You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Performance information should be saved for each code change (commit or PR) so we can identify performance effects. This can be done in different ways, but brainstorming here are our ideas:
- each log file written out by Haiying's derecho ci/cd workflow has a unique identifier so we can track performance. It's necessary that the names of the tar files are easy to associate with the commit/PR
- At the very least, we can pipe this information to the terminal (as well as log) so that it is saved in github actions and can be referred back to
The important information we've identified so far is:
- minimize time (total time to run solver). This is just to start with - in the future we should collect the detailed timing. 'minimize time' neglects everything except the solver.
- number of iterations
- date/time it was run
- maybe number of processors? Name of system? ex: output from nvidia-smi
- In the future, capturing memory usage
The text was updated successfully, but these errors were encountered:
Performance information should be saved for each code change (commit or PR) so we can identify performance effects. This can be done in different ways, but brainstorming here are our ideas:
- each log file written out by Haiying's derecho ci/cd workflow has a unique identifier so we can track performance. It's necessary that the names of the tar files are easy to associate with the commit/PR
- At the very least, we can pipe this information to the terminal (as well as log) so that it is saved in github actions and can be referred back to
- minimize time (total time to run solver). This is just to start with - in the future we should collect the detailed timing. 'minimize time' neglects everything except the solver.
- number of iterations
- date/time it was run
- maybe number of processors? Name of system? ex: output from nvidia-smi
- In the future, capturing memory usage
The text was updated successfully, but these errors were encountered: