Firefox 3 May Be More Memory Efficient Than Either IE or Opera

Edy52285 writes "Ars Technica has an article showing benchmarks pitting Firefox 3 Beta 4 against other browsers. Contenders include IE7, Firefox 2, Opera 9.5 Beta, and Safari 3.0.4 Beta. The piece includes a graph depicting FF3's memory usage well below that of the other browsers. The in-testing browser even trumps Opera, which has long been regarded as the fastest browser around."

Comments

Scale?

[frp001]

I just love this when someone provides a graph without even a detailed scale!

A Blessing!

[Ngarrang]

Firefox 3 will certainly be a blessing to my company if this holds up through official release. My company is standardized on Firefox for all web browsing and intranet apps. Our PCs are not necessarily cutting edge technology filled with copious amounts of RAM. The average speed is 1GHz and 512Mb RAM running XP. Now if only all apps took the route of less/improved memory usage with each new version instead of the bloat I am suffering with Microsoft Word, Citrix, etc.

Re:A Blessing!

[twistedsymphony]

This is one of the greatest selling points to open source software IMO.

When you have a product like MS Office, every year that they release a new version they have to load it up with new features to encourage people to buy it, despite the fact that most users only use a fraction of the feature set and rarely need any of the new features the new version offers. This can be applied to most for profit software.

When you have a product like Open Office it's being developed by people who are working more for their affinity for the software rather than a paycheck. The result here is that unneeded features are left out of the core application and once there is a solid interface and feature set they start turning towards making the product more stable and more efficient.

Of course there are exceptions on both sides of the fence, but this is something I've noticed with most of the OSS that I use. By running nearly all OSS alternatives I'm able to use the latest versions of my most common apps on my old P3 733 laptop and it feels just as peppy as the high performance rig I use at work loaded with MS apps.

Re:A Blessing!

[Anonymous Coward]

1) People may only use 20% of an application's functionality, but different people use a different 20%. So a competing application needs to implement at least 80% of the features to even get a look in.

2) "Open Office" ... "a solid interface and feature set they start turning towards making the product more stable and more efficient". Open Office is nice, and it is free, but it's not a great overall example of a wonderful application :)

Re:A Blessing!

[jlarocco]

You do realize that Opera works wonderfully on PCs with specs even lower than that, right? Guess it doesn't help you much now, but you should be kicking yourself for the past.

Re:A Blessing!

[WNight]

Nonsense. The "huge" cost of finding replacement software is $0. If it exists, it's most-likely free.

As for retraining users... Dump them on a website they want (sports scores, gossip, etc) and watch them figure it out. Sit them in front of an office app 90% like the one they're "trained" on and watch them be totally unable to find the File menu. These people can't be retrained because they aren't trying. People who do try rarely need more than a 15 minute 'get up to speed' intro.

Power users who write Office macros, script the app in VB, use all the weird features, they'd take a while to switch over. People who just write the odd document, bold stuff, etc, won't even notice unless you freak them out about it.

This claptrap about retraining is the biggest joke. The only people who can't learn are those who don't care to try. If your employees don't care to try to do what you tell them then they aren't very good employees and should be let go. It's not a question of intelligence, a parrot could use most office apps, but rather a question of motivation.

As for the CIO, some are MS happy, but I've rarely seen one that would care if the 95% of the company that hardly needs a computer were using etch-a-sketches as long as the costs were lower. You are providing some sort of evidence for your lower-cost and less-maintenance claims, right?

That's when testing with their own tool

[Adaptux]

Ok, so the Mozilla folks have succeeded in improving their browser's resource efficiency enough that it beats the competition on their own benchmark.

The more interesting question is of course whether the firebox beta also wins when other benchmarking tools including those produced by competiting browser developers are used.

Re:That's when testing with their own tool

[TripMaster Monkey]

Can you fault the methodology employed in the tool?

Re:That's when testing with their own tool

[edwdig]

Yes. The pages loaded by the tool (mostly different language versions of a number of mediawiki-based sites' home pages) are not representative of a broad enough cross section of people's browsing habits.

I think that's key though. If you start adding Flash heavy sites or whatever, you have to worry about things like plugins mismanaging memory. While that's a bad thing, it's not the fault of the browser, so it's pointless to let that influence the tests.

You could try browsing YouTube without the Flash plugin installed, but that would be an even less representative test.

Re:MAY...?

[Adaptux]

Does that mean that it MAY not be? Because the story itself and the accompanying graph seem to indicate that it IS, not that it MAY be. Just, like, clarify, you know?

This is just the result of one test with one benchmarking tool; on top of that, it was a test with the vendor's own tool. The "MAY" in the article reflects the uncertainty regarding whether this tool and the particular test conducted with it appropriately reflects real-life usage scenarios.

Not real world (for me)...

