Closed Bug 295848 Opened 20 years ago Closed 20 years ago

Huge performance hit when rendering pages with transparent GIF images

Categories

(Firefox :: General, defect)

Product:
Firefox
Component:
General
Platform:
x86
Windows XP
Type:
defect
Priority:
Not set
Severity:
major

Tracking

()

Status:
RESOLVED INVALID

People

(Reporter: levicki, Unassigned)

References

(
URL
)

Details

Attachments

(3 files)

Firefox performance hit on Pentium 630 at 3.0GHz with 1GB RAM
72.97 KB, image/gif
Details
In case anyone doubts...
33.25 KB, image/gif
Details
Example page without the java applet and images
13.86 KB, text/html
Details 
User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.7.8) Gecko/20050511 Firefox/1.0.4
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.7.8) Gecko/20050511 Firefox/1.0.4

DISCLAIMER: I cannot bother to search for the right bug # because this has
already been reported at least 100 times with different wording so if you think
it is duplicate of a duplicate of a duplicate then delete the other ones and
confirm this one because others do not bear any useful info as this one. Thank you.

It happens with other pages too but here is the most visible --
http://www.homepages.hetnet.nl/~brianpostma/pc.html -- my 3.0 GHz Pentium 4 with
1GB RAM becomes 90MHz Pentium 1 as I scroll over the page. I have narrowed the
issue down to the transparent GIF background image. Seems that the rendering of
GIF images is broken in Firefox. This issue also affects AMD CPUs. I am not sure
if it has anything to do with video card make and model (I have nVidia 6600 GT)
and video drivers (Forceware 71.89).


Reproducible: Always

Steps to Reproduce:
1. Just open the page and try to scroll up/down.

Actual Results:  
I got CPU usage >50% and the Firefox scrolled the page awfully slow.

Expected Results:  
It should scroll fast and smooth without taxing the CPU that much.
Does anyone out there consider this much CPU usage as a normal behaviour for a
lightweight browser?
This image should confirm that the high CPU usage comes from Firefox.
I also notice this. Are you sure it has nothign to do with the Java applet on
this page? I will research this more this afternoon.
Related to bug 265610?
Here, I localized the page and removed the Java applet. By doing so, I noticed
CPU consumption decreased by half to the level of CPU usage that is usual for a
page with this much content on it (many images, tables, etc.).

Therefore, I can say that due to the Java applet being on the page, the source
of higher CPU usage comes from there. In fact, if you visit the page in
Internet Explorer, you will notice much more processor lag than when you use
Firefox.

Marking this bug INVALID (not a bug).

Reporter, if you are willing to provide the URL for the search you conducted
that contained hundreds of similar reports, I'd like to see it. Thank you.
Status: UNCONFIRMED → RESOLVED
Closed: 20 years ago
Resolution: --- → INVALID
> DISCLAIMER: I cannot bother to search for the right bug # because this has
> already been reported at least 100 times with different wording so if you think
> it is duplicate of a duplicate of a duplicate then delete the other ones and
> confirm this one because others do not bear any useful info as this one. Thank
you.


If that is the case then you should not have filled a new bug. If the existing
bug is missing information then add it to that one. Please do NOT file duplicate
bugs when you are well aware they are duplicate bugs.

Also please use trunk nightly builds for filling bugs and not out of date
mile-stone releases.
# Here, I localized the page and removed the Java applet.
# Therefore, I can say that due to the Java applet being
# on the page, the source of higher CPU usage comes from there.

I do not care about Java applet, I didn't have Java enabled, not to mention that
I don't have it installed at all.

# In fact, if you visit the page in Internet Explorer,
# you will notice much more processor lag than when you
# use Firefox.

Isn't it obvious that I use Firefox because I do not want to use IE? 

# Marking this bug INVALID (not a bug).

Well I am sorry but I won't allow this to happen because this is THE BUG.

# Reporter, if you are willing to provide the URL
# for the search you conducted that contained hundreds
# of similar reports, I'd like to see it. Thank you.

Damn right I am willing:

http://instruct1.cit.cornell.edu/courses/ee476/FinalProjects/s2004/fci2/highleveldesign.html
http://www.pcper.com/article.php?aid=141

If you need more let me know. Another way to test it is to open that page I gave
as the first example and save it. Then open just BRIAN'S
HOMEPAGE_files\background.GIF using Firefox and try scrolling around. You will
see that it lags and taxes CPU considerably. That doesn't happen with JPG or
PNG, just GIF and mostly transparent ones used for backgrounds. I would ask you
to please check this again and don't be in such a hurry to ditch a valid bug
report next time.

# If that is the case then you should not have filled a new bug.
# If the existing bug is missing information then add it to that
# one. Please do NOT file duplicate bugs when you are well aware
# they are duplicate bugs.

Problem is that there are simply too many of them with different wording. Which
one should I add to then? Also, most of them are just one liners so if I am
already adding everything I would rather file new one.

# Also please use trunk nightly builds for filling bugs
# and not out of date mile-stone releases.

I am user not a developer so I am using fresh install of 1.0.4 final. I am
really not interested in nightly builds. If this bug is fixed in some nightly
build after 1.0.4 then please mark this bug as FIXED and let me know in which
release it will be included so I don't have to waste time downloading and
installing/uninstalling nightly builds. Thank you.
Status: RESOLVED → UNCONFIRMED
Resolution: INVALID → ---
this is about multiple bugs.
can't dupe to more than on ->INVA
Status: UNCONFIRMED → RESOLVED
Closed: 20 years ago → 20 years ago
Resolution: --- → INVALID
(In reply to comment #8)
> this is about multiple bugs.
> can't dupe to more than on ->INVA

Can you please explain this??!

More "slow" links:

http://www.kdge.com/index.html
http://fahmon.silent-blade.org/

This is interesting:
http://www.world-direct.at/mozilla/dhtml/75121/anim-test.htm

For me it gives:

IE 6.0 = 1562ms
FF 1.0.4 = 9062ms

This bug seems to be most related to the one I filed:
https://bugzilla.mozilla.org/show_bug.cgi?id=242426

It would be nice if someone investigated this with more attention.
Status: RESOLVED → UNCONFIRMED
Resolution: INVALID → ---
(In reply to comment #10)
> More "slow" links:
> 
> http://www.kdge.com/index.html
> http://fahmon.silent-blade.org/
> 
> This is interesting:
> http://www.world-direct.at/mozilla/dhtml/75121/anim-test.htm
> 
> For me it gives:
> 
> IE 6.0 = 1562ms
> FF 1.0.4 = 9062ms
> 
> This bug seems to be most related to the one I filed:
> https://bugzilla.mozilla.org/show_bug.cgi?id=242426
> 
> It would be nice if someone investigated this with more attention.
> 

they are different bugs that all have a separate duplicate.
Status: UNCONFIRMED → RESOLVED
Closed: 20 years ago → 20 years ago
Resolution: --- → INVALID
Status: RESOLVED → VERIFIED
> 
> they are different bugs that all have a separate duplicate.
> 

Well excuse me, but all I see is a slow scrolling and big perfornace hit. Can I
expect this issue fixed or should I give up using browsers based on Gecko engine
and leave you developers arguing whether it is a bug or feature after all those
reports you have seen?

If this bug is invalid, then please point me to the valid bug(s) for which I
should vote, thank you. This is the last time I am reopening this just so you
get notification, I won't do it again.
Status: VERIFIED → UNCONFIRMED
Resolution: INVALID → ---
(In reply to comment #13)
> If this bug is invalid, then please point me to the valid bug(s) for which I
> should vote, thank you. This is the last time I am reopening this just so you
> get notification, I won't do it again.

Igor, bugzilla is a system for tracking bugs to help improve firefox and for
work on its development it is not for assisting users. If you wish to find the
relevant bugs then please search for them yourself or ask elsewhere.
(forums.mozillazine.org)
Status: UNCONFIRMED → RESOLVED
Closed: 20 years ago → 20 years ago
Resolution: --- → INVALID
> Igor, bugzilla is a system for tracking bugs to help improve firefox and for
> work on its development it is not for assisting users. If you wish to find the
> relevant bugs then please search for them yourself or ask elsewhere.
> (forums.mozillazine.org)

I have searched, that is the point. All I found was a big nothing. I found a
bunch of similar yet as someone here claims different bug reports, you seem to
be very well aware of the issue and still those reports are either UNCONFIRMED
or DUPLICATE or NEW or FIXED and the problem is STILL THERE. I am very well
aware that this place is not meant for end user support but I am not asking your
assistance in using the browser, I am reporting an issue that stops me from
using it as intended.

I am asking you what in your opinnion should I do now? What benefits will I have
from searching the bug database and the forum? Will that fix the issue that I am
experiencing?
Why don't you check the forums, maybe you will then understand what I am talking
about here:

Here, check this:
http://forums.mozillazine.org/viewtopic.php?t=258573&highlight=cpu+usage
Status: RESOLVED → UNCONFIRMED
Resolution: INVALID → ---
If you're not interested in participating in the process, please feel free to
take your comments elsewhere.

Filing a generic bug that's a dupe of "hundreds of other bugs" is rude and
insulting to people trying to get things done here.  Yes, there are bugs that
cause performance problems in certain cases.  No, this can't be fixed magically
as part of a single bug.  There's a lot of different root causes caused by
specific combinations of factors, and fixing those without regressions is a
time-consuming process.  If you want to help, great, find those bugs and write
minimal testcases.  Don't be rude here, and waste people's time with useless
bugs (stuff is slow, make it faster seems to be the gist of the bug).
Status: UNCONFIRMED → RESOLVED
Closed: 20 years ago → 20 years ago
Resolution: --- → INVALID
Update:
I have tried to decrease hardware acceleration by one notch and the results for:
http://www.world-direct.at/mozilla/dhtml/75121/anim-test.htm

are as follows:

FULL acceleration    : 8984ms
REDUCED acceleration : 1453ms (comparable to IE rendering speed with FULL
acceleration)

So this seems to be the issue with Firefox and nVidia video cards/drivers only,
that is why it is hard to reproduce.
According to the VTune 97% of the time is spent in win32k.sys EngStretchBltROP()
function when rendering this animation:
http://www.world-direct.at/mozilla/dhtml/75121/anim-test.htm

And also according to VTune 99% of the time is spent in win32k.sys in
EngStretchBltROP() function when rendering transparent GIF image like this one:

http://www.pcstats.com/i/pcs_IEbkgrnd.gif
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: