Open
Description
Describe the current state
The benchmarks are running as tests, their output going to be stored in stderr
because the benchmark library is configured to do so.
This is not allowing to compare results effectively, or for long term, because the results are only stored in GitHub artifacts for 60 days.
Describe the desired state
Improvements that we should aim to get:
- add a build argument
-Dbenchmark-output=somepath.txt
and inject it into the build and the benchmarks step to store the results into a file - include the
--test-filter
argument (if possible) or use a second build argument (e.g.-Dbenchmark-filter
) to only run a subset of benchmarks while developing a specific signal or module