Closed Bug 269767 Opened 21 years ago Closed 17 years ago

Heavy CPU load on page load

Categories

(Firefox :: Shell Integration, defect)

Product:
Firefox
Component:
Shell Integration
Platform:
x86
All
Type:
defect
Priority:
Not set
Severity:
normal

Tracking

()

Status:
RESOLVED WORKSFORME

People

(Reporter: per.angstrom, Unassigned)

References

(
URL
)

Details

(Keywords: perf)

Attachments

(1 obsolete file)

Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.5) Gecko/20041107 Firefox/1.0 When Firefox is loading web pages, it seems to cause quite a heavy CPU load. If a page load takes longer than usual, due to a congested network or whatever, the CPU load continues at the same level, which makes the computer unnecessarily unresponsive. How to reproduce: 1) Start some system monitoring tool that is able to show the CPU load. 2) Point Firefox to a page that is consistently slow to load, such as <http://www.autark.se/slowpage.pl>. 3) While the page is loading, note the CPU load for Firefox. Expected behavior: After a peak when the page starts loading, Firefox's CPU usage should fall back to a lower rate while the browser is waiting for data. Current behavior: The CPU load goes up to 15 - 50% and stays there until the page has finished loading. It seems to make a difference whether the page is loaded in a foreground tab or in the background: the CPU usage goes down considerably when another tab is made active and rises when the loading page is brought back to the foreground. It seems that the problem does not occur with a fresh profile. When using my old profile, it doesn't seem to make a difference if I uninstall the few extensions I normally have installed (Web Developer and User Agent Switcher). The problem seems to be Firefox-specific, as I cannot reproduce the behavior in Mozilla 1.8a4. Maybe it's also specific to Linux.
Keywords: perf
im having very similar problems. a fresh install of either 1.0.1 or 0.9.3 will idle at 70-100% CPU on my machine on a few specific pages. among these are www.ynet.co.il and www.nrg.co.il im running firefox 1.0.1, freshly installed (no plugins, no nothing) on a windows 2000 SP4 machine
WFM with Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.8b2) Gecko/20050602 Firefox/1.0+, 0-1% CPU usage on Athlon64 3500+ at http://www.autark.se/slowpage.pl
*** Bug 276925 has been marked as a duplicate of this bug. ***
Reporter: Do you still see this with Deer Park Alpha 2?
Problem is still present in a nightly Deer Park Alpha 2 build from last night: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8b4) Gecko/20050727 Firefox/1.0+
(In reply to comment #4) > Reporter: Do you still see this with Deer Park Alpha 2? Yes I do.
http://www.ynet.co.il/home/0,7340,L-8,00.html The behavior of the firefox on this site is same from version 1.01 - now using latest Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.7.10) Gecko/20050716 Firefox/1.0.6 - same problem
I'm having the same problem. It first occured last week when I was trying to play WordRacer on Yahoo!, then it happened again moments ago when I was on linux.com. I could scroll the bars, but I couldn't maximize the screen, and my CPU load was at 100%. I checked the task manager, and firefox memmory usage was at 64mb, and peaked out at almost 70 before I killed the program. If I didn't have so much free memmory, it probably would have frozen my pc. After reading this thread, I visited the link posted by Rafi, and I had some interesting results. The page loaded as normal, but then my CPU load shot up from arround 15-20% to the mid 40's. It then went up even higher to the mid 60's and from there keep bouncing back and forth from mid 60's to low 80's back to the low 70's. Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.7.10) Gecko/20050716 Firefox/1.0.6 My system is an Athlon XP 2600+ (Barton Core) with 1gb Dual Channel PC2700 Memmory. I am running Windows XP Professional, with Service Pack 2. I am also on Dial-up (I know, it's pathetic).
WFM - Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.8b4) Gecko/20050901 Firefox/1.0+ This needs to be tested with a current nighly build.
Can anyone run with a current test and let's see if this still happens on Linux & Windows
Fedora Core 4, 2.6.13-mm1 kernel. Using the nightly build from 20050906, connecting via modem (so pages take a while to load). Opened a tab on 'about:', and another to http://slashdot.org . gkrellm is still reporting 2-8% user when looking at the about:, and toggling over to the slashdot tab pops it back up to 50-80%. I popped back and forth several times, and it was consistent. I've seen 100% spikes on occasion, but it's not as consistent as it was. That's still nuts though - 80% CPU on a 1.6Gz Pentium-4 to render slashdot as it comes across a 56K modem. There's not even any audio/video or flash or other CPU pigs on the page....
