261835 - Large transparent images as backgrounds cause major slowdown

Status: RESOLVED INVALID

(Core Graveyard :: GFX: Win32, defect)

Description

[Kelson]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.7.3) Gecko/20040927 Firefox/0.10
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.7.3) Gecko/20040927 Firefox/0.10

When viewing a page that contains a large transparent .gif as a background, the
UI slows down massively.  Anything that involves a redraw - hover effects,
resize, switching back into the window from another one, manipulating form
controls - causes a CPU spike and takes longer than in should.

For example: On an Athlon XP 1800, a 7-item dropdown box can take 3-4 seconds to
drop down.

Reproducible: Always
Steps to Reproduce:
1. Use a Windows version of Mozilla 1.7.3 or Firefox 0.10
2. Go to http://www.postman.net/ for the demo of Open WebMail
3. Log in with the demo login/password
4. Try manipulating form controls
Actual Results:  
Most form controls take at least 2 seconds to respond.

Expected Results:  
Form controls should respond instantly.

In an actual openwebmail mailbox, other effects are slowed down as well.

This appears to be a Windows-only problem, as Firefox manages just fine on my
350MHz Pentium II Linux box.

Disabling JavaScript and CSS has no effect, but removing the background image
solves the problem.  Similarly, editing the image so that it is no longer
transparent also solves it.

Comment 1

[Christian :Biesinger (don't email me, ping me on IRC)]

sounds similar to bug 260676 / bug 255648; are you using an nvidia graphics card?

Comment 2

[Kelson]

> are you using an nvidia graphics card?

I don't think so, but I'll check when I get into work tomorrow.

At home I have a Linux-only system with an NVIDIA card, and a Windows/Linux dual
boot with an SiS 300/305 card.  The NVIDIA/Linux system (32 bpp) is not affected
by this test case or either of the test cases from either of the bugs you
mentioned, but the SiS/Windows ME system is (both 16 bpp and 32bpp). Under Linux
(again, both 16 bpp and 32 bpp), the SiS system has no problems.

On both Windows systems that experience this, neither Opera nor Internet
Explorer is affected.

Comment 3

[Boris Zbarsky [:bzbarsky]]

ccing some Win32 gfx people; this is almost certainly a matter of graphics card
drivers sucking at some particular operation we request of them...

Comment 4

[Kelson]

OK, the system I first noticed this on (Win2K) is running a Hercules 3D Prophet
4000XT, which apparently uses a Kyro chipset. Screen depth is 32 bpp.  It also
trips on the test cases in the other two bugs.

So it looks like all three of these may be the same bug, and it's not
NVIDIA-specific, since I've seen this on two non-NVIDIA systems.

Comment 5

[Ere Maijala (slow)]

Could be related to our somewhat excessive use of StretchBlt. See bug 194971 for
small bits of information. It's unfortunately stalled, but maybe it would be
time to think about it some more. In any case, the issue is not easy and it
seems that every GDI call has issues that affect some systems but not the others. 

I find the comboboxes maybe a little bit sluggish but they still drop down
almost immediately. Anyone see this on Windows XP?

Comment 6

[Joël Kuiper]

I confirm at with Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.8b)
Gecko/20050124 Firefox/1.0+
on P4 1,6Ghz , 512MB DDR ram

Comment 7

[Kevin Ar18]

Attachment: Testcase based on code from the site

Comment 8

[Kevin Ar18]

Notes about testcase:
Bugs Present:
1) Click on the checkbox, and as soon as you release the mouse button move the
mouse.  You will notice lag.
2) Open the page in another tab and then switch to a different tab.  Once the
page is loaded, switch back to the tab with the testcase and as soon as you
release the mouse button, move the mouse.  You will experience lag.
3) Open the testcase page in the browser.  Then open another program and resize
the window of that program so that it only fills part of the screen.  With the
testcase page in the background, move the window of the other program around the
screen.  There is lag while moving the window.
4) Scrolling? not tested.

Note that the problems mentioned above are exactly the same as those described
in Bug 282600 (with the exception of maybe the scrolling issues which I haven't
tested).  However, if you were to switch back to an older build (like Phoenix),
then this bug report still has some of these issues present, while Bug 282600
does not.  I have yet to figure out why this is.  I'm thinking that it may be
due to the image used.

Comment 9

[Jasper Lieber]

(In reply to comment #7)
> Created an attachment (id=174705) [edit]
> Testcase based on code from the site
> 

The attached testcase is an excellent example of Bug 201198.

Comment 10

[Goi]

I can confirm this bug, except I don't have login/pwd for that site, but I've
experience massive slowdowns with transparent GIFs. They don't even have to be
the background. Example pages are http://www.pcstats.com and
http://www.cpemma.co.uk/index.html

Comment 11

[Gervase Markham [:gerv]]

This is an automated message, with ID "auto-resolve01".

This bug has had no comments for a long time. Statistically, we have found that
bug reports that have not been confirmed by a second user after three months are
highly unlikely to be the source of a fix to the code.

While your input is very important to us, our resources are limited and so we
are asking for your help in focussing our efforts. If you can still reproduce
this problem in the latest version of the product (see below for how to obtain a
copy) or, for feature requests, if it's not present in the latest version and
you still believe we should implement it, please visit the URL of this bug
(given at the top of this mail) and add a comment to that effect, giving more
reproduction information if you have it.

If it is not a problem any longer, you need take no action. If this bug is not
changed in any way in the next two weeks, it will be automatically resolved.
Thank you for your help in this matter.

The latest beta releases can be obtained from:
Firefox:     http://www.mozilla.org/projects/firefox/
Thunderbird: http://www.mozilla.org/products/thunderbird/releases/1.5beta1.html
Seamonkey:   http://www.mozilla.org/projects/seamonkey/

Comment 12

[Kelson]

Attached testcase still demonstrates this bug on today's Firefox beta build.

Product: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.8b5)
Gecko/20050927 Firefox/1.4.
Platform: Windows 2000 SP4
Hardware: 3D Prophet 4000XT.

Comment 13

[Wayne Mery (:wsmwk)]

testcase is toast