1199416 - Don't run non-major forgetSkippable so likely, and run them using runnables, not timers

Status: RESOLVED INVALID

(Core :: DOM: Core & HTML, defect)

Description

[Olli Pettay [:smaug][[e-mail]]]

We could possibly increase the magical 100 suspected limit to 1000.
100 comes from https://bugzilla.mozilla.org/show_bug.cgi?id=721543

Comment 1

[Olli Pettay [:smaug][[e-mail]]]

Attachment: less_cc_wakups.diff

Ended up hacking something else too, just trying out to reduce timer wake ups.
So, use already active event loop for non-major forgetSkippables.
Those aren't as critical, so exact timing or number of them isn't that critical.

Comment 2

[Olli Pettay [:smaug][[e-mail]]]

The patch is a

Comment 3

[Olli Pettay [:smaug][[e-mail]]]

oops. Don't know what I was about to write.

Anyhow, the patch does seem to turn most of CC wake ups to use already active event loop.
But I'm not sure how to measure whether this affects to the power consumption.

njn, want to try with your tools?

Comment 4

[Olli Pettay [:smaug][[e-mail]]]

mccr8 hinted about rapl. I'll try that tomorrow.

Comment 5

[Nicholas Nethercote [inactive]]

> mccr8 hinted about rapl. I'll try that tomorrow.

You should, it's easy and will save you having to ask me all the time.

Nonetheless, I tried the patch, and power consumption was unchanged AFAICT.

Comment 6

[Nicholas Nethercote [inactive]]

> Nonetheless, I tried the patch, and power consumption was unchanged AFAICT.

But the number of CCTimerFired events drops from 309 to 35 when I start the browser on BuzzFeed and let it sit for 60 seconds. So that's good.

Comment 7

[Olli Pettay [:smaug][[e-mail]]]

Yeah, I tried rapl and couldn't see really any difference to power consumption, but on my machine the results are really noisy.

Comment 8

[Olli Pettay [:smaug][[e-mail]]]

Forget skippable scheduling is all different these days, using mostly idle time.