Post a Comment
I agree. I've actually had to switch FROM Firefox because it leaks such memory. In fact, after a long period, it can eat up to 99% of my CPU too. I've recently downloaded Firefox 3 beta 4 to play, and I must say, between the integrated features and extensions like Stylish, it's making a pretty convincing case.
I've been using the nightly builds for about three months now, and I've only had it crash on me once...so you may want to give it a second chance.
From my experience, it definitely feels faster, the improvements in rendering are nice, and the new location bar makes it superior to any other browser. I love being able to type part of the title of a page (when that isn't even part of the URI) and have it pull up matching pages. With sites like OS News that have /story/#-style URIs, it's invaluable.
They need to add a grammar-checker for Thom, though, to catch that "it's" (a contraction of "it is" -- before "predecessor") should be "its" (possessive form of it).
One thing to note, though: I had to edit the install.rdf file for AdBlock (by default, it isn't set to work with 3.0.*, but you can tell it that it's compatible by editing this file). Long use of Firefox + AdBlock has made it my one cannot-live-without extension.
Should you want to hack AdBlock (or any other extension that will work, but isn't developer-approved for Firefox 3), you can find the install.rdf here:
firefox_profile_dir/extensions/{########-####-####-####-############}/ install.rdf
And you're looking for a block of XML like:
<em:targetApplication>
    <Description>
        <em:id>{########-####-####-####-############}< ;/em:id>
        <em:minVersion>1.5</em:minVersion>
        <em:maxVersion>3.0.*</em:maxVersion>
    </Description>
</em:targetApplication>
You want to make the maxVersion look like my example.
Note: {########-####-####-####-############} indicates that each # is a random hex value. If you're not sure where to find your Firefox profile, see:
http://support.mozilla.com/kb/Profiles
EDIT: alternate HTML characters worked in the preview, but not in the post...and my em-dash is apparently encoded incorrectly, despite Firefox using UTF-8 and the page using UTF-8... o.O;
Edited 2008-03-14 04:45 UTC
From my experience, it definitely feels faster, the improvements in rendering are nice, and the new location bar makes it superior to any other browser. I love being able to type part of the title of a page (when that isn't even part of the URI) and have it pull up matching pages. With sites like OS News that have /story/#-style URIs, it's invaluable.
...
I've had a similar experience both on Mac OS X and Windows. I'm a bit annoyed with the new "keyhole" look and functionality of the backward and forward buttons and list, but they work. The Mac OS X version retains the list on the buttons, which is extremely convenient but definitely not quite right. Speed is excellent and rendering is better than I had expected, though the improvements are of a lesser degree on Windows than on Mac OS X.
I tried beta 4 (not 3.0b5pre) the other day as my Firefox 2.0 replacement and I've got to say that it was horribly buggy for me, in contrast to the nightly builds. After a short time, it failed to respond to the keyboard or mouse and reminded me of a very early Firefox 1.x. It went away very quickly. Hopefully, beta 5 will be more usable for me.
Been using Swiftfox (http://getswiftfox.com/) on my Ubuntu for weeks ("Swiftfox is an optimized build of Mozilla Firefox" for Linux, based on FF 3.0 for a while now), very usable and much better looking on Gnome than 2.0
I've just tried Beta 4 in Leopard and and it's alot smoother, smooth scrolling is really smooth. It's much better than 2.x and the theme fits nice.
It's a straight fight between this and Camino for me at the moment, Firefox is still slower on startup and lacks a default ad-block like Camino but great work with beta4.
Firefox philosophy is plugin based, so i don't expect (and want?) adblock and other 'advanced' stuff (not so many people use ad blocking) to be builtin.
Camino does not support plugins, so having adblocking out of the box is mandatory.
Quality of this beta4 is so great that i don't think Camino will survive a long time. I was expecting (announced) improvements for the mac port, but firefox 2 was so bad that i am really impressed with this beta.
Well, Firefox' integration with Mac OS X is still not as tight as Camino's. For instance, Firefox 3 still does not support the system keychain or systemwide proxy settings (2 things I use on a daily basis). I'm especially disappointed in the former since it has been pointed out to me that the infrastructure for keychain support was implemented over half a year ago and it was allegedly "little work" to actually make it work but obviously it didn't land in time for the beta cycle or was regarded as a low priority target. On Windows such details might seem superfluous but on desktops with better frameworks (including GNOME/KDE) this makes Firefox stand out a bit.
