Firefox 3 vs Opera 9.50: Memory usage

What are we testing? We're testing the memory usage. The two serious contenders are Firefox 3 Beta 5 and Opera 9.50 Beta 2. Both aren't finished yet, but they've both reached a mature level in their development to compare the memory usage and give an indication of what the final release will be roughly.

In this test we look at the private memory usage of the memory alone, thus not looking at shared memory.

Hardware
- Intel Core 2 Q6600 @ 2.4 GHz
- 4 GB, DDR2-800 (3.2 GB addressable)
- 2x 500 GB Western Digital hard disks, RAID-0
- NVIDIA GeForce 8800 GTX, 640 MB

Software
- Microsoft Windows Vista Home Premium SP-1, 32-bit
- Firefox 3 Beta 5 (2008032620)
- Opera 9.50 Beta 2 (9945)

Preconditions
- Both web browsers are configured with their default memory settings, cleared caches/cookies/history.
- Each browser is cleared and restarted after each benchmark run.
- All pages are loaded completely.
- Firefox 3 has no additional add-ons enabled.
- Opera 9.50 has Opera Link enabled.

Daily sites
Dell, Aeon, AniDB.net, Fantasy Real, Formula 1, ING Card, InterfaceLIFT, Kotaku, MotoGP, Postbank, Tokyotosho, Wikipedia, AppleInsider, Ars Technica, BetaNews, Blue's News, CNET, Gamer.nl, Mac Rumors, Neowin.net, One More Thing, OSNews.com, Slashdot, Tweakers.net and Voodoo Extreme.

Single results (one tab)
- Firefox 50.888 K
- Opera 169.928 K

Parallel results (multi-tabs opened):
- Firefox 118.896 K
- Opera 213.668 K

Parallel results (multi-tabs viewed):
- Firefox 177.396 K
- Opera 261.020 K

Plug-in heavy sites
Indiana Jones, Wakfu, Bungie.net, Xbox.com, PlayStation, YouTube, World of Warcraft, and Guild Wars.

Single results (one tab):
- Firefox 90.976 K
- Opera 164.220 K

Parallel results (multi-tabs viewed):
- Firefox 344.488 K
- Opera 391.808 K

Conclusion
The results are surprising. The new king of reduced memory usage is Firefox 3, and not the browser you'd expected it to be, Opera. The Mozilla developers have truly listened to their community and put a hold to the often shouted weakness of Firefox 2, the memory usage (sometimes expressed as leakage by some).

You could argue that you can use the amount of memory that is freely available, but it's no excuse to claim unnecessary memory, as Opera does. Firefox 3's improvements in garbage collection and reducing memory fragmentation have proven that theory works in the real world as well.

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

Good job. I love benchmarks :-)

how do you know which memory is necessary and which isn't? you should have monitored the amount of disk accesses also, maybe opera uses less disk than firefox, who knows? if i have 4 gigs of ram, i'll gladly give programs all they need, ram is faster than disk, and should be used whenever possible...

And about response time of each?
Firefox becomes very slow when has several tabs open, that is fact.
And the forward and back pages in Opera is almost instantaneous, regardless of the number of pages that were previously accessed, while in Firefox this task is not instantaneous.

I wonder how many KB would Firefox use with 5 or 10 extensions.. :-P

Third commentor is an Opera user and still thinks this is Firefox 2. Wake up and smell the coffee, Firefox 3 never slows down even after opening and closing multiple tabs. I know, I ran the nightlies from Dec 2006, the day the Acid2 patches landed until now, I know how much Firefox 3 is a huge improvement over Firefox 2 which should not have been released at all.

Previously Kestrel was fastest, but now it's been eclipsed by Firefox and Webkit. Opera still has poor compatability with webpages as seen from the comments on the Opera Desktop Team blog about Kestrel builds including Beta 2.

Hey, I am using Ubuntu 8.04 in a 1.2Ghz Intel Centrino with 256MB RAM and Firefox 3 beta 5 is using more memory than Opera 9.50 beta 2. Firefox is definitely more sophisticated and everything.. nevertheless is slower.

http://blog.pavlov.net/2008/03/11/firefox-3-memory-usage/

Firefox 3 doesn't use much more memory even with a lot of add-ons because of the new memory management features. Opera users can believe it or not, Opera is no longer as lightweight as they used to think. Besides, anyone seriously use the crappy bittorrent feature in Opera over a dedicated one like utorrent? Bloating up the browser with useless feature and still can't compete with a dedicated, low resource, fully featured utorrent client is embarassing to say the least.

If I said it, is because I use Firefox too, and I can see that differ, if you can't see it, so sorry.

I confirm those tests, the lastest beta opera isn't even near of firefox WRT memory usage, and it's fast like a fast fox

As it stands, Opera's current release uses significantly less RAM compared to Firefox 3 Beta 5, and obviously FF2. I'm going to withold judgment until the final releases, when it really counts.

Firefox 3 Beta 5 with additional add-ons to try to get it remotely close to what Opera 9.50 Beta 2 is out-of-the-box:

Opera opens pages faster, period.
Opera responsiveness to tab-clicking and the like is better, period.
Opera synthesis of many Internet functions (searches, mail, chatting, newsfeed reading, etc.) is more polished and cohesive.
Firefox just seems much more cumbersome to use than Opera for most things.
Comparing Opera to Firefox without add-ons is an unfair comparison to me because Firefox with add-ons is pretty much useless (very limited in its capabilities) in comparison to Opera.

I'm not saying there isn't room for Opera to grow/improve, but that's the way I see it here without any numerical statistics to go on.

Firefox with 5+ tabs and 5+ add ons becomes very slow. Although the new version is comparitively faster, it is no where near the required speeds to satisfy users. Abhishek www.dibugs.com

By default Opera comes with turbo mode enabled. That means it will keep more stuff in memory for faster performance. I would like to know how Opera 9.5b2 behaves with turbo mode disabled.

Hey, what were the settings for RAM cache in Opera and in Firefox? Opera set on "auto" eats up to 15%(IIRC) of available physical RAM for caching pages. Firefox on the other hand caches only a few recent pages(also IIRC).

If the information about this setting is not included then the test is not valid in my opinion.

"You could argue that you can use the amount of memory that is freely available, but it's no excuse to claim unnecessary memory, as Opera does. "

No offense but I would say that doing a memory benchmark implies that the author knows something about memory handling and all that stuff. The whole test is basically worthless and the statement above proves that you are not very familiar with how these things work.

If the statement above was true, Unix and Linux OSs are basically crap when dealing with available memory:both OSs cache almost everything, keeping the cache means that if something needs the same data again, there's a good chance it will still be in the cache in memory thus improving performance... and yes... RAM is wasted if it isn't used, the question is whether the OS (or the application in this case) does a good job managing it, cached memory can be replaced quickly if a running (or newly starting) program needs the memory. Not using cache memory is as bad programming as badly use it.By your logic Unix and Linux systems "claim unnecessary memory", go figure.

That's why your test is worthless and shows nothing about memory handling, run a real test with multiple apps, heavy usage, hours of surfing, compare what features each application brings to the table (Opera's performance is dramatically improved by using cache, Firefox is basically a bare bones browser by default, etc) and see which browser is more responsible after all the tests, memory usage and performance is all about trade offs and releasing memory when other applications claim it...

Ok, I'm blind :) Too early in the morning my timezone.

Opera apparently was set to auto. I won't say that the Firefox team didn't do a great job, because they did. They did a wonderfull job.

But still, keep in mind that setting Opera's ram cache on auto with a high amount available RAM may give you easily a half gigabyte memory footprint over the day.

Cheers.

As others have mentioned, without specifying what each browser is doing in terms of cacheing this comparison is unclear, and "unnecessary memory" is a meaningless term.

