Browser

Firefox 3.0 memory improvements -- Fact or Fiction?

Firefox 3.0 is supposed to be less resource hungry than Firefox 2.0. Sadly it's not. I ran a test that showed that Firefox 3.0 uses more memory and releases less of it when it's done.

Firefox 3.0 has been a long-awaited update to Firefox 2.0. Mozilla.org has touted many of the improvements in Firefox 3.0 in the areas of speed, standards compatibility, and browser features in general. One of the biggest areas Firefox 3.0 was supposed to address was memory usage.

Firefox is a great Web browser, and I use it daily. One of the major irritations, however, is the fact that it likes to gobble up memory as you browse around the Internet with it. As you open and close sites, whether in separate windows or separate tabs, Firefox scoffs up more memory and very rarely releases most, if any, of the memory back to the OS when you close things. Eventually, the whole system slows down, and you have to close Firefox or, worse, kill the process to get the memory back. When memory usage grows too big, Firefox 2.0 invariably locks up or crashes. I've noticed this running Firefox in OS/2, Windows, and Linux, so it's not an isolated problem to one operating system.

Firefox 3.0 to the rescue?

With any new release of software, you can almost guarantee that the new version is going to use more resources and be more of a pig. Mozilla.org claimed that Firefox 3.0 was going to be different. So, when Download Day hit, I wanted to make sure to grab the final release of Firefox 3.0 as soon as possible. I then decided to see if the memory usage had improved any.

Sadly it did not. If anything, it got worse.  I conducted a semiscientific test on my production machine. This is a 2.8 Ghz P4 with 2GB of RAM running Windows XP SP3. I first ran my install of Firefox 2.0 and measured the memory usage of Firefox when visiting TechRepublic's Web site. I went to the home page, loaded all the tabs across the TR front page, measured usage, and then closed them. After that, I installed Firefox 3.0 and did the exact same thing all over again.

I compiled all my results in a Firefox 3.0 Memory Usage Photo Gallery. I was pretty disappointed. Not only did Firefox 3.0 start off using more memory than Firefox 2.0, it also failed to release all the memory back. In fact, it held on to 1Mb more memory than Firefox 2.0 did.

I tried the experiment a couple of times on the same machine. The memory usage varied each time, but the results were about the same every time. Firefox 3.0 took more memory and released less when it was done than Firefox 2.0. The next step is to install Firefox 3.0 on other machines to see if it's a trend or just some type of fluke.

(Update 6/19/2008 4:00 pm:  I reran the tests on another machine running openSUSE 10.3. The results were essentially the same. Firefox 3.0 took more memory than Firefox 2.0 and refused to give up the memory when I closed the TR tabs. Check the Photo Gallery for the new screen shots.)

The bottom line for IT Leaders

Although anecdotal evidence of a problem isn't the same thing as rock-solid proof, it can be strongly influential. Firefox 3.0's seeming memory problems show even more that you can't always trust the claims of a software vendor when they're telling you how great the next release of a product is going to be. Having a testing environment in place is imperative. Check the claims and make sure everything works the way it's supposed to before you blindly roll out a new product.

41 comments
notforattribution
notforattribution

It's not poor memory management. FF3 is not ready for anything but beta--if that. Font rendering is unstable (I've had pages switch to Greek, Cyrillic and Norwegian??) in mid page. The new UI is worse. I work on different computers, and Firefox was the one respite application I could count on to look and work the same way for Apple and Windows. No longer. FF now has Windows, Apple and Linux versions. No, they don't look alike, nor do they have a similar feel. There's no good reason that the same application should look and work differently on different machines. My favorite command--CtrlT for open a new tab--is GONE!!! Actually, it's almost gone. It doesn't work about half the time I try it. Why bother with tabbed browsing if you have to fumble with a mouse? Worst of all is that FF3 Windows looks like a Disney version of XP--lots of blobby, bright (distracting) and 'cute' but essentially useless icons and graphics clutter dropdown menus and popups, making them obstructively big, and consequently make them obstruct the browser window. The purpose of a browser is to retrieve data from another site, not to get in the way of real work by showing off what it can do while getting in the way of real work. Don't even start about the information bar: It blots out the top of the browser window while trying to second guess what you're typing. I can get google by clicking home when I want a search engine. I loathe sifting through dusty, useless guesswork URLs when I'm typing an address--especially one that I know & use. My experience with the 'smart' feature? It takes 2 - 12 times as many keystrokes to get to a url than the old history feature did. Tomorrow, I roll back to FF2--which I understand is fraught with difficulties and takes several hours. It's worth it

ijusth
ijusth

I find the IE also has this same problem with memory leaks. I end up needing to close running apps or windows to even OPEN taskmanager so I can try to kill processes. One thing that FF3 at least fixed was a know problem with lines across the screen. FF2 had issues when you scrolled with lines across the screen. SO far I have not seen this problem anymore. Mozilla giveth and taketh away.

