`conbenchlegacy` python library API documentation¶

Note: this is a legacy package, intended to be replaced by the other packages in this repo.

`conbenchlegacy.runner` module¶

class conbenchlegacy.runner.Benchmark¶

Bases: ABC

abstract run(**kwargs)¶

Parameters:

all (boolean, optional) – Run all benchmark cases.
case (sequence, optional) – Benchmark options sequence (rather than individual params): [<option1>, <option2>, …, <option3>]
iterations (int, default 1) – Number of times to run the benchmark.
drop_caches (boolean, default False) – Whether to drop caches before each benchmark run.
gc_collect (boolean, default True) – Whether to do garbage collection before each benchmark run.
gc_disable (boolean, default True) – Whether to do disable collection during each benchmark run.
run_id (str, optional) – Group executions together with a run id.
run_name (str, optional) – Free-text name of run (commit ABC, pull request 123, etc).
run_reason (str, optional) – Reason for run (commit, pull request, manual, etc). Probably will be used to group runs, so try to keep the cardinality low.

Returns:

(result, output) – result : The benchmark result. output : The output from the benchmarked function.

Return type:

sequence