But I have to admit that I fell in love with the new Mac theme. Although it shares some similarities with the Windows theme the grey colours suit me far more than the coloured glass buttons on the Windows platform.
Smooth scrolling is too slow on my old Intel Macbook to be usable (compared to Safari at least).
Auto-completion in the address bar is more informative and clearer now but I have experienced some problems with it becoming jerky if you start to delete an address character by character (via backspace) as Firefox will try to find a match in the history each time.
Anyway, I feel rather positive about the new Firefox. Oh, and to add at least *something* which is on-topic: I'm glad that Mozilla has been working on the memory consumption. I'm curious how it will compare with Opera 9.5.
But you can't trust anyone's statistics on memory usage. Everyone has an angle (and the potential for bias here is pretty clear).
Then, of course, Firefox's memory usage has become such an internet cliche that I doubt these sorts of reports will be enough to kill it. I hope they do though. This is one of the dryest flamewars I've ever seen.
"We’ve taken a look at how much they cache and how long they cache it for. In many cases we’ve added expiration policies to our caches which give performance benefits in the most important cases, but don’t eat up memory forever."
I'm glad that they have finally revisited their policy of caching everything and keeping it forever. It was a design decision Mozilla made that many of us have found quite disagreeable.
Is not the leaks in Firefox itself, but rather the HUGE leaks it triggers in the X server. I've tested 2.x on the last several releases of X and the behavior is always the same. Sites with lots of images are prone to do this more quickly. And then only way to undo the leak is to restart X which is, well, inconvenient in many cases. I hope they have fixed this issue.
I'm facing the same problem on my Linux (fedora 8) machine. Firefox b3/4 feels often very slow and not responsive. And this is not due high loads of FF itselves.
I already expected that it's the X server.
You have any bug reports from it or people who are complaining too?
No, it's a leak. I've used xrestop to compare what X thinks it has allocated and what cat /proc/`pgrep X`/smaps says it has in its heap. The differences are often astounding. I've had the X server have over 700MB allocated in its heap and xrestop claims that it only has, say, 20MB of resources allocated. It's ridiculous.
You may want to check out the GrApple themes, then:
http://takebacktheweb.org/
I've been using the Yummy variant (and its Gran Paradiso precursor) for a few months now. With my settings, Firefox doesn't use any more screen real estate than Safari does.
You can also right-click the navigation bar, pick "Customize", and check "Use small icons."
...using Mac OS X 10.4.x with UNO 1.52 and the Firefox UNO theme from http://www.takebacktheweb.org add the following extensions:
Fission
Safari View
Stop or Reload Button
SwiftTab
...and optionally the Bookmark Keys extension (if you can make it work exactly as you'd expect on a Mac (command + 1-9 selects between Bookmarks Toolbar links), and you'll be able to get an almost exact Safari clone
Edited 2008-03-14 08:30 UTC
There should be more stuff like this and less stuff like the Mono/Miguel/Microsoft blog that was linked to recently. Even though I don't get excited about browsers, I'm interested in reading about how problems like Firefox's memory usage are dealt with in a concise article (I'm pretty sure all this information was already "out there" in the form of discussions, mailing lists and whatnot).
As someone noted in a comment on the original site, a nice side effect of the Firefox memory issue is that, in tackling it, some useful/interesting tools and tests have emerged.
I like the speed and I even like the new location bar. I haven't noticed it getting between me and the pages I want at all. Usually three keypresses are more than enough to bring up the page I want.
The only thing I don't like about Firefox3 is its XP theme. Hoping to get winestripe for FF3 soon. But I used to do the same with FF2, so nothing changed.
Have been using FF3B4 for 2 days solid now, switched from FF2 as soon as i was accepted to the Foxmarks beta program, was a show stopper without this ... Got to be said its fast and so far problem free for me ... Its ben open on my Mac for 2 days now, many tabs open and closed, all manner of sites visited and the memory footprint is still only 168meg (i have a few tabs open just now).. not too shabby ... Well worth a shot if you ask me ... Come to think of it he performance is pretty good as well, feels a bit slicker than FF2, that may be subjective though as im in a good mood with FF3 just now :-) ...
