Bug 245783 (GLibPerformance) - [GTK] Slow performance issues tracker bug (scrolling, animations, drag & drop, input)
Summary: [GTK] Slow performance issues tracker bug (scrolling, animations, drag & drop...
Status: NEW
Alias: GLibPerformance
Product: WebKit
Classification: Unclassified
Component: WebKitGTK (show other bugs)
Version: WebKit Nightly Build
Hardware: PC Linux
: P2 Normal
Assignee: Nobody
URL:
Keywords: Performance
Depends on: 170347 208249 215045 220085 221738 222902 223970 225512 233126 234700 238578 240865 250988 250998 251016 252249 252846 254985 256105 256454 258926 258931 259227 264966 265048 267774 271233 271234 274218 276845 278692 278695 143868 185579 198857 238415 239934 245124 247748 248224 249145 250978 251017 251336 252545 252948 255713 256123 256460 256756 271232
Blocks:
  Show dependency treegraph
 
Reported: 2022-09-28 07:54 PDT by Jeff Fortin
Modified: 2024-08-26 20:20 PDT (History)
6 users (show)

See Also:


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Jeff Fortin 2022-09-28 07:54:27 PDT
Here's a meta/tracker bug related to all the performance issues I've found to be debilitating for real world usage of Epiphany. I am typically testing with open source Radeon graphics on my desktop computer (AMD Radeon R9 270 GPU, 8-cores Intel Xeon CPU) and on my ThinkPad X220 (Intel Sandybridge CPU and graphics). Issues occur both on X11/Xorg and in a Wayland GNOME session.

Most of the issues are what Mozilla calls "jankiness", i.e. visible lagginess when scrolling in particular, though some may be experienced with animations or drag & drop interactions. Combined, they make the browser unusable for me, for anything except occasional testing out of curiosity. Firefox and Chromium are butter-smooth in comparison.
Comment 1 Michael Catanzaro 2022-09-28 08:24:15 PDT
Collecting these all into one tracker is useful, thanks!
Comment 2 Jeff Fortin 2022-10-01 11:01:38 PDT
While I noted my hardware configuration in the description above, I might add an extra little piece of information: it actually helps to be running older hardware in order to truly exacerbate the problems in a way that makes it easier to debug/test/benchmark. Indeed, if I try on a much newer (and, to my standards, overpowered) Intel chipset, like Intel UHD Graphics 620 on an Intel Kabylake Core i7-8550U CPU, then websites such as the ones mentioned in bug #215045 or bug #221738 (among others) will be a little bit janky and slow to scroll, but not incredibly noticeable on that front, so the problem may go unnoticed except by attentive eyes. The newest hardware chipsets are so ludicrously overpowered that they tend to mask problems.

My recommendation is to benchmark/test your fixes on much older hardware or suboptimal drivers, such as the Intel Sandybridge found in the very popular ThinkPad X220/T420/T520, or some open source radeon graphics from the radeonsi era; any improvements to machines from the 2011-2015 era will also benefit machines running newer hardware, but it will be easier to spot and measure the difference with older hardware, which is still very capable (and works fine on other browser engines).
Comment 3 Peter 2022-10-03 02:52:11 PDT
I want to thank you for the tracker bug! I'm using myself an X220, which seems to be a favorite device for many people after ten years.

What wonders me most is the low performance and high load on Gitlab. Maybe using Epiphany there is a good test case. Just a feeling, we struggle with JavaScript?

Epiphany doesn't allow to turn off JavaScript. It would improve resource consumption, security, and debugging if this would be possible. I remember that some special facilities internal to Epiphany use JavaScript? Good websites run without JavaScript even Amazon and Stackoverflow.
Comment 4 Michael Catanzaro 2022-10-03 07:36:02 PDT
(In reply to Peter from comment #3)
> I want to thank you for the tracker bug! I'm using myself an X220, which
> seems to be a favorite device for many people after ten years.
> 
> What wonders me most is the low performance and high load on Gitlab. Maybe
> using Epiphany there is a good test case. Just a feeling, we struggle with
> JavaScript?

No, we have the best JavaScript. The problem with GitLab is scrolling performance. It won't be fixed here as this is a tracker bug.

> Epiphany doesn't allow to turn off JavaScript. It would improve resource
> consumption, security, and debugging if this would be possible. I remember
> that some special facilities internal to Epiphany use JavaScript? Good
> websites run without JavaScript even Amazon and Stackoverflow.

You can try testing in MiniBrowser if you want to disable JavaScript.