Big kudos to Mozilla developers, I think FF3 is a great browser, it has shed a lot of the sluggishness of FF2 (wow, their JS optimizations are amazing!!!) and the work on memory reduction is noticeable. They've done really great work going from FF2. Oh, I love the new UI features like Places — really brilliant stuff (wish it had made it into FF2)!

But if Firefox is also deliberately flushing data out of memory (IIRC is does so more aggresively than FF2) and Opera is cacheing it, then both browsers are simply using different priorities.

On my older tests with Opera 9 vs. FF 2 (open 3 Google image searches, 20 results per page, navigate forward 10 pages in each, then navigate back) — Opera retained ALL images in RAM for faster history navigation using less memory than FF 2 did. Now it seems Opera is still faster for history navigation though uses more memory to do so. That seems a fair choice (and personally I prefer the history navigation speed over seeing large free RAM numbers in Process Explorer). Note that Opera's memory cache scales to the system memory, so the more memory, the more the cache will utilise. Remember that Opera's core runs on mobile phones, so it can work in utterly spartan conditions. But as the RAM restrictions lift, it will scale to the machine.

You may also want to read this excellent article by Mark Russinovich of Sysinternals fame about memory optimization, just to re-emphasise that constantly flushing things out of memory is suboptimal performance-wise:

http://windowsitpro.com/article/articleid/41095/the-memory-optimization-... — click on reprints to get a free reprint for personal use.

Actually, FF2 was really keeping all images uncompressed in memory, regardless of tab, fastback cache or the actual amount of physical ram available. That was IMO the biggest source of memory consumption of FF (esp. since sometimes, FF held onto images even if there was no obvious reason to).

Opera did (and probably still does) this much smarter - it kept the actual image files in memory, but flushed the decompressed bitmaps (probably with respect to the memory cache size), and it decompressed the images on demand without disc access. This won't be noticable if you go back/forward and esp. not on google image search, which shows small thumbnails, but on my computer there was a noticable jerk whenever I scrolled a page with screen-sized images and a new came in view.

FF3 is much more aggressive than FF2 - it flushes all image bitmaps after 45seconds, regardless of the amount of ram still available. I'm not much of a fan of this either, I'd prefer if the images were kept in memory unless the memcache was full, but it's better this way, and at least FF now actually has the ability to flush image data that's still used, instead of having to keep everything in memory.

Thanks for the informative comments (do you have the bug number?) :beer: I do remember seeing the bug report for this but had forgotten its details.

Sure, the one where the 45s flushing was implemented is #296818. There's still another bug realted to this - #213391 because the images are only flushed after 45s, so it's still possible to bring FF to it's knees by quickly loading a set of large bitmaps (before the 45s kick in). The attached testcase is artificial (it uses very small files so they get loaded quickly), but it nicely demonstrates that FF really does keep all images in memory (but I haven't tried it on a late nightly).

add some addons to firefox (so it will be near to opera) open 10 pages and you see how much faster is Opera. Scrolling is just terrible in firefox.

Sorry FF people, its a fact.

Also Opera has build in more features then any other browser.

Also Opera beats ALL browsers in ACID2 and ACID3 tests.

Only thing i love more about firefox is automatic speller, opera has just a speller and u need to install it... also some addons are great. Like adblock plus and element helper cleans up webpages better then others.

Also Opera beats ALL browsers in ACID2 and ACID3 tests.
Actually, Opera is on par with WebKit (the engine behind Safari). Sure, WebKit's implementation of SMIL still sucks, but enough is implemented for it to pass ACID3.

In broad terms, I want to know:
- how accurate is it (error-free, standards-compliant, faithful rendering)
- how fast is it (rendering speed, startup speed)
- how light is it (efficiency of resource utilization)
- how integrable and extensible is it (openness, OS/user component architectures)

1. While Gecko has been fine-tuned for ages, it has become bloated and inordinately complex. It was originally designed for Windows. It needs a greenfield redevelopment. Everything else is faster -- Gecko has become the Giant Sloth of web browsers.