NickNielsen
NickNielsen

Although it does appear FF3 is using a bit less. But that's probably because 90% of my installed add-ons are disabled because they are incompatible with the new version.

Bedwetter
Bedwetter

I see an improvement after using it for 24 hours on Vista Ultimate. Less memory usage and it's released rather quickly on close. I'm using the dashboard memory usage meter on Vista. Not terribly scientific but ... It also works better from a user point of view, loads page views quicker than FF 2.x

ptomes
ptomes

"Check the claims and make sure everything works the way it???s supposed to before you blindly roll out a new product." Maybe you should make sure to do better testing to prevent your false claims. http://ejohn.org/blog/firefox-3-memory-use/ http://blog.pavlov.net/2008/03/11/firefox-3-memory-usage/

John Sheesley - TechRepublic Pro
John Sheesley - TechRepublic Pro

Check the screen shots and judge for yourself. The screen caps were all made on within 1/2 hour of each other, so the content on the site was essentially the same. I loaded exactly the same number of tabs. I made sure I showed the full screen monitor, not just bits and pieces so it wouldn't be possible to say that I was running some program that changed the results. There's nothing 'false' about those results. I did mention it might be something isolated to this machine or OS. But for now I wouldn't want to fatten up a memory hog like Vista by feeding it Firefox 3.0. Have YOU tested it personally or are you just doing exactly like I said and trusting the words of a developer?

ptomes
ptomes

(Sorry for my English, please) Yes, of course, I did test it extensively on several machines. http://lh4.ggpht.com/ptomes/R9a8DVjwwQI/AAAAAAAACkY/tBvqxIklrPI/3.png http://lh4.ggpht.com/ptomes/R9a3ZVjwwPI/AAAAAAAACjg/FBJgLjzbEqw/2.png http://www.phil.muni.cz/~ptomes/cache.swf http://www.phil.muni.cz/~ptomes/fedora.swf Mr. Sheesley, it seems you should use the browser for some time to see improvements first and then do relevant conclusion not based on one short experience (anomally?). I believe you agree with this.

John Sheesley - TechRepublic Pro
John Sheesley - TechRepublic Pro

As I said, it was only a semi-scientific test on one machine at the time although a second XP machine has shown similar results. But thats exactly what I was looking for from the TR audience - a confirmation of the results on multiple machines to see if it was a trend, or as someone else said here, some anomaly on one machine.

John Sheesley - TechRepublic Pro
John Sheesley - TechRepublic Pro

In Decision Central, I pointed out how shocked I was that Firefox 3.0 was as much, if not worse, of a memory pig as Firefox 2.0. http://blogs.techrepublic.com.com/decisioncentral/?p=117 Of course, it's a pretty unscientific test that I plan to replicate on other machines, but what better test platform than the TechRepublic audience? If you've tried Firefox 3.0 yet, what's your experience so far? Is it an improvement or just as much of a pig as Firefox 2.0?

boxfiddler
boxfiddler

What a nightmare. More than two tabs open at a time and it froze up, badly enough in some instances that I had to close out, then reopen to return it to functioning. Page hangs everywhere I went on the web. For some bizarre reason it didn't appear to be able to deal with whatever differences exist in varying versions of Flash. I installed two different versions five different times. Dumped it after a week. Back to running 2.11 with no problems. Sheesh. etu

techrepublic@
techrepublic@

I have lost count to the number of FF3 I have already installed. It is the default in all desktops I have put my hand on. In all those systems, FF3 has worked with no problem. Unfortunately, your experience is very negative. To improve FF3, Mozilla needs bug reports from people with problems. As a programmer, I know very well how difficult it can be to spot the problems without good quality bug reports. Please make a small effort and post bug reports of the problems you are encountering. Bug reporting guide: http://developer.mozilla.org/en/Bug_writing_guidelines

antithesis77
antithesis77

We're currently developing some Flash/AIR apps with a touch of Java. Unfortunately, the RAM usage on FF3 has skyrocketed compared with FF2... I would to be able to proclaim to our potential users that our app isn't resource hungry, but this kinda sucks for us as developers!

TroyW
TroyW

Because on all 3 machines here at this residence that I've installed FF3 on, memory usage has dropped considerably, and it's been much more responsive then FF2.x I even took a screen capture of the task manager showing FF2 being closed and FF3 being opened with the same tabs on the oldest machine here: http://img.photobucket.com/albums/v411/aussiemadman/Computer/firefox2_vs_3.gif

Toretto84
Toretto84

an open-source product which, by definition, is always a "work in progress", is an "example" how software vendors don't always deliver what they claim? Last time I checked, I didn't PAY for firefox 3.0... I downloaded it, knowing very well that there's still issues that will be fixed during the next updates. Memory problems or not... Firefox 3.0 > Opera > Safari > IE :D

LinuxConvert
LinuxConvert

Firefox 3.0 seems to be following the MS bloatware-path. I've switched to Opera. I still use Firefox but, as you pointed out, it seemed to slow the systems the longer it was running. I use Opera, Firefox then IE in that order depending on what I need. Systems: WinXP SP3, 1.8G, 1.2G RAM GNU/Linux Ubuntu 8.04, 1.5G, 512MB RAM

techrepublic@
techrepublic@

A while back I read an article about browsers memory usage and how firefox 3 was a great improvement over firefox 2 and even opera. Curiosity got me so I did a memory usage test on Firefox 2, Firefox 3 beta, Opera. The test used javascript to load 1000 pages (collected from my history and bookmarks). The script opened and closed pages on a 10 seconds period, keeping 15 pages open. It made a 5 minute pause every 100 pages and in the end closes all pages. I then waited ~30 minutes and cleared the cache. Externally a shell script sampled the systems (not just the browsers) memory usage every second. I choose to use memory system usage instead of only browser memory usage because browsers store data and cause change in other processes (X server). I set browser cache sizes to the same value and cleared caches at start. Firefox 3 won hands down, on average using less memory and recovering memory faster. The Firefox 3 graph was almost always well below all others. There was no clear second place with Firefox 2 and Opera competing for it. One thing that surprised me was that Opera even after clearing the cache had a much larger memory usage than at start. If I have time and inclination, I will rerun the test and post them.

John Sheesley - TechRepublic Pro
John Sheesley - TechRepublic Pro

That would be pretty interesting to see. As I said, this was a pretty semi-scientific test and was only limited to doing stuff on TR's site.

techrepublic@
techrepublic@

WARNING: This benchmark was run on my laptop using scripts with little testing or review (a.k.a. bug infested :) ), using URLs from my browser's (firefox) bookmarks and history. Your results WILL DIFFER. The browsers: Firefox 2.0.0.15 Firefox 3.0 Opera 9.50 Firefox testing was run using my (almost) normal setup (noscript and adblock where turned off), with 47 extensions enabled. Description: The benchmark was run on a GNU/Linux Mandriva 2008 Spring with 2G RAM. The browsers memory cache was set to 40MB and the disk cache was set to 100MB. The cache was cleared at the start of the test and at the end to see how much memory was used by the cache. 200 pages where loaded every 5 seconds and the memory and CPU usage was recorded using a shell script. The system was left untouched during the benchmark runs. Results: Firefox 3 used the least memory, Opera used 69% more than Firefox 3 and Firefox 2 used 77% more memory than Firefox 3. Opera used the least CPU, Firefox used 15% more than Opera and Firefox 2 used 79% more than Opera. In the end, clearing the cache released 23 MB in Firefox 2, 14 MB in Firefox 3 and 8 MB in Opera. Comments: In this test, Firefox 3 shows a great improvement in both CPU and memory usage over Firefox 2 and even Opera. Final warning: These results are (I hope) valid for this benchmark. Other benchmark will shows other results. CPU Usage Graph: http://pmc.com.pt/browser_benchmark_comparisson/browser_benchmark_comparisson_cpu.png Memory Usage Graph: http://pmc.com.pt/browser_benchmark_comparisson/browser_benchmark_comparisson_mem.png Data: http://pmc.com.pt/browser_benchmark_comparisson/browser_benchmark_comparisson.ods HTML Page: http://pmc.com.pt/browser_benchmark_comparisson/browser_benchmark.html Shell Script: http://pmc.com.pt/browser_benchmark_comparisson/browser_benchmark.sh The list.js file with the list of URLs is not included. You have to make your own if you want to run the benchmark.

linux for me
linux for me

His results of testing Firefox seems to be the opposite of yours. And I am running Firefox 3.0 on my Vista system and it appears to use less memory and is faster. Your mileage may vary, but I'll keep FF3 over IE anytime.

steven.taylor
steven.taylor

While not a big fan of IE 7, I do have to use it while at work. I knew there would be the comments here about it ("it's not as bad as IE", etc), I thought I'd put it in perspective. On my PC, which is almost the exact dupliate of John's, I did the same experiment with IE 7.0 IE with 1 tab: 33.5 MB vs FF3 at 54.7 IE with all tabs: 101.5 MB vs FF3 at 81.7 IE back to 1 tab: 65.1 vs FF# at 54.7 On a 2 GB machine, I'm not going to split hairs. There just isn't enough difference to make this a selling point on either browser. Other than that: 1. FireFox 3 breaks a lot of plug ins. 2. It has security issues with it's new malware detection feature 3. While the debate is still out, memory usage doesn't seem to have improved. While it may be faster and have other cool stuff, it's just another peice of software.

jdclyde
jdclyde

it also seems to respond quicker. Handling it better? is total memory used the bench mark we should look at, when up until now benchmarks were ALWAYS done based on how long it takes to complete a task? Only been on it for two days, but so far so good, and haven't seen any reason to move more systems over (personal systems) for further testing. I have also only seen one person complain and that was about gmail, which works just fine for me and others who replied, so I have to assume there is something else on the posters system that is pooched.

jphallett
jphallett

I tried using it for about a week when the final release was available. I kept experiencing memory errors and the browser would close. I usually have about 5-6 tabs open on different pages. I uninstalled and went back to Firefox 2 and I have not had the issue. I am running Windows Server 2003 SP2.

jmgarvin
jmgarvin

My only gripe with FF3, is that about 1/2 my plug-ins don't work (yet). I'll wait a few weeks to migrate completely, but I see no reason not to.

The Maverick Phantom Wanderer (formerly Macoza, No
The Maverick Phantom Wanderer (formerly Macoza, No

On my laptop, 3.0 was much worse than 2.0 (though neither were as bad as IE7). On the other hand, Firefox 3.0 worked exceptionally better than 2.0 on my friend's laptop. So, I believe it just may be a difference between the machines used. Nodice, who suffers the plague of being machine-unfriendly.

boxfiddler
boxfiddler

not the only plague Nodice suffers.

boxfiddler
boxfiddler

self-incrimination in all its glory.

The Maverick Phantom Wanderer (formerly Macoza, No
The Maverick Phantom Wanderer (formerly Macoza, No

Rather than counter your statement by asking you why someone must be self-important just because they are flippant (though more likely the result of ill-conceived sense of humor) and possibly overbearing [only one other person besides yourself- who I will decline to name- that appears to think this (though that makes you neither wrong or right)], I will simply attempt to answer your question. Then again, maybe I won't (too lazy). Maybe I'll let myself be called flippant, imperious, and self-important. After all, the truth- whatever it may actually be- is for the its seeker to find. Nodice, who accepts what you think of me.

boxfiddler
boxfiddler

how you can be flippant and imperious [i][b]without[/i][/b] being self-important, I may rescind my vote.

The Maverick Phantom Wanderer (formerly Macoza, No
The Maverick Phantom Wanderer (formerly Macoza, No

the qualities of those locusts are qualities that some believe I possess. Unfortunately, I must agree that I am flippant. My flippancy is something that rarely shows itself outside of the online arena, though it appears frequently to those of my own generation (with good reason). As for being imperious, that is something will have to be dealt with soon. However, I must protest being called self-important. Maybe some of the things I have said have led people to believe this of me. If so, this is probably a result of several things that I may not mention here ( In general any perceived self-important is nothing more than a reaction to real life circumstances). Fiddler of boxes, rescind thine vote. Nodice, who leaves to cry for the first time in 18 years.

CharlieSpencer
CharlieSpencer

"Nodice (formerly KP, Anonymous User's Champion)" How about the plague of Cyber Identity Indecisiveness?

boxfiddler
boxfiddler

of the imperious, flippant, and self-important variety that would likely be a good vote. etu

Editor's Picks