Edit :- spelling
Edited 2008-03-14 03:09 UTC
Been using it heavily for 3 days now. Completely replaced my FX2. Gotta say it feels pretty solid, renders 100% correctly (the pages which I visit), crashed only once for me when opening a page from the history list.
There are still issues, like uncomfortable bookmark usage, clumsy URL bar drop-down, not all extensions are compatible due to the new PKI/authentication mechanism (including my own SmoothWheel, although it does work on a private test build), the new theme could use some refinements, etc, but overall, a pretty solid build.
Memory usage is definitely down, especially after a long period of time with many tabs opened/closed. A definite plus.
Another definite plus, is the snappiness. IMHO it feels like (damn, finally) each tab to its' own. I can open many tabs from the same page much MUCH faster. On FX2 after I opened a tab, I had to wait a sec or two (or more, depending on the general "load") for FX to "stabilize" before I could open another tab. No more. big YAY
that must have been the most annoying issue of FX2 for me.
So overall, FX3 seems like a winner to me. Can't wait for the final release, although I can fully recommend the (slightly) adventurous to check out beta 4. Well worth it IMHO.
Edited 2008-03-14 03:58 UTC
But I have a feeling that after Firefox 3.0 is released, there will be no further effort to improve the memory situation or the lack of proper integration on non-Windows platforms. It annoys me for something which talks about being the product of opensource, such little resources are actually used by Mozilla to improving the experience on alternative platforms.
I know I sound bitter but I've been waiting for over 2 1/2 years for Mac OS Aqua form buttons to *FINALLY* appear in the Mac OS X build, how long before we start seeing an effort for proper integration with Mac OS X beyond the superficial?
hope you have tried it yet btw after your last misinformed comment On firefox another over 70 memory leaks have been plugged.
Can't wait till launch I need to start getting my quotes ready.
I never denied the fact that there have been improvements, the problem I have is that they're just going to sit around and stop improving it after the dust settles over memory leaks. The tracking of memory leaks should be an on going project, and they should be part of the routine updates in future releases.
The only reason they've actually done a damn thing about memory leaks was the pissing and moaning from people like me; screaming from the rafters over the crap memory management. Now that they've fixed *some* but not all of the memory issues, are we going to see them de-allocate resources now that the majority of alarmists like me have chosen to go silent? that is the issue I have.
I want to see Firefox succeed, but I have a problem with organisations who, after one achievement, they lack the motivation to keep moving forward; if Firefox developers were *truly* committed to the memory leak issue, they would assemble a auditing and benchmarking team whose sole job is to fix memory related issues. When I see that happen, and when I see the fixes at the top of the list to be merged, then I'll believe that they're committed to it for the long term.
They always have been, it's just that a lot of the stuff that's gone into FF3 has been major architectural changes like the garbage collector, not simply freeing a pointer.
Isn't that exactly what's happened? Maybe they don't have a formally designated team, but they've created a bunch of tools and tests to help keep this from being a problem in the future.
I never denied the fact that there have been improvements, the problem I have is that they're just going to sit around and stop improving it after the dust settles over memory leaks. The tracking of memory leaks should be an on going project, and they should be part of the routine updates in future releases.
The only reason they've actually done a damn thing about memory leaks was the pissing and moaning from people like me; screaming from the rafters over the crap memory management. Now that they've fixed *some* but not all of the memory issues, are we going to see them de-allocate resources now that the majority of alarmists like me have chosen to go silent? that is the issue I have.
I want to see Firefox succeed, but I have a problem with organisations who, after one achievement, they lack the motivation to keep moving forward; if Firefox developers were *truly* committed to the memory leak issue, they would assemble a auditing and benchmarking team whose sole job is to fix memory related issues. When I see that happen, and when I see the fixes at the top of the list to be merged, then I'll believe that they're committed to it for the long term.
lol, that's the funniest thing I have read all day. Like I say its not name and shame time yet...but seriously Firefox has better than pleasing people who don't try their products, like say the 10's of millions in cash they receive, the none too subtle attack on their market share by apple, the waste of space Internet Explorer showing willing to get back in the market again...by simply being good enough, their move into the portable market traditionally dominated by opera. Hell why not even pride in their work.