2. Kestrel (or whatever Opera is calling it these days) is not open source. It's a great browser that's blazing fast, but these days a browser's architecture should be that of pluggable internet middleware, not a black box. If you don't want integration, it's fine. But integration, shared components, etc., have value. For example -- virtual filesystem navigation (an "OS" function) is very close to what the web browser does; why have redundant components with separate configuration? When a user or application interacts with a remote application via ftp, ssh, dav, or xml, why can't that happen via shared components and a single configuration?

3. WebKit is the way to go. It grew out of KDE's KHTML, which is lean and fast. Safari uses it. Gnome is switching to it (from Gecko). KDE is still using KHTML, but would likely change to WebKit if there is critical mass. (Actually, the "way to go" is for Opera to open source their engines.)

Also, your "benchmarking" is somewhat unscientific (i.e. without value). As noted above, Firefox sans plugins != Opera. Private memory usage is but one small criteria one might consider when choosing a browser. Also, it would be useful to see this on a variety of OS and hardware platforms and with a broader range of more performance-oriented tasks.

You could argue that you can use the amount of memory that is freely available, but it's no excuse to claim unnecessary memory, as Opera does.

That's a pretty lame counter-argument you've got there.

Overall I would say this is a pretty useless benchmark these days now that everyone has shitloads of RAM to spare. It's 2008 man, not 1995, it doesnt matter anymore.

wrong there, you having shitloads of ram, but if every developer asumes that then you are just like at the begnining eating MORE ram than you need to, why do you think most people ahve shitloads of RAM? it is not because they just wanted to have enough, it is because most software now eat SHITLOADS of ram, so we are just stuck at the begining, thats how business works, now you have to have at LEAST 2GB for you to say you have enough, and about 4GB for you to say you have 'shitloads', nowadays most configurations have only 1gb or 512mb even, thats too far from 'shitloads' considering todays memory hungry bloated apps.

Useless
This term best describes this article. As I am very interested in software, I have tested those browsers and more (IE8 and Safari 3.1 also) my conclusion was that Opera was faster and took less memory. Will I say : "Guys Firefox is heavy, it's shit" ? No, because it really depends on your settings and your computer. So just stop with those stupid benchmark and try it for yourself !!! It's not like you're paying those ...

Remember One is always best served by his own self !

lets stop hating, wait for ie8!

gosh... i hate ie! and ie8 sucks!!! firefox is so much better, opera is fine but not as good as firefox

Our company relies on getting web developments working in all browsers. Our experience shows that the best route is to code for Opera and then add the exceptions for FF2/3 Safari IE6/7 etc. We found that coding for Opera leaves us with the fewest conditional markup requirements for the other browsers. Coding for FF 2 left us with the most.(IE6 just made us dizzy) FF 3 in our test harness has shown no change in this area.

Recently we dropped IE6 compatibility - a lot of people may have it, but why support a dead and clunky browser.

Don't get me started on Safari of any flavour - Safari may be fast (ish) but being able to crash faster and on more websites than any other browser is not a reasonable goal. And yes we do use macs for development but business pragmatism wins out over zealotry.

Firefox is (2.x) and remains (3.x) a largely poor web browser. Better than both IE6 and Safari (any flavour) but not as good as Opera 9.27 and IE7 against FF2.x. FF3 is still too clunky. We looked at the code for 3 - they have taken steps to reduce memory usage that will bring it to a crawl on 'real' sites. Further, the FF answer to their inabilty to be standards compliant is to say 'everyone else is wrong and they are right' (in public), but a quick trawl through the bug reports shows this is all just hype and hyperbole.

Opera 9.5 beta still has bugs that prevent us from using it as a baseline, but we expect that to change.

For most, choosing a browser is like choosing a car. Personal needs and brand bias are often more important than the reality of ownership. People who hate IE will always hate IE - even though IE7 is actually not that bad now - people who love Firefox will always love Firefox - even if it has become broken and insecure to an extent even IE6 is unable to repeat.

