622492 - Scrolling performance choppiness due to fixing bug 590568

Status: RESOLVED WORKSFORME

(Core :: Graphics, defect)

Description

[[not reading bugmail]]

I notice smooth scrolling performance started lagging more with the patch in bug 590568.  Its worse when a plugin is visible or being scrolled in or out view.

Comment 1

[Timothy Nikkel (:tnikkel)]

Do people besides Dennis see this?

Comment 2

[kolubinowicki]

(In reply to comment #1)
> Do people besides Dennis see this?


Please check the recorded video files.

Recorded by digital camera (making a screencast at high fps eat a lot cpu cycles).
Video recordings are made at 60fps in resolution 320x240 pixels.
Keep in mind watching these video, about few things: moire pattern (distort image), LCD screen ghosting(!), and Youtube video codec recompression artifacts.
But I hope chopiness in scrolling are visible enough (look at region of youtube video control button).


Up to 9 seconds in video Opera 11, after Minefield (Gecko/20110102 Firefox/4.0b9pre):

http://www.youtube.com/watch?v=EzxbB6BLaDw


Minefield recorded 60 fps (smooth scrolling=off, scrolling using scroll bar and mouse wheel):

http://www.youtube.com/watch?v=OqOROvjWVao


Minefield recorded 60 fps, played at 4 fps (slow motion :)

http://www.youtube.com/watch?v=m1cgVgNWhBo

Comment 3

[Timothy Nikkel (:tnikkel)]

Thanks for those videos.

I wonder if these scrolling problems happen everywhere, or only on computers that also had bug 590568.

Comment 4

[[not reading bugmail]]

Agreed, thanks for the videos!  If its in windows/nsWindow code, wouldn't it affect all Windows users?

Comment 5

[Timothy Nikkel (:tnikkel)]

The fix for bug 590568 was in windows/nsWindow, and that bug only affected some computers.

Comment 6

[Joe Drew (not getting mail)]

Please renominate if this happens to more people; one computer isn't enough on which to block, sad to say.

Comment 7

[Csaba Kozák [:WonderCsabo]]

Scrolling is very laggy on this page (not the whole page, but scrolling the thumbnails in the frame) : http://www.youtube.com/user/google#g/u

Can it be related to this bug?

Comment 8

[Timothy Nikkel (:tnikkel)]

There doesn't appear to be any plugins on that page, in which case, no, it must be a different bug.

Comment 9

[Ashley Bischoff (blog at handcoding.com)]

(In reply to comment #7)
> Scrolling is very laggy on this page (not the whole page, but scrolling the
> thumbnails in the frame) : http://www.youtube.com/user/google#g/u

Perhaps that may be bug 625253?

Comment 10

[[not reading bugmail]]

I checked the builds before and after retain layers.  It appears we restored the choppiness (ie the slow and choppy scrolling that occurs when a plugin is visible, if the plugin is not visible, scrolling is faster) that was there before retain layers landed, that retain layers set out to fix.

Comment 11

[[not reading bugmail]]

I realize that between all the bug fixes with OOPP and Retain layers, we might just have to live with it for now, unless retain layers with hw acceleration can be tweaked without regressing all the regression fixes.

Comment 12

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Retained layers improved scrolling speed on many pages, including those without plugins. Also, fixing bug 590568 only affected D3D9.

Comment 13

[[not reading bugmail]]

True.  And i noticed this lag also affects D3D10 hardware acceleration experience with plugins on smooth scrolling.

Comment 14

[Alice0775 White]

Confirmed. Rendering of "video controls bar" is lagged with smooth scroll.
Mozilla/5.0 (Windows NT 6.1; WOW64; rv:2.0b10pre) Gecko/20110122 Firefox/4.0b10pre ID:20110122030336
Works:
http://hg.mozilla.org/mozilla-central/rev/378fe2ca7eb4
Mozilla/5.0 (Windows NT 6.1; WOW64; rv:2.0b9pre) Gecko/20101230 Firefox/4.0b9pre ID:20101231112335
Lagged:
http://hg.mozilla.org/mozilla-central/rev/39db16b78175
Mozilla/5.0 (Windows NT 6.1; WOW64; rv:2.0b9pre) Gecko/20101231 Firefox/4.0b9pre ID:20101231122305

Shockwave Flash
    File: NPSWF32.dll
    Version: 10.1.102.64
    Shockwave Flash 10.1 r102

Graphics
    Adapter Description: ATI Radeon HD 4300/4500 Series
    Vendor ID: 1002
    Device ID: 954f
    Adapter RAM: 512
    Adapter Drivers: aticfx64 aticfx64 aticfx32 aticfx32 atiumd64 atidxx64 atiumdag atidxx32 atiumdva atiumd6a atitmm64
    Driver Version: 8.801.0.0
    Driver Date: 
    Direct2D Enabled: true
    DirectWrite Enabled: true
    WebGL Renderer: TransGaming Inc. -- ANGLE -- OpenGL ES 2.0 (git-devel Jan 22 2011 03:39:02)
    GPU Accelerated Windows: 1/1 Direct3D 10

Comment 15

[Timothy Nikkel (:tnikkel)]

According to Alice's config the problem is affecting d3d10 too? I'm not sure why.

Comment 16

[[not reading bugmail]]

I see it with this too:

ATI Radeon HD 5400 Series
Vendor ID 1002
Device ID 68f9
Adapter RAM 512
Adapter Drivers aticfx32 aticfx32 aticfx32 atiumdag atidxx32 atiumdva
Driver Version 8.762.0.0
Driver Date 8-3-2010
Direct2D Enabled true
DirectWrite Enabled true (6.1.7600.20830)
WebGL RendererTransGaming Inc. -- ANGLE -- OpenGL ES 2.0 (git-devel Jan 18 2011 13:07:16)
GPU Accelerated Windows1/1 Direct3D 10

Comment 17

[[not reading bugmail]]

Using Flash 10.2.161.23, not that it matters.

Comment 18

[Alice0775 White]

Sorry, 
Please ignore the range in Comment 14. 
Because, it happens on Minefield before landing Bug 590568.

Comment 19

[[not reading bugmail]]

Yeah, let me do some more analysis.

Comment 20

[[not reading bugmail]]

This seems to be less choppy or slow when scrolling plugins in and out of view now that bug 627651 landed.  It seems painting feels more evenly distributed regardless of plugins being in view or not.

Comment 21

[Gregory Pappas [:gregp]]

Plugins are no longer with us.