Post

PASC

Posted
By Robert Caddy 1 min read

Platform for Advanced Scientific Computing (PASC) Conference

I spent most of the week remotely attending PASC. It was really interesting and got me connected to the US-RSE group.

Optimize Characteristic Reconstruction

I refactored the characteristic reconstruction so that the eigenvectors are only computed once. This and Trey White’s suggestion of adding __launch_bounds__(256) to a few kernel definitions improved performance by about 40%.

Paper

I made the plots for the MHD Blast Wave and am mostly done with the Orszag-Tang Vortex plots.

Other

  • Profiled Cholla with HPCToolkit
  • Opened a PR for the linear wave convergence test. PR #305
Paper Writing PPMC
This post is licensed under CC BY 4.0 by the author.