No its down to people like yourself who shamelessly promote alternatives, and use subterfuge instead of reading the article presented.
Like I say if as a anonymous poster on the internet you take credit for what has been a fabulous beta whose features far outstrip that of reducing memory leaks, using an engine who's development started before 2.0 release seems crackers, the mere suggestion that they would sit around drinking tea, and smoking fags is a little beyond belief. They do not have the luxury of being a desktop monopoly; They are in the web browser market competing with one. That is ignoring the fact that this will their 3rd Major and their development work is available to all.
...but if it makes you feel better, you can take credit for a fantastic release of Mozilla latest browser, rather than all those people actually doing the development. Perhaps you should contact Google and ask for a few million for your tireless efforts on their behalf.
Well, while I certainly appreciate Mozilla Corp's flair for good marketing, Kawai has a point. Marketing and new whiz bang features have overshadowed code quality and security to some extent. People have been complaining about FF memory leaks for years and FF devs have been claiming that those memory leaks are features for as long. IE still dominates on the desktop, despite FF's relative success. But everyone expects that, so it's not really an incentive. But with webkit beating the pants off them on acid3, showing them up on code quality, blowing them away for speed, and being open source, as well, I think Mozilla Corp has received a wake up call.
Give me a fast and memory efficient browser that doesn't suck up memory needlessly. And one that does not take hours (yes, hours) to render long tables, as one needs in reporting applications, and I'd be happy.
It seems to me that this aggressive freeing of memory would eat the CPU, and this on a browser that can already peak my CPU usage when rendering pages. Reclaiming memory over time for Firefox is good (a must!) but I do hope that they allow the policy to be tuneable in about:config so I can have some control over how aggressively it behaves.
I don't imagine, and it does not appear from the writeup, that they tested *my* average usage patterns, namely 100+ tabs left open for months at a time... or until the kernel kills Firefox for using too much memory (thank goodness for Session Manager)! Still, leak fixes are good and any improvement in handling retention of cached information is worthwhile, even if it doesn't help me much.
CPU usage is cheap in terms of user perception (unless they're doing one of the rare sustained CPU operations, like mathematical caluclations or encoding/decoding). Excessive memory usage is far more expensive (since it often results in hitting the disk for things, which is far more noticeable than CPU spikes).
Hear! Hear!
I run about 70 concurrent Gnome users on a dual Xeon 3.2GHz box. Not really much processing power by today's standards. And yet I rarely feel constrained on processor. I do, however, feel constrained on ram. I really need a minimum of about 64-96MB of physical RAM per user. Running the latest Fedora x86_64, it really takes about 128MB before I feel I have some breathing room. So I don't take kindly to the cavalier attitude Firefox devs have traditionally taken with regards to trading memory for processor. Especially, when the cost is unpredictable. For example, caching prerendered pages. We generate some pretty large reports through our browser based report generator, and I do not particularly care to have FF caching the last 8 runs.
And don't even get me started talking about the way FF just looks at the physical ram of the machine it is running on and sets its values based upon that... without ever considering just how many dozens of copies of FF might be running.
I know that this is strongly user dependant, but I have a measley 2Gb ram in my machine (cost all of £30) and I feel that Firefox has more CPU related issues than memory issues, not that I have any real performance issues with any of the current browsers. Even running in Virtual machines, I don't feel any issues.
In my experience, the only cpu-related problems firefox has had were when opening many new tabs at once (supposedly much improved in FF3) and in plugins. With 15-20 tabs open, Firefox 2 was basically unusable on my old machine with near 100% cpu usage. Then I installed AdBlock, which shut off all those Flash ads and cpu usage was back under 5%.
Edited 2008-03-14 19:12 UTC
I've done some informal testing on FF2, FF3, and WebKit, just to get an idea how they fare as terminals for my most demanding intranet app. It is an adhoc report generator. I set up a report that generates a 12 column, 15000 line report of items with item number starting with S thru Z. The actual data to be displayed is about 900kb in size.
FF2 seems faster than FF1.5, which is the last version I had tested until now. I don't have actual data for 1.5 to compare to, but as I recall, 1.5 was very slow for this kind of work.
