Opened 10 years ago
Last modified 10 years ago
#7082 new Feature Requests
Add support for benchmarking
| Reported by: | Owned by: | Gennadiy Rozental | |
|---|---|---|---|
| Milestone: | To Be Determined | Component: | test |
| Version: | Boost 1.51.0 | Severity: | Not Applicable |
| Keywords: | test benchmarking | Cc: |
Description
One useful thing with boost test would be to run the tests in a "benchmarking mode". The general idea is to be able to run many iterations of all tests to get an idea of their performance. I believe the following changes would be sufficient:
- BOOST_AUTO_TEST_CASE could have an additional optional parameter: number_of_iterations. This is set by the user manually. For example: BOOST_AUTO_TEST_CASE(my_test_case, 10000). If not passed in, it would default to 1.
- A new command line parameter would trigger benchmarking: say --benchmarking. When passed in, all tests would run number_of_ iterations times.
- the elapsed time for a test would then be the accumulated time taken by all iterations
This would be great for nightlies.
This issue was discussed a bit on the mailing-list on this thread:
Note:
See TracTickets
for help on using tickets.
It just occurred me that, as a side effect, setting number_of_iterations to zero is also a great way of ignoring tests. However, Gennadiy has stated there is already a solution for this particular problem:
http://lists.boost.org/boost-users/2012/07/75075.php