As a sanity check, loading the same http://slashdot.org as the *only* tab shows CPU utilizations down around 5-8% while loading.
This problem happens to me as well, typically on complex pages with many form elements or embeddings. It also may not happen until Firefox has been in use for several hours. I am using 1.0.6 and have seen this problem from early betas with no improvement.
I'm still seeing the problem in "Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8b4) Gecko/20050907 Firefox/1.4", although not as rampant as before. The example URL, when loading in the foreground, now causes Firefox to grab some 6 % CPU while X usage climbs up to round 15 %. I don't have any other extensions loaded than the default ones. Different from some other reporters, I'm on a fast broadband connection. I think the time has come to tentatively confirm this bug.
Status: UNCONFIRMED → NEW
Ever confirmed: true
OS: Linux → All
I've found a similar bug at page http://carpidiem.it/~diegoliz/page26.do.html see https://bugzilla.mozilla.org/show_bug.cgi?id=246974#c31 It seems that when the page source download takes more than 40 seconds, the browser freezes at 100% of CPU for some minutes, even if all the data has been transferred and there is no more network activity. Is it related to this bug instead?
(In reply to comment #15) No, your problem is that the page is greater than 1 megabite in size. Because the html is so large it takes a long time to get the information to display the page.
Comment on attachment 193906 [details] URL - when enter the site - 95-99% CPU consumption by firefox. When closing the page - back to normal I'm obsoleting this attachment, as it is too complex to isolate the bug.
Attachment #193906 - Attachment is obsolete: true
Seeing this problem with FF 1.5.0.7 on Breezy Badger. My system has 450 MHz Pentium III, 256MB mem, and FF open with some bugzilla tabs, none loading. Followed the link and CPU usage stabilizes on 13%-16%; after pressing Esc to stop the page from continuing to load, usage is 1%-4%. If I let the page continue to load but I switch to another tab, usage is 5%-9%. Accounting for the variation and the rest of the CPU usage, I have a 4% CPU usage increase when the page is loading in the background and 8% in the foreground. Since the page load is nothing a <p>About to sleep for 3 seconds&nbsp;...</p> every few seconds, this is indeed quite alot. Two other points: - I'm not sure it's related to the ynet.co.il issues. Can anyone try to explore this question? - Need to reproduce this with a debug trunk build and do profiling, but I can't devote time to this in the near future.
Here are my observations of Firefox 2.0 rc2: With the test page loading in a foreground tab, Firefox steadily takes 4 - 5 CPU %, while X uses some 10 %. After switching to another tab, Firefox goes to near-zero CPU utilization, while X goes down some 2 - 5 % (which is still too much, IMO). I'm going back to marking this bug as Linux-specific, as I think this is related to the way Firefox uses the X Window System. Furthermore, I'm tentatively assigning this bug to the OS Integration component. Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8.1) Gecko/20061003 Firefox/2.0
Component: General → OS Integration
OS: All → Linux
(In reply to comment #19) At least check it with trunk on Windows before you do that...
(In reply to comment #19) I'm having exactly the same behavior as in the initial description. Running on Win XP Pro SP2, all updates installed. FF 2.0.0.0, just updated to 2.0.0.1 and behavior seems to be similar. Barton 2800+ + 1 GB PC3200 RAM. Normally 2 FF instances open, approx. dozen or more tabs open in both. Using > 300 Megs of RAM according to taskmgr (heavy, uh ?). The interesring part seems to be that when starting FF it seems to be running more or less OK, nut after using some time (an hour or more ?) problem gets worse and worse ending up in a situation where CPU usage is 100% until (almost) the whole page is loaded. Have not tested it with some other tab active, because does not let you to change active tab in that point. Scrolling of the window eats up any free resources possibly left. Scrolling takes CPU resources in general something like 20-70 % with recently started system also... I have installed quite a few extensions, none of them seem to be listed in problematic extensions, though. I wouldn't call this Linux specific...
