934720 - Battery load higher at idle since Maverick update.

Status: RESOLVED INCOMPLETE

(Firefox :: Untriaged, defect)

Description

[Bryan]

Attachment: FF-27-Drain.pdf

User Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.9; rv:27.0) Gecko/20100101 Firefox/27.0 (Beta/Release)
Build ID: 20131104004000

Steps to reproduce:

Over the last week, I have used Safari and Firefox separately, on separate boots with 100% starting charges on my battery. Each time each browser was used under similar reproducible configurations and circumstances (my workflow doesn't change much). Running Firefox on my machine consistently reproduces similar hardware loads and battery drain that on average far exceeds Safari 7 in my comparison on my machine.


Actual results:

Hardware: late 2011 Macbook Pro, 2.4 GHz i5, 8GB RAM, SSD, 94% of batter capacity left according to Coconutbattery app. 
Boot ups are with full 100% charge on battery, Screen 50% brightness, keypad lighting OFF, bluetooth OFF, WiFi is ON, 1 program running in dock; Firefox Aurora or Safari 7. 

Today's example: 
approximately 35 minutes use on battery with 100% charge to start and indicator is showing 88%. Safari does not display this rate of battery drain with the same pages opened. 
Various screen shots of activity panel and the open tabs in Firefox are included. The screen shots are of course snapshots and CPU load varied between 2% and as high as 7% at idle. Idle in this case meaning nothing in the tabs is running (to my knowledge); ads are blocked, flash is blocked (they are mostly contextual pages) and I am writing this or looking at activity monitor.


Expected results:

In my usage over the last week, I have consistently seen Firefox produce CPU loads between 2% and 6% under similar circumstances: between 4-8 tabs that are all mostly text, no video's, with extensions adblockedge, flashblock and lastpass running. Memory loads also seem to remain consistent between 550MB and 750MB. 

Safari 7 shows somewhat similar numbers of CPU and memory in usage under the similar criteria. Safari 7 CPU usage seems to not fluctuate as much and it actually lets go of CPU load in contrast to Firefox at idle. At times, CPU and memory numbers seem to converge to a more one to one relationship, but OS X 10.9 is still reporting their arbitrary energy usage numbers as much higher and the battery usage is significantly higher using Firefox over Safari 7. 

For a period in time, Aurora was flaky in even showing up in the energy usage graph. Reboot fixed that. I would love for Firefox to be at least a little more energy conserving. I understand this may be unique to this machine, but I hope it helps button something up making FF better.

Comment 1

[Steven Michaud [:smichaud] (Retired)]

This is all rather vague -- too vague to do anything with.

But it's possible bug 932602 may make a difference here, once the work
there is finished.

> approximately 35 minutes use on battery with 100% charge to start
> and indicator is showing 88%. Safari does not display this rate of
> battery drain with the same pages opened.

What is the corresponding figure for Safari?

Please re-run this test, as follows, and report your results here.

1) Choose a single page, which you will load in both Firefox and
   Safari and include in your report.  It should be "active" (cause
   the page to redisplay periodically), but not contain any plugins.
   For example http://www.mozilla.org/en-US/ or http://www.apple.com/.

   (Later it'd also be interesting to test with plugins, but for now
   let's keep things very simple.)

2) Set your computer to not sleep or go into screen saver mode for at
   least 30 minutes.  Also disable all extensions in both Safari and
   Firefox.

3) Charge your battery to 100%, then take your computer off AC.

4) Close all other apps and run Safari.  Then load your chosen page in
   Safari, making sure only a single browser window is open,
   containing a single tab that displays the page you chose in step 1.
   Also make sure Safari is "active" (for example that it's menu is
   showing at the top of the screen).

5) Let your computer sit idle for 30 minutes.

6) When the 30 minutes is up, observe and record your computer's
   remaining charge (by percentage).

7) Go back to step 3, but this time use Firefox instead of Safari.

Comment 2

[Steven Michaud [:smichaud] (Retired)]

(Following up comment #1)

I just performed this test myself (on a Retina MacBook Pro), and found that Firefox's performance is actually *better* than Safari's!  At the end of my test of Safari, my computer's remaining charge was 93%, but for Firefox it was 97%.  The page I tested with was http://www.apple.com/.

What we need here, as in any bug report, is comparisons of Firefox with other browsers that are testable and verifiable.  Without that, we don't have a viable bug report.

Comment 3

[Steven Michaud [:smichaud] (Retired)]

I should add that I tested on the Mavericks "release" (build 13A603), with FF 25.

Comment 4

[Bryan]

I'll run what Steven suggested tonight and get back to you guys. 

I read really good reviews for Firefox on the new OS so I was surprised when I saw the battery meter taking such a dump so fast. Thats when I started comparing the two browsers more in-depth. If it was closer, I would have not said anything, but this seemed dramatic enough and I didn't see anything about it specifically in various reviews or bugs. I'll add more later.

Comment 5

[Steven Michaud [:smichaud] (Retired)]

> What we need here, as in any bug report, is comparisons of Firefox
> with other browsers that are testable and verifiable.  Without that,
> we don't have a viable bug report.

I said this already above, but it bears repeating.

Without these characteristics, a bug report is basically worthless.

There's probably not much point in just repeating my experiment from
comment #1.  But do let us know if you do repeat it and get
significantly different results.

> I read really good reviews for Firefox on the new OS so I was
> surprised when I saw the battery meter taking such a dump so
> fast. Thats when I started comparing the two browsers more in-depth.

Please translate this into a repeatable experiment that demonstrates
what you say.

Comment 6

[Bryan]

Okay, both Safari and Firefox were run for 30 min. Both single tab at  http://www.mozilla.org/en-US/ . No screen saver. 100% charge.

Safari's numbers were both 95% after 30 minutes. 

Firefox 27's numbers were 89% and 91%

Comment 7

[Steven Michaud [:smichaud] (Retired)]

Weird.  Thanks for running the test.  I'll test again with http://www.mozilla.org/en-US/, just to be sure.

Why didn't you test with http://www.apple.com/, as I did?

What machine did you test on?  Did you have any "non-standard" software running in the background?  (By "non-standard" I mean anything that wouldn't have been running on a machine with a clean install of Mavericks.)  Have you changed any of the system settings from their defaults (aside from the sleep and screen-saver settings)?

My machine is a Retina, mid 2012 MacBook Pro with a 2.6 GHz Intel Core i7 processor, 16GB of RAM and Intel HD Graphics 4000 hardware built in, plus NVIDIA GeForce GT 650M hardware on the PCIe bus.  I tested in a clean install of Mavericks (not an upgrade), with none of the system settings changed save for the sleep and screen-saver settings.

Comment 8

[Bryan]

Well, I didn't do apple.com just simply because I ran the 4 runs with Mozilla.com and with homework ran out of time. I just wanted to have more than one run to see if there was any variation. I will actually run the apple.com trials tomorrow out of interest and post the results. If you have any sites with flash or HTML5 stuff, I can do that too if you want to compare results. 

As far as non-standard software. When I did the trails, I closed the two that I have running usually, dropbox and F.lux. During all of this (the last week or so), one the first things I did was kill dropbox incase there was some soft of download that was maybe running continuously or something that I didn't see or notice. In system prefs/users&groups/login items, there is nothing right now at all and hasn't been for a week or so. For the browser tests, only the browser is running.

Only settings changed really are the energy saver settings wart from when I ran the browser tests. 
Battery settings are dim display; I see put hard disks to sleep was chosen, display sleep 2 min, computer sleep 10 min. No screen saver at all.

My machine is a late 2011, MacBook Pro, 2.4 GHz i5, 8GB Ram, Intel HD Graphics 3000 512MB, Samsung 830 SSD 256GB. I didn't do a clean install this time just simply because of where it fell in the semester for me. So Mavericks upgrade over 10.8.5.

Comment 9

[Liviu Cirdei]

Considering a lot of time passed since last activity on this bug, could you confirm if the issue is still reproducing on the latest Firefox versions (e.g. Firefox 43 / Nightly 46)?

Comment 10

[Liviu Cirdei]

Closing this as incomplete due to lack of response from the reporter. Feel free to reopen the bug if the issue still reproduces on a current build.