FF2 renders the table in 26 seconds and uses 310MB at the end.
FF3 renders the table in 18 seconds and uses 180MB at the end.
WebKit renders the table in 2.5 seconds and uses 120MB at the end.
I tested printing by printing to a file and observed the results. FF2 and FF3 both print in 2 phases. First, a window pops up that says it is "preparing". This lasts about 30 seconds for both FF2 and FF3, and the main windows *does not redraw* for that entire period, so any windows that move over it create very ugly results that make it seem like the user's computer is "locked up". After that, you get another window with a progress indicator. While that is up, the main window *does* redraw, and the user sees progress on the bar, so even though it takes a while to complete, its not really a problem.
Total printing time for FF2 is 6.5 minutes. Total priting time for FF3 is 5.5 minutes. GTKLauncher, which I am using to test WebKit since I don't have the WebKit enabled version of Epiphany, does not support printing, so I'm not able to test that.
This would also be a good time to point out where FF's caching of prerendered pages can really cause it to fall on its sword. The browser starts out with about 30MB of resident memory. (I have lots of memory on my test machine so all pages are in ram.) FF2 ends up at 310MB. So I figure 310MB - 30MB = 280MB is about the size of the rendered data. Now, imagine that a user runs several reports. The XDMCP server the browser is running on has 8GB of memory. So FF2 autotunes, based upon total system ram, to cache 8 prerendered pages.
8 * 280MB = 2240MB
So in this case, FF2 is caching 2.2GB of useless pages. The other 69 users on the system tend to notice when someone does this. The Mozilla site says that an average rendered page takes 4MB. Kind of like the average American family has 2.3 children.
Of course I've taken steps to ensure that the above nightmare is not allowed to occur. But still, the FF devs would do well to think beyond Aunt Tillie and her whiz kid nephew watching YouTube videos and reading GMail, and think about the wide range of things that the browser is called upon to do when someone like me has enough faith in the power of web apps to try to use them in a heavy duty fashion.
Edited 2008-03-16 15:42 UTC
What about it? ;-)
Actually, I could test with that, but all I have available is IE6 under Crossover Office.
I did test Opera 9.26. It rendered in about 2 seconds and used 180MB. It also prints in 2 phases. The first popup lasts about 15 seconds and the main window does not redraw during that time. Then it pops up a window that says "Printing page 1" and sits for a long time. Although the main window will redraw at that point, it might take 5 or 10 seconds before it does, which still looks pretty embarrassing. I presssed cancel after 6.5 minutes and the popup went away, but screen redrawing was still quite slow, and the opera process was still pegged at 100% processor.
Actually, my major complaint about all of them is lack of proper progress indicators during prolonged periods, and their nonthreaded nature, which makes them all quite clunky. The user doesn't so much mind waiting as long as he knows that its not locked up and has some idea how long it might take to complete.
Another notable point is that we are talking about displaying 900kb of raw character data in a table... and even WebKit sucks up 120MB to do it. So even the best of them is exhibiting well under 1% efficiency on memory use.
I know tables have fallen out of favor, of late. But this is not an abuse of tables as a formatting tool. This use is exactly what tables are for.
Edited 2008-03-16 22:06 UTC
:) I feel your pain; You are the unluckiest man alive. Time and time again. Its you that has all these problems. I do I feel for you, with all those roles you have, with bosses that make you use a particular selection of hardware, write programs a certain way, or satisfy those disgruntled customers with only a Microsoft solution.
Its like spooky or something, your patience and understanding of FOSS and its products that you just wish fitted that need, but just can't.
I can only find it somewhat ironic that someone who is so disrespected he can't choose his own hardware; cant deliver futurepoof solutions based on standards to his own customers, leaving his customer tied to frighteningly unsafe technologies like ActiveX...perhaps there is a connection. I think you should rethink your choices. I think they have tapes for that kind of thing.
On a unrelated note, I wonder if you has noticed the whole 44million on Yahoo's table left there by Microsoft. Its not left there for companies...its left there for Aunt Tilly to buy more quality produce advertised in her browser mind you its not just silver surfers its Uncle Tom; Little Jimmy etc etc.
Seriously if you need help and advise on asserting yourself I'm here for you.






