Closed Bug 1038752 Opened 11 years ago Closed 4 years ago

High energy usage on Huffington Post

Categories

(Core :: Graphics, defect)

x86_64
Windows 8.1
defect
Not set
normal

Tracking

()

RESOLVED INCOMPLETE
Tracking Status
platform-rel --- -

People

(Reporter: joseph.k.olivas, Unassigned)

References

Details

(Keywords: power, Whiteboard: [Power:P1][platform-rel-HuffingtonPost][platform-rel-Intel])

Attachments

(3 files)

Using Intel Platform Power Estimation Tool, and letting Huffington Post sit on the home page, measurements show that Firefox has higher overall energy consumption, stemming from the GPU. It appears that there may have been a regression, as previous measurements (http://people.mozilla.org/~rvitillo/dashboard/#huffington) show that Firefox was reasonable compared to other browsers.
Attached image Huffington - Google.png
Adding Chrome energy measurements
Attached image Huffington - IE.png
Adding IE energy measurements
Blocks: power
I did 30 seconds measurement of the front page on mac using tools/power/rapl and powermetrics. Firefox: - processor+memory: 21.93 Watts - firefox process 10% 32 wakeups/s - plugin-container process 63% 41 wakeups/s (web content) - plugin-container process 6% 32 wakeups/s (plugin content) Safari: - processor+memory: 10.5 Watts - WebContent process 27% 215 wakeups/s - Plugin.64 process 8% 37 wakeups/s Chrome: - processor+memory: 15.4 Watts - Chrome Helper 48% 96 wakeups/s - Chrome Helper 8% 85 wakeups/s - Chrome Helper 8% 36 wakeups/s - Chrome 5% 45 wakeups/s
Status: UNCONFIRMED → NEW
Ever confirmed: true
Keywords: power
Looking at a profiler, it seems like the biggest chunk of time is being spent on painting. Using the paint flashing tool, it looks like the painting is from the scrolling bar in the middle of the page.
Whiteboard: [Power]
Updated Mac measurements taken with |mach power| are as follows. Nightly: > total W = _pkg_ (cores + _gpu_ + other) + _ram_ W > #01 15.95 W = 13.22 ( 2.39 + 2.41 + 8.41) + 2.73 W > > 1 sample taken over a period of 30.000 seconds > > Name ID CPU ms/s User% Deadlines (<2 ms, 2-5 ms) Wakeups (Intr, Pkg idle) GPU ms/s > com.apple.Terminal 724 604.75 234.74 110.39 136.56 > firefox 24827 125.41 68.87 6.83 0.27 83.66 41.48 136.56 > plugin-container 24829 453.45 92.58 28.09 28.45 91.36 49.88 0.00 > plugin-container 24830 27.03 63.93 0.10 0.00 59.21 18.83 0.00 > Terminal 86609 5.92 88.39 0.00 0.00 0.43 0.20 0.00 Chrome: > total W = _pkg_ (cores + _gpu_ + other) + _ram_ W > #01 13.30 W = 10.93 ( 1.70 + 1.48 + 7.75) + 2.37 W > > 1 sample taken over a period of 30.000 seconds > > Name ID CPU ms/s User% Deadlines (<2 ms, 2-5 ms) Wakeups (Intr, Pkg idle) GPU ms/s > com.google.Chrome 790 527.42 376.27 256.69 66.14 > Google Chrome Helper 24894 382.14 89.31 0.00 0.00 131.45 99.00 0.00 > Google Chrome Helper 24890 74.12 66.53 0.00 0.00 140.35 102.50 66.14 > Google Chrome 24884 46.13 65.99 0.60 0.27 69.31 41.92 0.00 > Google Chrome Helper 24895 32.01 65.41 0.00 0.00 35.19 13.30 0.00 Safari: > total W = _pkg_ (cores + _gpu_ + other) + _ram_ W > #01 7.68 W = 5.68 ( 0.39 + 0.56 + 4.73) + 2.01 W > > 1 sample taken over a period of 30.000 seconds > > Name ID CPU ms/s User% Deadlines (<2 ms, 2-5 ms) Wakeups (Intr, Pkg idle) GPU ms/s > com.apple.Safari 791 134.15 50.35 43.65 6.51 > com.apple.WebKit.WebContent 24936 130.25 87.51 0.00 0.00 49.55 43.08 6.51 > Safari 24922 4.31 55.49 0.00 0.00 0.23 0.13 0.00 > com.apple.WebKit.Networking 24925 3.35 66.51 0.00 0.00 0.57 0.43 0.00 > com.apple.WebKit.Plugin.64 793 22.91 102.16 85.43 0.00 > com.apple.WebKit.Plugin.64 24937 23.42 53.63 0.03 0.03 102.16 85.44 0.00
Component: General → Graphics
Product: Firefox → Core
Whiteboard: [Power] → [Power:P1]
platform-rel: --- → ?
Whiteboard: [Power:P1] → [Power:P1][platform-rel-HuffingtonPost][platform-rel-Intel]
platform-rel: ? → ---
platform-rel: --- → ?
platform-rel: ? → -

With a recent Firefox Nightly ii do not see a high CPU load on MacOS. But I don't have Windows to test.

Reporter, if you are still around and using Firefox, is this problem gone for you?

Flags: needinfo?(joseph.k.olivas)

It appears the front page has changed significantly since this was reported, so it is fair to close this as no longer valid. The scrolling headline widget is no longer there, which was the main contributor.

As an aside, we reached out to some of the news agencies (years ago, maybe 2014) as their pages tended to be quite active and culprits for battery drain. Their headlines change so often and quickly that best practices for battery life can be hard to follow, and the narrative will drive the visuals, rather than adhering to consistency/performance/battery life.

Flags: needinfo?(joseph.k.olivas)

Thanks a lot for your feedback Joe! As said I'm going to close this bug as incomplete given that we don't have a way to reproduce it anymore.

Status: NEW → RESOLVED
Closed: 4 years ago
Resolution: --- → INCOMPLETE
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: