To run an algorithm for multiple graphs and collecting their results, the launcher script is used.
- alg=algorithm-filename (with or without .c)`
- df=path/to/datasets/folder (default ./data)
- make-flags="flags (e.g.
threads=t,no_ht=1,debug=1,wait_passive=1, and/orenergy=1) to be passed to themake, separated by space, and double quoted" (default empty) - program-args="arguemnts to be passed to the
programbymake, separated by space, and double quoted" (default empty) - sf=m (start from dataset m, default: 0, i.e., all datasets are processed)
- sa=n (stop after processing dataset n, default: -1, i.e., all datasets are processed)
- -ld (just list datasets, default: 0)
- -iw (include weighted graphs, default: 0)
- -shm-store (store graphs in shm, default: 0)
- -shm-delete (delete shm graphs at the end, default: 0)
As the first step, the script reads the arguments.
Datasets should be stored in / linked from a folder that is passed to the script using the df argument.
After reading the arguments, the script lists the datasets and sorts them by number of edges in the ascending order.
The script creates a log folder to store the output of running the algorithm for each dataset.
The script, also, creates a report file that contains the final results collected from each dataset.
The path to report file is passed to the algorithm using LL_OUTPUT_REPORT_PATH.
Procesing may be stopped by the script if the sa argument has been set.
The last step is to print the report file.
