I have used the Firefox 3 beta on a couple of test machines, and it seemed fast and reasonably lightweight. So I was surprised to see this post from NeoSmart today. After a snippet from a Mozilla paper touting big improvements in memory usage and a concerted effort to stomp out memory leaks, the author (there’s no by-line on the post) notes:
Firefox still uses a lot of memory – way too much memory for a web browser.
We haven’t seen it reach 1GiB+ like we have with previous versions, but it’s quite normal for Firefox 3 to be sucking up ~300MiB of memory right off the bat, without a memory leak (the difference between memory leaks and normal memory abusage is that in a memory leak you’ll see the memory usage keep increasing the longer the browser is open/in-use).
I’ll take a closer look later, but meanwhile I throw it out to you. The final release is just around the corner. Who’s been using Firefox 3? Are you happy with it? Does it demand a lot of your system resources?
Ed Bott 
I’m using Firefox 3.0RC1, and though I am quite happy with it, it’s using as much memory on my system as v2.0 did.
It definitely uses a ton of memory… the main difference is that I don’t think I’ve seen Firefox 3 go over 200mb more than once… wheras with my same workflow on Firefox 2 it would regularly see 500mb (and then hang and die)
But that’s also likely a result of the CONSTANT crashing problems I have with Firefox 3. There’s no way it can use up that much memory since it’s constantly rebooting itself =)
Am using Firefox3 rc1 and loving it.
It is lighter definitely lighter than FF2. I didn’t write down exactly how much RAM FF2 was using, but it’s obviously lighter.
Anyhow, I did not conduct precise tests (opening the same pages in both FF2 and FF3 and comparing), but it does start up faster and runs noticeably faster than FF2… I’d say the article you are referring to is simplistic FUD.
What does “way too much for a web browser” mean? How much exactly is a Web browser supposed to use? Is there a norm?
I don’t see FF3 using 300MB right off the bat… and am running Vista x64 Ultimate with 8GB of RAM. My FF3 uses ~80MB at stratup (my home page being iGoogle, with a dozen gadgets).
I just opened the same set of web pages in FF3 and IE7 after freshly starting them both, FF3 is using 95 MB, IE7 is using 110 MB. Frankly, the difference is peanuts here, who cares? If you really want to nitpick on memory usage, then sit down, open the same pages in all the browsers you wish to compare (on the same system) and write down what you see.
Otherwise, you’re just wasting everybody’s time writing insipid FUD with words like “quite normal”, “right off the bat” and “way too much”. No facts basically, just random ramblings by some random person on the internet.
Here’s a fact for you Ed: one of your readers uses FF3, loves it, and disagrees “right off the bat” with what that “NeoDumb” guy has to say.
@The How-To Geek:
My FF3 runs for days without the slightest hiccup nor any increase in memory usage (I never turn off my machine, nor do I close my browser). It’s very simple: I close some tabs and memory usage goes down, I open some tabs and memory usage goes up… FF2 used to go on forever as well (as longs as the OS would), I did catch FF2 using ~500MB every now and then indeed. I did not notice any such thing with FF3 yet. Closing all tabs basically puts FF3’s memory usage down to ~80+MB on my system… which seems very correct to me)
BTW, I’m the guy who used to have BSODs with Vista (I left a comment here some time ago complaining about it). I was having trouble with Vista up until 2-3 weeks ago, where I got fed up, took WinDbg, and debugged the minidump from all my BSODs (i had around one BSOD every 2 weeks… followed all what the diagnostics I could find had to say etc, but could not pinpoint where the BSODs were coming from).
Anyhow, I learned how to debug minidumps (thanks to these guys: http://blogs.msdn.com/ntdebugging/default.aspx) and went ahead. I found out that my problem was coming from the NT kernel module, when handling a page fault. I knew immediately what the problem was: I set the page file size manually to 1GB (because I have lots of RAM already… figured I didn’t need virtual memory much). That made Vista crash apparently. I reverted the setting to let “Automatically manage paging file size for all drives” and did not have any BSODs since.