As for us, we need a good development tool not a lifestyle choice. Opera is the tools for now. If FireFox version 3.4778987366763 IE8 Konquerer or some such is that in the future then we shall change. For now Opera is our preferred solution.

Have 15 tabs opened in the latest Firefox 3 nightly build and it isn't slow at all for me. It scrolls perfectly on every page. This is on a AMD Ahtlon 1.60 GHz based notebook with 1.25 GB of memory and Vista Home Premium. Maybe some of the people saying how slow Firefox 3 is with many tabs opened are on slower computers.

I've been using Opera for a long time, that is to say before it was ever free. And I must say that no other browser even comes close. I just install it, select the faux emerald skin and browse away. The tabs were in Opera far before any other browser. And their implementation is perfect. Everything is just INSTANT. Let's not forget about the forward and back funtions, directly from the mouse, a feature that I have come to take for granted, and when I try that in IE or FF, I get so mad because they just DON'T HAVE THIS FEATURE. Or at least preinstalled, I guess. I did use firefox too. I have a friend which is a ff dude and he wanted to convince me to switch. But after retesting ff, I came t the conclusion that.. well, Opera is just better. I reccomend it to anybody and everybody.

My job requires the computers/browsers to be always-on or stand-by. Some computer may go through hibernation, but none may be powered off. We have to open/close 10-12 pages (browser, tab, page, instant, what-so-ever) at a time. Each page for one web site.

The IE-7 machines can withstand this sleepless operation no more than a week. Much less if open/close pages after pages from multimedia-enriched sites frequently. We have to shut down those machines every Friday, else we'll have trouble on every Monday.

However, the Opera 9.2x machines can run for months. Three machines are powered on for more than four months. We never have to shut down Opera even once (to claim memory, or else).

All machines are from the same lots. Each have at least 1GB (1024 MB) RAM. So .. if Opera eat 300-400 KB (0.3-0.4 MB) RAM for lunch and then throw away every days, but never crash even once, nor slow down, WHO CARE?

If Opera is such that divine, why do we have to use IE? Darn! That 1 M$-way of IIS! Opera can't work with several websites that rely heavily on M$_STANDARD Web Server Functionalities.

Heh, as I said .. my memory is pretty bad. Seem I forget to talk about some browser that .. errr .. errr .. what's its name?

But I still remember that .. THIS YEAR IS
2008, not 1995.

I work for Opera and I like Opera and FF pretty equally. Out of loyalty to Opera, I use it on a daily basis, but before I was hired I was a huge FF fan.

I like Opera for general cleanliness and speed. It is a nice minimal browser with plenty of built-in features. For general browsing and web development I now prefer it over FF. It's just CLEAN.

The only reason I might use FF over Opera SOMETIMES is the extensions. It adds a lot of functionality that Opera may not have.

So yeah, I'm kinda torn. With 9.5 and Opera Link though, it makes it nice when I need to synchronize my bookmarks and speed dial between my desktop, my work laptop, and my phone, which runs Opera Mini (by FAR, the best free mobile browser). So yeah, I know I'm biased, but I love Opera. I still love FF too though, and Opera works a lot with FF to open web standards.

As long as people avoid IE, I'm happy. I think we can all agree on that :)

I have used firefox for some years now, and tried this browser out trough this day. And guess what? I am really impressed!!! Looks good, acts great and is full of feaures out of the box. As for now, Opera 9.5 will be my new browser if nothing crazy happens! :)
I can honestly recommend this browser!

Memory usage comparison is only fair when both apps are doing exactly the same thing. Opera 9.5 is offering and processing more features (apart from being more than a browser, it's a mail reader, feedsreader, etc) than an extensionless FF3, therefore it's hard to compare memory usage.

Opera doesn't consume even 100mb when I use it with multiple tabs opened.
This test is not valid!

