1270841 - Incredible slow canvas on certain site

Status: NEW

(Core :: Graphics: Canvas2D, defect, P3)

Description

[Alice0775 White]

Build Identifier:

Steps To Reproduce:
1. Open URL http://10cloverfieldlane.jp/
   wait until 100% progress bar, then animation using canvas will start.

Actual Results:
The animation using canvas is incredible slower than other browser,
although subsequent trailer video is no problem.

Comment 1

[Alice0775 White]

Build Identifier:
https://hg.mozilla.org/mozilla-central/rev/369a5ee3a2880a4a98df3a00bf3db8d8f36b181b
Mozilla/5.0 (Windows NT 10.0; WOW64; rv:49.0) Gecko/20100101 Firefox/49.0 ID:20160505030327

Graphics
--------

Features
Compositing: Direct3D 11
Asynchronous Pan/Zoom: wheel input enabled; touch input enabled
WebGL Renderer: Google Inc. -- ANGLE (AMD Radeon HD 6450 Direct3D11 vs_5_0 ps_5_0)
Hardware H264 Decoding: Yes; Using D3D11 API
Direct2D: true
DirectWrite: true (10.0.10586.0)
GPU #1
Active: Yes
Description: AMD Radeon HD 6450
Vendor ID: 0x1002
Device ID: 0x6779
Driver Version: 15.301.1901.0
Driver Date: 2-26-2016
Drivers: aticfx64 aticfx64 aticfx64 amdxc64 aticfx32 aticfx32 aticfx32 amdxc32 atiumd64 atidxx64 atidxx64 atiumdag atidxx32 atidxx32 atiumdva atiumd6a atitmm64
Subsys ID: 23111787
RAM: 1024

Diagnostics
ClearType Parameters: Gamma: 2200 Pixel Structure: R ClearType Level: 100 Enhanced Contrast: 300
AzureCanvasAccelerated: 0
AzureCanvasBackend: direct2d 1.1
AzureContentBackend: direct2d 1.1
AzureFallbackCanvasBackend: cairo
ClearType Parameters: Gamma: 2200 Pixel Structure: R ClearType Level: 100 Enhanced Contrast: 300

Comment 2

[Mayank Bansal]

Attachment: memory-report.json.gz

Plus, the memory use increases a lot. I noticed 1GB memory usage for that particular content-process.exe while the demo was running.

Comment 3

[Jeff Muizelaar [:jrmuizel]]

Bas, can you reproduce?

Comment 4

[Alice0775 White]

Gecko profiler:
https://cleopatra.io/#report=35b9da94807389a10f92c9d8efe5205915cef128

Comment 5

[Bas Schouten (:bas.schouten)]

(In reply to Jeff Muizelaar [:jrmuizel] from comment #3)
> Bas, can you reproduce?

Yes, I can, the symptoms, combined with the information from the profiler, lead me to believe that this Canvas demo is using a rectangular clip, we don't make a distinction for canvas between rectangular and complex clips.

Indeed if we look at the code:

            _context.restore();
            _context.beginPath();
            _context.moveTo(0, 0);
            _context.lineTo(0, _col);
            _context.lineTo(_row, _col);
            _context.lineTo(_row, 0);
            _context.closePath();
            _context.clip();
            _context.save();

So this adds too bits of complexity to making the very naive approach work:

1. The rectangle is wound counter clockwise.
2. It's emitted in lines instead of using the rect instruction.

We could add code complexity to detect this being a rectangle. I'm not sure if that's what we want though.

Comment 6

[Jeff Muizelaar [:jrmuizel]]

How does Edge do on this example?

Comment 7

[Bas Schouten (:bas.schouten)]

(In reply to Jeff Muizelaar [:jrmuizel] from comment #6)
> How does Edge do on this example?

It does fine. It's detecting that it's a rectangle from what I can tell.

Comment 8

[Milan Sreckovic [:milan] (needinfo for best results)]

The page is gone - did anybody save that canvas?

Comment 9

[Milan Sreckovic [:milan] (needinfo for best results)]

Attachment: WIP: test for rectangle and call that API

The patch to detect the rectangle with Skia/SkiaGL is fairly easy, but without a test case, it doesn't make sense to land a change that we can't tell is making a difference.

Comment 10

[Jonathan Watt [:jwatt]]

Mass bug change to replace various 'parity' whiteboard flags with the new canonical keywords. (See bug 1443764 comment 13.)

Comment 11

[BugBot [:suhaib / :marco/ :calixte]]

Clear a needinfo that is pending on an inactive user.

Inactive users most likely will not respond; if the missing information is essential and cannot be collected another way, the bug maybe should be closed as INCOMPLETE.

For more information, please visit BugBot documentation.