(In reply to comment #21) I think you may be complaining about generally heavy CPU usage rather than this specific problem. Can you reproduce the usage increase when loading the page linked from comment #0 ?
(In reply to comment #22) You are probably right. I did some testing and could not reproduce the problem with test page of comment #0. However, the problem seems to be related in the way FF operates with network protocol stack. Pages with lots of content, possibly from several servers at the same page seem to cause this. Especially if due (possibly) heavy traffic there are delays in handshakings. And as mentioned before, problem gets worse when FF has been up for a longer period of time. At the moment my FF has been up for approx. 2 days, 21 tabs open in 1 browser and I'm experiencing the problem. BTW, I started IE6 without closing anything else and ran my ISP's speed test; my personal transfer record :-) 8M/1M ADSL, and could get as high as 7,4 Mbps, never achieved that before... TCP & NIC settings and drivers seem to be OK. In brief: - FF using all the CPU resources when loading certain pages. FF halting, shows often "Not responding" - Problem gets worse when FF up and running for a longer period of time. - Seems to be network protocol related issue. If there is a bug discussion which would be better arena for this kind of a problem, pls. give a link.
(In reply to comment #23) When posting the previous comment, FF halted for 1-2 minutes (CPU load 94-97 %)after clicking "commit" button when it was somehow "negotiating" with Bugzilla...
I'm seeing the same behavior as the initial poster, with Firefox 2.0.0.1 running on Windows XP. All my testing was performed in Safe Mode, so no addons are running. I can reproduce this both with the URL mentioned in comment #0, and also with an ASP.NET page that I created that simply sleeps for 1 minute and then returns a very simple page. If I have just one tab, I see Firefox using < 1% of the CPU while it's waiting to load the page. That's good. If I have 2 tabs (one set to the slow page and one empty), and the active tab is the slow page, then CPU use is constantly ~20%, even though the browser is receiving no data off the network. If the other (blank) tab is selected, CPU use is < 1%. I see this behavior as soon as Firefox starts - no need to wait hours or days to reproduce.
Judging by some fairly recent comments, the problem might not be specific to Linux after all, so I'm resetting the OS field to "All". Incidentally, my personal experience of the 3.0 branch (Linux) is that the problem is much less prominent, maybe even insignificant.
OS: Linux → All
I can reproduce with a page returning no data for a long time with an old profile but not with a new profile on the same installation of Firefox 2 / Linux.
Assignee: bross2 → nobody
QA Contact: general → os.integration
Could be bug 270312.
I see the same behavior in FF3.0 and 3.0.1 but the heavy CPU load only happens if the tabbar is visible, so it is more likely related to bug 358336. I never saw this problem with FF1.5.X nor 2.0.X.
Also for the heavy load to occur the FF window MUST be visible. If you iconize FF the page loads normally. Someone please check this. I have Proccess Explorer monitoring the CPU usage and if you minimize/restore FF several times there is a peak load when its visible and a very low/normal load when minimized or obscured by other app.
do you see this problem with current beta of 3.1? http://www.mozilla.com/en-US/firefox/all-beta.html WFM Mozilla/5.0 (Windows; U; Windows NT 6.0; en-US; rv:1.9.1b4pre) Gecko/20090313 Shiretoko/3.1b4pre
I'd say it works for me too, in a fairly recent Linux trunk build. Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.9.2a1pre) Gecko/20090211 Minefield/3.2a1pre
Cannot reproduce the problem in the latest Linux trunk build either. Resolving to WORKSFORME. Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.9.2a1pre) Gecko/20090323 Minefield/3.6a1pre
Status: NEW → RESOLVED
Closed: 17 years ago
Resolution: --- → WORKSFORME
I can confirm this bug. Page loading indicator (which is displayed in the place of favicon) in the tab makes Xorg consume 70%-100% CPU. You must have tabs visible. Try sending a file or opening a page that takes some time to load.
Mozilla/5.0 (X11; Linux i686; rv:6.0) Gecko/20100101 Firefox/6.0 Iceweasel/6.0
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: