csf-testing-teaser.png

About this project

This project looks to unify our approach to automated testing, making it easier for our software engineers and HPC support staff to support. This approach will use GitHub Actions to run testing workflows on GitHub resources and on our on-prem cloud, Cirrus. The automated testing workflows will run linters, code coverage, model builds, and unit, system, and performance tests on CPUs and GPUs.

Why this work is important

Automated testing will ensure software excellence and confidence in its quality. This workflow replaces the first-pass testing that our Software Engineers perform on each pull request. When a contributor submits a code change to GitHub, this workflow automatically flags and notifies the contributor of issues, allowing software engineers to bypass this step and spend time on contributions that pass the first round of testing.

How does this fit within the CSF

  1. Fits within Better Practices
  2. Tests for code portability across different architectures
  3. Integrates the use of AI/ML

NSF NCAR Labs involved in this project

CISL
MMM
CGD
ACOM
HAO
 

External partners

DOE
NOAA
 

More information

MPAS-A CI repository: https://github.com/ncar/mpas-model-ci