[alyawn]

I think a much better test would be to see a single window with 30 tabs. I don't know anyone that would have 30 windows of a browser running. All this proves to me is that FF does a better job of sharing resources across instances. Does anyone use windows rather than tabs to manage their browsing?

comes at a cost

[buddyglass]

Can't remember where I read it, but I recently read a description of how they achieved some of this efficiency. Much of it has to do with using a different memory allocator which avoids fragmentation. That's good. However, a lot of it also comes from "expiring" cached data according to some time-based policy. That's probably a good idea too, but it's not a memory savings that can be considered "for free". You're actually expunging cached data from memory, which means you may have to reload it again later, and you're spending CPU cycles to enforce that policy. It probably requires minimal CPU to do that, but if they implement it via polling it could screw up the processor's ability to sleep, which in turn jacks up battery usage on laptops. Witness the recent effort on linux to get various apps to "fix" the way they behave in order to play better on laptops. This could end up being a regression in that area.

Threading

[Thaelon]

But what about threading?

I'm tired of every browser tab and window I have open locking up so Flash can render in one of the windows.

Even IE doesn't do this!

Memory Leak?

[X3J11]

I used to hear from a buddy about how much he disliked Firefox because it was a memory pig, but never saw it myself until a few days ago. I'm not sure of the why or how, but after browsing http://www.deviantart.com/ for an hour or so, opening each deviation in a new tab, my system started crawling. Checking task manager I found Firefox to be using 1.7GB of memory. Closing every tab did nothing to release it, closing Firefox did.

Re:I knew IE7 was bad, but...

[Knuckles]

Along that vein; be advised that notepad has a VERY small footprint.

you seem to be inferring that feature-wise, IE7 is better then FF3. Care to elaborate?

It's not the average speed that matters

[goombah99]

While I know you were making a joke youre point was actually insightful. Namely the browser speed wars is something of a game of leapfrog. Any browser that is reasonably fast is a good browsers. But what matters is that the browser maker keeps the browser among the best at all time.

That is to say if every 3 years browser X gets a big update and becomes the fastest for a few months and then gets severely eclipsed for 2 years. it's not the best browser.

Speaking of Karma hell, a good example of this is Thunderbird email which occasionally shines but then goes and wnaders in the woods for years at a time

Re:It's not the average speed that matters

[GoodbyeBlueSky1]

Well fast is one thing, memory is another. I did notice that FF3b4 "seemed" faster when opening a new tab, but I think that's more of a Gecko performance thing (i.e. browsing the web isn't any faster but drawing window elements is, a bit).

Anyway I always thought Firefox was fast enough. What I'm most surprised with (shocked, even) is the BIG leap in memory management, even from the last beta. Every release gets touted as being better at this, but this is the first time I'm really impressed with the steps they took. I accidentally left work Friday with my workstation still on and Firefox open with multiple tabs. I came in Monday, checked the Task Manager on a lark and FF3b4 was using about 30MB RAM. If you've used Firefox for any significant amount of time (Firebird 0.7, baby!), that number should impress.

Just one man's experience, on Windows (traditionally the best performing version of FF, the Linux version can be pretty buggy and the Mac version actually scares me into using Safari). Let's hope it doesn't regress from here.

Re:It's not the average speed that matters

[whackco]

Ok, just because a program uses LESS memory doesn't make it faster or slower. In fact, the more memory it uses to cache more, the less disk thrashing generally, and the faster experience.

This is also a common misconception in Vista's memory management. It fills the empty space in memory with things 'pre-fetched' for faster loading, etc. I like it, and it works well for me.

Jezz Slashdot - I expected more from the worlds largest concentration of geek power.

Re:plugins

[AKAImBatman]

IE lies. Much of its memory is hidden in system DLLs that don't show up in the Windows Task Manager. To get an accurate read on how much memory IE is using, you need to use special tools that track memory across the system.

opera memory usage

[perlchild]

While reducing wasted memory is a good thing(and memory leaks are worse). Opera's or Firefox's memory usage can include caching resources... Opera's been talked to as the "fastest" browser around, not the lightest... There's a difference, and I'm surprised so few people on Slashdot caught it.

Having less memory leaks makes you faster, but being faster can happen using more memory.

Re:Crash

[o'reor]

I guess, if the Mozilla developer had his cup of jo the morning he came up with that feature, that he decided to store the contents of the deleted tabs in the Mozilla cache, that has a user-defined maximum size on disk or RAM. That would sound logical to me.

Note to submitter : memory-footprint != speed

[mxs]

The in-testing browser even trumps Opera, which has long been regarded as the fastest browser around. And how does a smaller memory-footprint yield faster performance, EXACTLY ? The two are not necessarily related -- or even related in a way you might like. There are plenty of algorithms that sacrifice memory to become more speedy (and vice versa). The size of a memory-footprint of an application tells you exactly ... nothing ... about its speed, or its relative speed to other programs with different footprints.

Quite honestly, I don't care about memory consumption so long as it remains reasonable. My Opera-process has been running for weeks with, at times, heavy usage (dozens of open windows, some with highly dynamic pages). It's been stable and quick throughout that time, and did not grow to a size where I'd have to wonder what the hell is causing swapping.
Yes, you can crash Opera (often related to badly coded plugins), and yes, you can make it unresponsive. I found, however, that it's far easier to do that to Firefox than Opera, and that Opera has been consistently snappier. Maybe that'll change with FF3. Hopefully it will, competition in that arena is always good.