Right now my Opera 9.5 is using 49,46MB of RAM + 55,816MB of virtual memory.
TOTAL= 105,276MB

The Opera is running since 37 hours with about ~3200 page hits in history.
(usually i use 3 tabs).
You just need to go preferences and choose smaller cache memory size (i use 4MB) to
get similar results.

The machine have 256MB of RAM and 512MB of virtual memory.

I just tested firefox 3.0 final and opera 9.5...

In my tests opera is much better.. In some cases it takes only half of the memory that firefox eats..

And btw.. where is this ffox brilliant memory management?
Looks like ffox is very lazy in freeing the unused memory..

Pretty weird results. I was expecting firefox to take lead in memory management.

I am using opera and firefox for about 4 years. I like opera than FF for the several reason:
1. the initiator of download management
2. the initiator of tab browsing
3. the google translator widget is the best i've ever used

A new test should be conducted between the recent releases of Opera 9.5 and Firefox 3, seeing as they have no beta-status anymore.

Please see http://avencius.nl/content/firefox-3-vs-opera-95-memory-usage-take-2 for a comparison of the final releases, as well as explanation between memory usages as well as supposed tweaks for Opera.

Firefox 3 requires more memory than Opera
This test is a fake

hmm, i've had over 50 tabs open in ff2 (just under 200kB per winxp taskmanager), at which point it can begin having problems.
usage w/ opera9.nn has maxed below 90kB, IIRC.

i usually have only 3-7 explorer(to 35kB), 2-7 tabs in text editor (to 14kB), and 1-2 agent ransack(~11kB each) running. i run services necessary to see the router (net & LAN). same kind of
firewall at ~20kB, sum of remaining antibaddies 20-30kB.

Still I'd been avoiding going to ff3 from ff2 precisely to avoid heavier ram usage. I'm hoping to find more solid comparison, that accounts for various manner of use. (eg, i also use googlebarlite, userchrome, usercontent, and some profiles have 5-20 searchplugins. i rarely view swf. In Opera, I've added a few search engines + a few buttons, else it's stock "classic installer" opera w/ the most restrained skin possible (skins were broken a while in 9.5.)
Then there's kmeleon, which seems to use less than Opera, but crashes unpredictably. i've never had Ie7 on this computer cuz i've seen what a slug it is on others. presumably IE7 is actually IE8 beta.

~800MHx duron, 640 pc133 (sdram). gf3 ti200 (64mb?). winxp

thx for writing this up.

oops, "same kind of" was junk that should have been deleted during editing.

Nice article, thank you. To my mind it is necessary to test the memory usage of the two serious contenders are Firefox 3 Beta 5 and Opera 9.50 Beta 2 because they have both reached a mature level in their development to compare the memory usage.

What about when you have to load Firefox with FLST or LastTab, Tab Scope, & Tabs Open Relative, to get the same features for tabs as Opera has built in, also opera is a full email client. Lets run Thunderbird too and see who eats more memory.
But also Opera needs Aspell & a dictionary just to spell check.

This page came up when I was looking for a reason why Firefox needs to be restarted to change the skin, or add or disable a plugin, where as Opera doesn't.

WAS an Opera user, now a Firefox user. Opera even with 9.52 out still has issues with some sites I go to.

With turbo mode disabled (type about:config in the address bar), Opera uses much less memory and does not seem much slower.

Firefox 3.03 and Opera 9.61 freshly installed on my computer.

At the moment Opera with 9 tabs is using 101.6 MB of memory
while Firefox with 5 tabs opened is 103.2 MB of memory.

I have no plugins or extensions installed in Firefox.

I rest my case.

firefox 3 is rubbish i'm using opera 9.62 and this uses less memory than firefox 3.0.4 when using single tab, how the mighty have fallen!

I have an 800MHz, 256MB RAM. The FF3 is significantly slower due to large amount pagefile access. FF2 takes a while until that happens. And Opera takes even longer. I'd love it if someone would come up with a lean, fast, highly optimised for older machines.

Navigation

Chrome

Chrome tracker

Firefox

Firefox tracker

Opera tracker

User login