Closed Bug 288472 Opened 20 years ago Closed 18 years ago

High CPU load after a few hours

Categories

(Firefox :: General, defect)

PowerPC
macOS
defect
Not set
critical

Tracking

()

RESOLVED DUPLICATE of bug 273310

People

(Reporter: martin.otten, Assigned: bugzilla)

Details

Attachments

(4 files)

User-Agent: Mozilla/5.0 (Macintosh; U; PPC Mac OS X Mach-O; en-US; rv:1.7.6) Gecko/20050326 Firefox/1.0.2 (PowerBook) Build Identifier: Mozilla/5.0 (Macintosh; U; PPC Mac OS X Mach-O; en-US; rv:1.7.6) Gecko/20050326 Firefox/1.0.2 (PowerBook) After a few hours sometimes firefox uses up to 80% of the CPU-power (even if firfox should have nothing to do). As I do not restart my powerbook for weeks I know that it seems to happen every time running firefox and using it for a few hours. Closing windows and tabs does not help Reproducible: Always Steps to Reproduce: 1. Start Firefox on MacOSX 10.3.8 2. Open some windows and tabs 3. wait some hours (maybe more than 24) 4. open the terminal 5. execute "top" Actual Results: top shows me up to 70% of cpu load Expected Results: cpu useage near 0%
Please attach output of sample when Firefox is using cpu
Before I mage this screenshot I ran some java-applets. The tabs containing this applets are closed now. currently firefox has nothing else to do than displaying 12 pages opened in tabs in one window. The pages only contain animated gifs. But currently none of them are visible while I am writing this comment.
Related to or duplicate of bug 254884 ?
I am using FireFox on a Win XP Pro machine & getting similar results. I tend to keep many tabs open in many windows. I also tend to shut down with "hibernate" to save my state, including where I've been browsing. At first, CPU usage seems normal, but after a day or two, FireFox is pegging 97-99% of CPU. I shut down tabs one at a time trying to find a "guilty page", but it doesn't seem to matter what the content is: static html or anything else. After each closure there's a slight reduction for a a second or two to 70-80%, then it springs right back up to 99. Finally I get it down to a single tab with static content and the Task Manager still shows FireFox consuming 98-99% CPU. This impacts my usage primarily when starting new processes. When they start taking forever to start up, I check & see FireFox is busy doing nothing with 99% of the CPU. Interestingly, other interactive stuff usually seems to be unaffected (perhaps because other interactive stuff is so lightweight it can get by with the remaining CPU power?).
I've been using Firefox since "back in the day". Ever since 0.9 or so, I'm been seeing CPU hogging (in Windows 2000). At first, Firefox just used to be slow to open a window that had been minimized - almost like Firefox had gone to sleep and was having trouble "waking" up. Now, I get this a LOT, even while typing this. Firefox uses almost 100% of the CPU and appears to just lock up. It usually recovers in 10-30 seconds, but I find myself opnening up IE (gasp!) while it's hung just so I can keep working. If you search for "CPU" in the open bugs list (https://bugzilla.mozilla.org/buglist.cgi?query_format=specific&order=relevance+desc&bug_status=__open__&product=Firefox&content=cpu) you will see a LOT of bugs that appear similar in nature. Sounds like something being seen by enough users under enough different situations to warrant a serious investigation!
I checked if camino has got the same bug, and well it has. Take a look at the attatchment.
* Firefox 1.0.4 on XP pro SP1, ie. not Mac only, also earlier versions and Mozilla After several hours of use with many tabs, Firefox starts to consume all available CPU time. At the same time, Windows Task manager shows "I/O Other" to increase by some 100000 requests per second and "I/O Other Bytes" by some 1 MB per second. (see talkback TB6717605X, Firefox actually crashed then but it is rare) Prior to the steady 100% load there may be brief periods of 100 % load when starting a download, sometimes throughout an entire download. Sometimes the condition is triggered upon resume from hibernation, and then it can almost freeze the entire system for several minutes (any operation in windows taking several seconds). In one form or another, similar CPU hogging has occurred in most Mozilla and Firefox versions. Even Mozilla email client alone may start it.
Firefox is getting slower on every little release since several months. No idea, if this is due to any changes in OS X, but on my machine, Firefox is almost unusable now (well, I started with Mosaic on a SE and it was faster..). Opening pages, clicking links, typing here on bugzilla - everything is very slooow. I have so many beachballs here - I could open a shop :/ I tried to alter the setting in many ways, removed/added fonts, tried everything I could but no luck. Safari is about 500% faster or even more, esp. after some hours. I rarely restart my machine, but only Firefox shows this behaviour. After these few lines of text, Firefox draws 1 char/s and I have to wait several seconds just to see what I wrote. This happens on all websites. Typing here highers the cpu-time up to 84% (1GHz G4, 1,5GB RAM) with only one window open containing 6 tabs from bugzilla. Copy from top: PID COMMAND %CPU TIME #TH #PRTS #MREGS RPRVT RSHRD RSIZE VSIZE 23584 firefox-bi 59.5% 13:16:52 3 100 1370 213M 60.9M- 128M 611M All version I tried show this slowness, DeerPark is even worse. Sad to say, but I have to say goodbye to Firefox and start to use Safari :( Even the old IE is *way* faster for me. I read several comments like mine on several pages, so I am not alone. *sigh* This is really a critical bug(?), because it renders Firefox unusable on many machines I know of/use.
