Hi Telemeters (especially reviewers),
tl;dr Android Telemetry Benchmarks now act like unittests:
chromium.perf runs everything that's enabled.
Thanks to Tianyin, the Android perf bots are now using Recipes. This
has allowed us to enable device sharding (the Nexus 5 now shards
across 8 devices) and to get the list of benchmarks to run from
'run_benchmark list'.
This means that we no longer need to edit master configs and restart
the master to add or change benchmarks. So authors and reviewers
should think about where a benchmark should run and use the
@benchmark.Enabled/Disabled annotations accordingly.
Desktop bots are next on the agenda[1].
-Tony
[1] Help out or follow along at
http://crbug.com/388885