Thanks for the pagefile story, Joe. I’ve been recommending for a long time that users not mess with the pagefile; this is a great example of why.
Yes indeed, it is a great example of why one should not mess with the page file size, I learned it the hard and frustrating way 🙂
However, I would argue that Microsoft should then deactivate this possibility altogether. This is a serious bug in the virtual memory system. How come I am allowed to set the pagefile size when it makes my OS unstable? How come one can not turn off virtual memory safely? I have 8GB of RAM, isn’t that equivalent to having 4GB of RAM + 4GB pagefile? Why do I need a 8GB pagefile? I have a Raptor 10KRPM hard drive, which unfortunately has only 139GB of formatted capacity. The pagefile is eating ~6% of my hard disk space, for no real apparent benefit to me, the end user/developer. I bought lots of RAM precisely because I don’t want my system to swap.
If I had 16GB of RAM, I would have to sacrifice 12% of my drive capacity to the pagefile… 🙂
Many questions for the kernel guys at MS 🙂
I will not change the page file size ever again, I promise that 🙂
Great questions, Joe, and worth a post on their own, which I will do.
A couple quick hits though:
“I have 8GB of RAM, isn’t that equivalent to having 4GB of RAM + 4GB pagefile?”
Nope. You’re thinking in 12 year old terms, when Windows had a swap file that was used to augment physical memory. Today, the pagefile is used as a way to quickly map pages in and out of memory to maximize performance. There are circumstances under which it is useful to adjust it, which is why the option exists. But it’s also hidden under the Advanced tab of System Properties, which should indicate it’s not for casual use.
“If I had 16GB of RAM, I would have to sacrifice 12% of my drive capacity to the pagefile…”
No, you wouldn’t. The page file resizes dynamically, and available disk space is one consideration it takes into account.
I’ll have more to say later. Thanks for bringing this up.
I did some testing on Firefox 3 Beta 4 versus Firefox 2 memory usage here:
http://www.brightrev.com/reviews/web/5-web/20-firefox-3-beta-4-first-look.html
In short, Firefox 3 B4 uses less memory than Firefox 2 in every circumstance. It also recovers more memory when browser tabs are closed.
I actually never had a big problem with Firefox 2, because I’d take 10 seconds to close and reopen the browser a couple times per day. Regardless, based on my testing I’d call the original article bunk, or at least suggest that they check their extensions.
FF3 Beta 2 sucked. Crashed all time. I went back to FF 2.x.
But FF3 RC1 is great. It’s fast and stable. It crashed yesterday, for the first time since it was released. I’m not going back to FF 2.x. I think memory usage is better but I’m not sure if it’s significantly improved. But I don’t care. It’s super fast and I love the features.
IE8 has a tough hill to climb.
So I’m not sure what the tip about the pagefile is IF you do go ahead and manually adjust it. Is it not to have too much of a differential between RAM and the pagefile (in the example, it was 8:1)? In my case, I see that my pagefile is about 80% the size of RAM and I’ve not had an issue with crashing or out-of-virtual-memory errors, so it’s a good fit. I’m aware of the old rule-of-thumb about making it 2X RAM, or whatever, but it was overkill here.
On the subject of the post, I think there’s more to that story, as having FF3 take 300MB+ RAM with no extensions, few pages, and within a half hour is on the edge of impossible. It should be less than 1/3 that without even trying. I’d chalk it up to an Ubuntu thing, but he says it happens on Windows too.
Rick, look up “page file” in Windows Vista Inside Out (or in XP Inside Out, for that matter).
Ed, I meant more in terms of avoiding BSOD’s. Joe seems to have hit on a specific case of having a ton of RAM and very little pagefile. That’s not something I’ve heard about before. I’ve seen plenty of pagefile-setting advice, but aimed at things like performance, not to avoid a severe problem.