I've been seeing this too, on OS X 10.4.2. This sample output was taken right after I had clicked on a link in bloglines. If Firefox has been started recently, this link will load very quickly (a few seconds at most) and the CPU is usually around 12% for firefox-bin. When this sample was taken, it took almost two minutes for the page to load. CPU load for firefox-bin was between 32% and 60%.
I'm using Firefox 1.5 beta 2 on Mac OS 10.4.2, on an iBook G4 (1 GHz, 768 MB RAM). Although I never noticed any particular high CPU usage (apart from pages with Flash content, but that's why I have the Flashblock extension installed), I can tell you that, after long browsing sessions, Firefox surely starts becoming slower: scrolling, page loading, tab handling... everything becomes slower and slower as the sessions continues; when opening the preferences window I can even see the buttons in the window fade in. Quitting Firefox and launching it again fixes the problem, so it does not seem to be related to particular web pages, or to the amount of history items. Also 1.5 beta 1 had the same problem, and I REALLY hope it can be fixed before the release, as it is a real showstopper for me. I sometimes have to launch Safari to decently browse a web site when Firefox has slowed down and I have other open tabs that I do not want to lose (or I sometimes save all of them as a bookmark folder and quit the browser). As far as I can tell, versions 1.0.x didn't have this problem.
As Cammino got the same problem I am sure that this is a problem of the gecko-engine. I hope that it will be fixed in the new vectorgraphics version.
Mozilla/5.0 (Macintosh; U; PPC Mac OS X Mach-O; en-US; rv:1.9a1) Gecko/20051118 Firefox/1.6a1 I have gotten the same problem running Firefox 1.5 through Deer Park 1.6a1 on Mac OS 10.3.8, 600 Mhz iBook. 1. STATIC PAGES After 1-2 day's work on the machine (no restarts), Firefox uses 15-60% of CPU cycles, sometimes more -- even if Firefox is hidden, and even if all open pages are plain HTML with no moving images. Closing tabs or windows will not free up the CPU cycles; I must restart the program. 2. ANIMATED PAGES If Firefox is displaying pages with animated GIFs or other moving images, it can chew up ALL CPU cycles left available by other programs, even if: (a) the animated images are in hidden tabs (that is, the top tab on all animated windows is a static page), or (b) Firefox is hidden, and shouldn't be processing anything. In this case, closing the tabs with animated images will free up some CPU cycles, BUT the program will continue to use too many CPU cycles a la #1, depending on how long I left it running. CPU cycle usage does not increase during time the machine is in Sleep mode. Reproducible: Always Steps to Reproduce #1: 1. Start Firefox/Deer Park on Mac OS 10.3.8 2. Open at least one window, with several tabs (more windows and more tabs seems to make the problem worse) 3. Leave Firefox/Deer Park open for some hours -- doesn't matter if you Hide it, stick the windows in the Dock, or leave them visible, and doesn't matter if you use or don't use the program 4. Check CPU usage using Activity Monitor (I usually do this when I notice a program -- not necessarily Firefox -- responding slowly) 5. CPU usage will be at 12% even if the program is hidden, more usually 15-60%, and is more (a) the longer the program runs and (b) the more windows and tabs are open while it runs 6. Closing tabs and windows will not free up the CPU cycles; restart is necessary. Steps to reproduce #2: Same as above, but open some windows with animations. The effect is greater with more animation windows open, even if all are hidden. Closing tabs and windows will free up some CPU cycles, but not those consumed by Problem #1. Expected results: CPU usage in the 1-2% range when the program is inactive and hidden. A slew of other bug reports seem to point to the same problem: 312048 (Mac) 311242 (Windows) 293689 (Windows) 313751 (Solaris)
I think this has been fixed in Firefox 1.5 final, I did not experience this problem anymore, although Firefox still "feels somewhat heavy" on my iBook G4 1GHz, but there's not much that can be done about it at the moment, I'm afraid.
Several times when opening the bookmarks list and no other programs running, Direfox takes 100% control of the CPU and becomes unresponsive. The only way to close it is by going to the task bar and issuing a close comand.
The same problem occurs in Camino. Mac OS X 10.3.9 Camino trunk NB 2006041619 (v1.2+)
Attached file Sample of Camino
When the problem occurred, the sample was taken from the activity monitor.
*** This bug has been marked as a duplicate of 273310 ***
Status: UNCONFIRMED → RESOLVED
Closed: 18 years ago
Resolution: --- → DUPLICATE
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: