Closed Bug 936338 Opened 11 years ago Closed 2 years ago

Chrome is more responsive on the D3.js based MIT Media Lab's Immersion application

Categories

(Core :: Graphics: Layers, defect, P2)

defect

Tracking

()

RESOLVED INCOMPLETE
Performance Impact low
Tracking Status
firefox56 --- affected

People

(Reporter: birtles, Unassigned)

References

(Blocks 1 open bug, )

Details

(Keywords: parity-chrome, perf, perf:responsiveness, Whiteboard: [in-the-wild])

At MozFest 2013, Deepak Jagdish introduced me to MIT Media Lab's Immersion application which uses SVG to visualize social connections. They are currently using Chrome to demo the application since the animation of SVG shapes is too slow on Firefox. If you load the URL in Chrome the animation is very smooth and responsive. In Firefox it is very jittery and uses a lot of CPU.
FWIW, SPS profile against Mozilla/5.0 (Windows NT 6.1; WOW64; rv:31.0) Gecko/20100101 Firefox/31.0 ID:20140321030203 CSet: 72c2f9ebf4bc: http://people.mozilla.org/~bgirard/cleopatra/#report=197f612102ebf2fe8ba4abacb04d4b71ec600e7a (played with the "Charge", "Nodes" and "Links" sliders) This is - again - slow D3.js perf.
I put the Nodes slider to maximum and jiggle the Links slider around I can still see Chrome is a bit smoother. That said, there's a big improvement since we started the Moz2D work.
Bug 1090934 should also help a bit more.
Depends on: 1090934
I'm going to close this, but it would be good if you could check you're happy with that, Brian.
Status: NEW → RESOLVED
Closed: 8 years ago
Flags: needinfo?(bbirtles)
Resolution: --- → WORKSFORME
Summary: Immersion animation of SVG is slow → Chrome is more responsive on the D3.js based MIT Media Lab's Immersion application
Chrome still seems significantly smoother to me. When you click a circle we drop a lot of frames and only the second half of the transition animates. When I drag the slider I get about half the frame rate of Chrome.
Flags: needinfo?(bbirtles)
OK.
Status: RESOLVED → REOPENED
Resolution: WORKSFORME → ---
Whiteboard: [in-the-wild] [parity-chrome] → [in-the-wild] [parity-chrome][qf:investigate]
Status: REOPENED → NEW
OS: Windows 7 → All
Priority: -- → P2
Hardware: x86_64 → All
Flags: needinfo?(cku)
Depends on: 1385159
Depends on: 1385164
Flags: needinfo?(cku)
Assignee: nobody → cku
Turing on webrender make perfromance worse...
(In reply to C.J. Ku[:cjku](UTC+8) from comment #7) > Turing on webrender make perfromance worse... A bug worth checking further.
Flags: needinfo?(howareyou322)
Flags: needinfo?(howareyou322)
Painting - 8 ~ 9 ms Scripts - 4.8 ms in average. DisplayList - 1ms in average LayerBuilding - 2~3 ms Reflows - 0.3 ms Styles - 0.4 ms FF FPS is around 55, and Chrome can render frames in 60FPS constantly.
As per comment 5 I see a lot of dropped frames on Windows when you click the circles to expand/collapse them but I don't see (m)any dropped frames in Chrome, and even Edge seems better.
Depends on: 1385861
Assignee: cku → nobody
Mass bug change to replace various 'parity' whiteboard flags with the new canonical keywords. (See bug 1443764 comment 13.)
Keywords: parity-chrome
Whiteboard: [in-the-wild] [parity-chrome][qf:investigate] → [in-the-wild] [qf:investigate]
The delay when clicking a circle, at least on Mac, is dominated by tile allocation. It seems like we're creating too many layers and not reusing tiles effectively. I've filed bug 1514047 about the WebRender part of this.
Component: SVG → Graphics: Layers
Whiteboard: [in-the-wild] [qf:investigate] → [in-the-wild] [qf]
Whiteboard: [in-the-wild] [qf] → [in-the-wild] [qf:p3:responsiveness]
Performance Impact: --- → P3
Whiteboard: [in-the-wild] [qf:p3:responsiveness] → [in-the-wild]
Severity: normal → S3

Link is gone so I can't test but in my experience in a M1 mac Firefox now is on par with Chrome rendering SVG and Canvas graphics.

I can't find a working archive of this anywhere.

Status: NEW → RESOLVED
Closed: 8 years ago2 years ago
Resolution: --- → INCOMPLETE
You need to log in before you can comment on or make changes to this bug.