Benchmark

The directory Benchmark contains a simple benchmark system.

Usage

• Before starting to change the code, run the benchmark
• If you are done with your work, or if you are stuck, commit and run the benchmark again with the new code version
• Compare the results before/after

In most cases you expect to have not broken/changed the code when updating say, from revision 0001 to revision 0002, during your xmas holiday project. So to be sure, just diff the benchmark result:

$ diff bench_0001/181225/bbh_q1_s0s0_10Hz/waveform.txt diff bench_0002/181225/bbh_q1_s0s0_10Hz/waveform.txt 

In other cases you do expect a difference because you fixed the code and/or made it better. So you will need to inspect the data and see if you have done a good work.

PLEASE USE THE BENCHMARK, it is easy and can safe you/us lot of time.

What to run?

Parfiles for several relevant test cases are collected in the dir

$ C/par/

Those parfiles can be added in the list of FILES at the beginning of the makefile in the Benchmark directory.

How to run

Type

$ cd Benchmark
$ make build # prepare directory and compile exe
$ make run # run the test

or simply

$ make # build and run

Notes:

• The benchmark output folder is named after the last git revision, a subfolder with the date is created.
• Information about the git revision are stored in a text file; the whole source code is backupped in the output folder.