Performance Analysis performed during WaSH development
They were written and run on the University of Warwick's specific batch compute system. Adjustments must be made if one wishes to run these elsewhere.
The Python notebook parses and collates the files generated by the run scripts.
It also contains a vast array of representations of the data, which can be adjusted to visualise any other collected data. Running the scripts should automatically make the slurm file with the results readable to the notebook, and the numbers will be available once the script has its time output.
One run script exists for each number of nodes and cluster combination, since these scripts do not easily take in node numbers. This may be an idiotic way to do it, but slurm wasn't taking the flag from the commandline as intended when I tried.
The slurm files are individual runs of WaSH or SPH-EXA, with the runtime in seconds printed first, followed by the remainder of the program's standard output.