSS) I spend most of this week prepping for the IHPCSS. I made a poster that you can find in this post. V1835 Aql I reviewed a late dra...
Welcome! You can download my IHPCSS Poster here or viewed below. If you’ve found your way here hopefully it’s because you want to learn more about me, my research, Cholla, or the details of imple...
MHD Debugging With some help I found a major bug. Several indices had a * instead of + when calculating the shift based on the number of passive scalars. After fixing that the constant, no magneti...