Closed mlaveaux closed 3 months ago
hyperfine is undesirable since it measures total time instead of only rewriting time. However, the benchmark task should be adapted to aggregate several runs, but it should stop when a process fails.
Made my own repeat and compute the average, not as sophisticated but should be fine for simple benchmarking. The benchmark setup could be further improved by also allowing the REC input files instead of mCRL2, but this is a separate issue.
We should use hyperfine (or some other aggregation) tool for benchmarking the rewriters, and write the results to a JSON file to be able to easily merge the results into a single table.