I’m looking forward to the article on page file 🙂
My understanding of it is indeed extremely limited. I thought the page file was simply a holder for virtual memory (extra RAM on disk basically…). I may be completely wrong. I don’t see however how using the hard drive can speed up RAM management 🙂
The topic is indeed not related to the post, but I wanted to share my newfound stability for Vista (I come to this blog only occasionally… hope you pardon my “off-topic-ness”).
Also, for some reason, my pagefile seems to remain constant, Windows is not resizing it apparently. Every time I look, it is using 8GB precisely. I do not uncheck the “Hide protected OS files”… because I don’t want to accidentally do anything to them 🙂 So I look only every now and then, like today.
I should say that I run a ton of extensions, and almost always have 20 tabs open… so my results might not be average.
I’ve noticed that RC1 seems to use more memory than beta 3 did. Not sure if it’s just me (or my extensions)
I’m no Windows kernel expert, I may be wrong again. But I did not have a BSOD since I made that change (I will have to wait more to confirm of course). Here’s a typical stack trace from my minidumps. Most of these BSODs were triggered by a user-level application (one of them was firefox.exe BTW, 2 of them were my own exes, which do consume a lot of RAM, on the order of 500MB). In almost all stack traces, a certain ‘KiPageFault’ function from NT kernel was involved.
nt!KeBugCheckEx
nt!KiBugCheckDispatch+0x6e
nt!KiSystemServiceHandler+0x7c
nt!RtlpExecuteHandlerForException+0xd
nt!RtlDispatchException+0x22f
nt!KiDispatchException+0xc3
nt!KiExceptionDispatch+0xa9
nt!KiPageFault+0x1e5 (TrapFrame @ fffffa60`0a00ed80)
OK, that’s my last post on page files. I’ll post back again on Ed’s thread on this.
Ed please delete this comment if it becomes too much of off-topic stuff, I won’t be mad 🙂
I haven’t had a problem with FF3 since the first beta was released. From what I’m reading, it must have to do with how long you keep one session open for. I constantly open and close any browser I’m using so must not ever get to a point where I”m having problems. I haven’t monitored my memory usage because I don’t feel I have a reason to. But when I do look at memory usage FF3 seems to use anywhere from 60-100 MB of RAM.
Ah, the page file. Back when I worked IT we had to kill it on all servers for a government project. I question the usefulness/security of that strategy, but whatever. (logic was a tough sell in that environment)
As far as Firefox 3, I’ve been running various Mac betas for a few months. Other than some of the beta flakiness you’d expect (FF 3, Beta 4 was crash prone), I find it more snappy and stable – doesn’t drag my machine to a crawl after a few hours with 15 tabs.
As a Flock user, I’m hoping that it is lighter….I have to run a lot of tabs all the time to keep up with the tech news and i sometimes see as much as 600MB taken up by it. Flock will be getting all of the back-end FF3 enhancements shortly after launch. Then it will be Flock 2.0, btw! Flock on!
Hi,
i used FF 3.0 dev and now i`m using FF 3.1a and both are using less memory than FF 2.0 was using.. so there are improvements with the so called Firefox Memory Leak/Bug. In my eyes it is not as perfect as it should be (for me..) but it is quite ok and great, that FF 3.x is using less memory than all versions before did. I cannot agree to the refered article and that Firefox 3.x is eating up to 500MB of Ram..
Sorry for my english
sincerely
Chris
Firefox 3, running on Windows XP, does seem to use less memory in my experience. and most importantly the memory usage seems to stay largely constant; i.e. it doesn’t keep going up and up as you use it.
Speed-wise it’s faster on one XP computer and about the same on another XP computer that I use, so your mileage may vary. It is much, much better on a Mac, however.
Firefox doesn’t use as much memory as it predecessor, and uses a similar amount in comparison to other browsers (i.e., Safari, Opera, IE, ect). However, there is a VERY small tool called the, Firefox Ultimate Optimizer, and instantly reduces FF memory usage. I am currently using it, and FF is running on 276k w/o delay. This tool has been tested by such antivirus kings such as, Kaspersky and Trend Micro, and has been found to be clean. If you’re an FF you’ll love it; it can be download at http://firefox-ultimate-optimizer.en.softonic.com/
Those guys at neosmart are looking for some web traffic & need some ass whipping..so do you ed!
