OmniSciDB  4201147b46
 All Classes Namespaces Files Functions Variables Typedefs Enumerations Enumerator Friends Macros Groups Pages
Conbench Client is a conbench client script for running and publishing benchmarks to a running conbench server.

You don't need to run this manually. It should be taken care of by Jenkins. The below instructions are in case you would like to run the client elsewhere.


  • Install conbench via pip install conbench.


  • Run conbench --help from the directory with the script. The output should look something like ``` $ conbench –help Usage: conbench [OPTIONS] COMMAND [ARGS]...

    Conbench: Language-independent Continuous Benchmarking (CB) Framework

Options: –help Show this message and exit.

Commands: StringDictionaryBenchmark Google Benchmark list List of benchmarks (for orchestration). version Display Conbench version. ``

  • Create a.conbenchfile in the same directory with contents: `` url: email: YOUR_EMAIL password: YOUR_PASSWORD ``` You may sign up for an account at .
  • The test corresponding to the benchmark name, e.g. StringDictionaryBenchmark, must be compiled and runnable in the current repository.
  • The current git commit, as output from git show -s --format=H, is the commit that will be associated with the recorded benchmarks. This commit must be known by github. Do not run conbench on an unpublished commit, otherwise conbench will not be able to obtain necessary data on the commit and will result in errors in the reporting.
  • The build_dir is assumed to be build-$GIT_COMMIT. It will cd to its Tests subdirectory, run the benchmark, capture and parse the output, and report the results to the conbench server.
  • View and analyze the results at