We need regression tests for Firefox power consumption in common uses cases, e.g. browser idle with 0 tabs, browser idling on popular websites, video playback, etc.
Summary: [meta] Create a Talos test for Firefox power use → [meta] Create Talos tests for Firefox power use
I am not sure if making this a talos test would be the best use. We could add it to the talos harness, but it won't be run as part of normal talos. If we need to do something out of band, I would rather make this simpler and not have it conform 100% to talos proper.
It could also be worth to track the number of idle wakeups and the number of cycles spent executing FF code on the CPU. On Linux this could be quite feasible using the perf subsystem by tracing the context switches (sched:sched_switch), wakeups (sched:sched_wakeup) and the C-states of the CPU (power:cpu_idle). I suspect that it would be enough, at least initially, to collect that sort of data only on a single platform to provide us with useful metrics.
You need to log in before you can comment on or make changes to this bug.