Chris, I asked a question of the community here and got some great input. What’s your point?
It’s NOT a memory hog. Period. It uses very little ram on my aging PC. Memory usage is usually b/w 50 & 100 MB’s..I’ve never seen it go above & i usually have at least 7 tabs open on an average.
I’ve had a wonderful experience with FF 3 in my Ubuntu installation. I’ve done a couple of tests and have found that FF seems to operate faster, and use less memory in Linux compared to Windows
Now that I’ve located beta versions of extensions I like to use (TabMixPlus, FireFTP, Firebug) I’m very happy with Firefox 3. Been using it three weeks/ nine hours a day. The more I use it the more I notice small improvements. No crashes and no memory issues.
The most problematic application I use regularly is Internet Explorer (IE8 beta).
I use Firefox with multiple user profiles which I start or stop several times a day (profiles for casual browsing, research, another for development, et cetera) I have a Windows XP computer with 512 MB of RAM. The CPU is Athlon XP clocking at 1.47 GHz.
I rarely have memory issues to speak of. I’m reasonably careful to avoid overtaxing the computer. I don’t use heavy IDE’s or wrangle huge photos. If I have any issues whatsoever (which is rare) they’re related to CPU usage or wireless network hassles.
I do web development and programming all the time. Granted, I don’t use virtualization, heavy-duty image processing, and DVD pirat..err, encoding. Gaming is out of the question.
I could have mentioned the following:
A couple of web sites which are flash-heavy can prevent Firefox from responding momentarily while the page is loading (or perhaps being processed). This has recently started to annoy me on sites like Tom’s guide or Anandtech.
I used to block flash long ago, but I stopped as it didn’t seem worth the trouble once I got DSL. A dual-core proc and more memory would likely erase these problems, I suspect. The same problem affects Fx 2.
I doubt this is a Firefox issue per se. It never crashes Firefox unless I get impatient and try to close the window or use task manager to interrupt it.
“…on our Windows PCs, it’s normal to find Firefox 3 taking up ~350MiB”
I have ten tabs open to popular web sites. Mine is using 110 MB. I don’t know where they get these numbers. I’ve never been able to duplicate the complainer’s memory usage stats in the past. I wonder what nutty extensions they might be using?
Could it be related to huge numbers of RSS feeds? That’s something I just don’t use much.
FF3 is not a memory hog. Memory usage is usually between 50-80 MB’s.
I have never seen it go above and i usually have at least 3 tabs open on an average.
Furthermore, FF3 rocks …. it’s so fast!!!!!!
I’ve seen ‘Firefox is a memory hog- it uses 300MB’ stories regularly over the last few years. OK, using 300MB out of 512MB RAM could probably be called ‘hogging’, but what about using 300MB out of 1GB- isn’t 30% just, well, ‘use’. And the blog author in question here used machines ‘ranging from 2GB to 8GB of memory’. 300MB out of 2G is 15% and out of 8G less than 4%. When does 300MB stop being hoggish and become piffling?
(Currently FF3b5 ~100MB/1G.)
Ed, I have been using the FF3 betas for a while now, and while in general I like it and it uses much less memory, I have a few issues I’d like to resolve:
1) The embedded pop-up blocker resets to “off” every time I start it. I must have a setting wrong somewhere.
2) Continental.com will frequently not let me login. I find I have to switch over to IE7 when that happens.
3) Most of my add-ins have not been migrated over yet. I suspect the official release will hurry up that process.
4) My most annoying issue is something which may cause me to avoid switching on my primary machine. I scan a lot of sites daily. While I ultimately may switch over to an RSS reader for that, today I mainly accomplish this by opening up an entire folder of bookmarks in one go. When I’m done, I open the next folder, which (in FF2) overwrites the current folders (which is what I want to have happen). Apparently, the consensus on the Mozilla development team is that this is a destructive condition, and they have eliminated this option in FF3. I am an advanced user, so I don’t care if this option is an advanced setting, but as of FF3 RC2, they are still refusing to allow this as an option. There are some add-ins that attempt to resolve this issue, but I’d rather it be a native option. Opening ten tabs, and then ten more (and having 20 and not 10 at that point) is extremely annoying to me. A discussion of this problem can be found here:
https://bugzilla.mozilla.org/show_bug.cgi?id=395024
Clearly, I’m in the minority, but there are a lot of people that feel strongly negative about this change, especially because they are not even allowing it as an advanced setting.
@Stephan: I for one will welcome the change where opening a bookmark folder will not overwrite current tabs. I would use the feature more often if that is the case. I currently have a “startup” folder that I check right away. If I ever want to check it later, I open a new window and then open the folder. I really like the tabs, and only want to see one window ever. However, given that there are people (or at least one) who like the old behavior, it seems reasonable to me (without reading any of the issues surrounding it) to have a hidden about:config option to get the old behavior back.
I ran into a bug with SSL not working on any reasonably new apache2 servers configured with the defaults, so I uninstalled, and haven’t tried again. FF2 is working perfectly fine for me, so I don’t have a reason to upgrade. I also haven’t any issues with memory like other folks that I have heard of. My machines are typically 512MB, linux and windows. I don’t have lots of tabs open usually, though when browsing phpbb3 sites, I typically open up every issue that I am going to read, so might have 30 tabs open. I don’t see any issues then either. I suspect it is javascript/flash/etc related, and since I run a webwasher proxy filter, I don’t see most of that stuff.
I’m using FF3.0 RC2. I’ve got 20+ tabs open in one window. The browser has been up all day. Memory usage according Sysinternals Process Explorer: Virtual Size 434,000K Working Set 282,000K.
Better than any other browser I’ve used.
Cheers
Mark
sadly firefox 3 still does have huge memory hog
for me its 75Mb after 5 mins..even trim doesnt helping
yes firefaux is the new ie. try maxthon and you will see that with many option, many features, secure browsing you still don’t need more than 100mb, cpu usage of not more than 5% at any time and be fast too.
it is just ridiculous how much ram firefaux uses if you leave it on especially using something related to java….190mb?…15% cpu usage? come on now. and when i try to save images or favorite site it stops a second or 2 and then starts. thats just bs.
people have fallen for the marketing PR campaign of “best browser”…it was in 2004, not now anymore. that is maxthon, try it.
This thread is a little old, I found it by searching for Firefox memory issues. I DO have an occasional problem, let me describe. First off, FF 3.0.1 (but I’ve seen it on previous 3x’s), Win XP SP3, monoprocessor, a gig of memory.
Hardly any add-ons, just stuff like Google toolbar.
Normally with about 3 tabs open, it cruises along at 100-200 or so MB. But occasionally the whole machine slows to a crawl, and I catch FF steadily consuming more and more memory, upwards of 600 MB, even without doing ANY interaction with the web pages open (although they are things like Yahoo Finance, which updates constantly).
Suddenly the usage drops back to the more normal 100-200 MB, then proceeds with the same increase pattern, the memory graph looks just like a sawtooth plot.
If I catch it doing that, I’ll close it out (and that sometimes takes MINUTES because the system is so bogged down), and after restarting it’ll behave for a while.
I haven’t run across a description quite like this; of course, I need to do more testing to see what happens when closing certain tabs, sites, etc, but usually by that time I’m so frustrated I just want to get it working again so I can get working.
I downloaded and installed a Windows XP update that fixes a GDI memory leak:
http://www.microsoft.com/downloads/details.aspx?FamilyID=9b5edfc8-a4bb-4080-9063-6518166e2dab&displayLang=en
and after that, FF3 memory has been very stable. I’ve been running this about 3 or 4 days now, and FF3 memory use is OK. So maybe that will fix the problem. I would like to hear feedback from others.