35 posts in this topic

Ok, so I've personally done a memory usage comparison between 33 browsers.

How I Tested :

In order to keep the test environment as constant (and thus fair) as possible, I created a Vitual Machine with VMware Workstation 6, installed the Windows OS, installed all the browsers, set all browsers' start pages blank, restarted the system, opened taskmgr, and saved a snapshot of the Virtual Machine. Everytime I finished test a browser, I loaded the snapshot back to test the next browser, so the enviroment would remain the same before the start of each test. All browsers tested with clean installs, and English interfaces, unless an English interface is not available. All memory values are rounded to 1 digit after the decimal.

Host System :

Intel Xeon 5110 Woodcrest dual-core CPU @ 1.60GHz

2.0G RAM

Windows Server 2003 SP2

Guest System :

2 Processors (Intel Xeon 5110 @ 1.60GHz)

1.0G RAM

Windows XP SP2

The Tests :

1) Blank Page

The memory usage after starting the browser with a blank start page and wait for the memory usage to stabilize, and record the value.

2) One Page

Open http://www.msn.com, IE's default home portal page, wait for the memory usage to stailize, record the value.

3) Six Pages

Open five other pages, wait for the memory usage to stabiize, record the value. The five other pages are :

1. http://www.yahoo.com, one of the biggest portal sites in the world

2. http://www.sohu.com, one of the biggest portal sites in China, and as many Chinese portal sites, (in)famous for heavy animated pictures and flash stuff, and tons of non-standards-compliant stuff.

3. http://en.wikipedia.org, the largest encyclopedia in the world

4. http://www.zol.com.cn, another Chinese site, one of the biggest tech portal sites in China, and again tons of animated pictures, flash, and non-standards-compliant stuff.

5. http://www.apple.com/startpage, the default Safari home page, and quite heavy HTML stuff.

4) Trident Memory Leak Bug Test

Using the memory leak bug test here ( http://im.qq.com/tt/images/release/testpage.html ), provided by Tencent, if the browser's memory usage doesn't imcrease constantly, or the browser can do garbage collections thus periodic decreases in the memory usage during the test, to keep the total memory usage under less than 10M extra as compared to before the test during the entire test, then it counted as not affected by the bug. Else it's counted as affected by the bug.

The Candidates and Results :

===== Trident-based =====

1) Internet Explorer 7

The market leader and most widely used web browser

ie7266.jpg.xs.jpg

Home Page : http://www.microsoft.com/windows/products/...ie/default.mspx

Download size : 14.7M

Blank Page : 17.8M

One Page : 32.4M

Six Pages : 109.1M

Memory Leak Bug : Yes

2) iRider 2.52.0018U

A commercial web browser based on Trident, with interesting thumbnail interface rather than tabs. Claims to have better memory management.

irider119.jpg.xs.jpg

Home Page : http://www.irider.com/

Download size : 3.72M

Blank Page : 11.3M

One Page : 27.3M

Six Pages : 95.7M

Memory Leak Bug : Yes

3) Crazy Browser 3.0.0 RC1

A Trident-based MDI browser

crazy_browser545.jpg.xs.jpg

Home Page : http://www.crazybrowser.com/

Download size : 691K

Blank Page : 14.8M

One Page : 27.6M

Six Pages : 96.5M

Memory Leak Bug : Yes

4) SlimBrowser V4.10

Another Trident-based MDI browser

slimbrowser781.jpg.xs.jpg

Home Page : http://www.flashpeak.com/sbrowser/

Download size : 1.84M

Blank Page : 15.5M

One Page : 29.1M

Six Pages : 98.3M

Memory Leak Bug : Yes

5) Donut RAPT #126

A highly rated Trident shell from Japan.

donutrapt911.jpg.xs.jpg

Home Page : http://rapt21.com/download.php

Download size : 713K

Blank Page : 7.4M

One Page : 27.4M

Six Pages : 96.6M

Memory Leak Bug : Yes

6) Donut L 32

Another Trident shell from the Donut family.

donutl897.jpg.xs.jpg

Home Page : http://rapt21.com/download.php

Download size : 255K

Blank Page : 6.0M

One Page : 26.4M

Six Pages : 95.6M

Memory Leak Bug : Yes

7) Donut Q 77 (Japanese-only)

Yet another Trident shell from the Donut family.

donutq417.jpg.xs.jpg

Home Page : http://rapt21.com/download.php

Download size : 731K

Blank Page : 6.9M

One Page : 26.7M

Six Pages : 95.6M

Memory Leak Bug : Yes

8) Sleipnir 2.7.1 w/Trident

The most popular alternative browser in Japan. With Trident engine.

sleipnir_trident482.jpg.xs.jpg

Home Page : http://www.fenrir.co.jp/en/sleipnir/

Download size : 4.47M

Blank Page : 27.5M

One Page : 41.3M

Six Pages : 116.4M

Memory Leak Bug : Yes

9) Grani 3.1 (Japanese-only)

Another Trident based browser from the maker of Sleipnir.

grani648.jpg.xs.jpg

Home Page : http://www.fenrir.co.jp/grani/

Download size : 4.47M

Blank Page : 27.0M

One Page : 41.6M

Six Pages : 109.6M

Memory Leak Bug : Yes

10) Lunascape 4.6.1 w/Trident

Another popular alternative browser in Japan, with a lot of extra toolbar and scrolling news stuff, and apparently a custom garbage collector that works best with fewer pages. With Trident engine.

lunascape_trident303.jpg.xs.jpg

Home Page : http://en.lunascape.tv/

Download size : 8.10M

Blank Page : 7.2M

One Page : 34.2M

Six Pages : 102.1M

Memory Leak Bug : Yes

11) Bura 1.2.04 (Japanese-only)

A tabbed Trident shell with a funny icon, from Japan.

bura639.jpg.xs.jpg

Home Page : http://www.geocities.jp/tabbrowser/

Download size : 467K

Blank Page : 6.2M

One Page : 28.4M

Six Pages : 94.5M

Memory Leak Bug : Yes

12) Maxthon 2.0.9.1640

The most popular alternative browser in China, based on Trident engine.

maxthon2317.jpg.xs.jpg

Home Page : http://www.maxthon.com/

Download size : 3.8M

Blank Page : 18.3M

One Page : 34.8M

Six Pages : 101.9M

Memory Leak Bug : Yes

13) Avant Browser 11.6 build 11

A popular tabbed browser by a Chinese programmer, based on Trident engine.

avant582.jpg.xs.jpg

Home Page : http://www.avantbrowser.com/

Download size : 1.90M

Blank Page : 19.6M

One Page : 34.4M

Six Pages : 105.2M

Memory Leak Bug : Yes

14) TheWorld Browser 2.1

A light-weight Trident-based browser from China.

theworld253.jpg.xs.jpg

Home Page : http://www.ioage.com/en/index.htm

Download size : 558K

Blank Page : 16.9M

One Page : 31.0M

Six Pages : 98.1M

Memory Leak Bug : Yes

15) GreenBrowser 4.5.0423

Another Trident-based browser from China.

greebrowser827.jpg.xs.jpg

Home Page : http://www.morequick.com/IndexEn.htm

Download size : 1.23M

Blank Page : 13.2M

One Page : 27.6M

Six Pages : 89.4M

Memory Leak Bug : Yes

16) GOSURF v2.81

Yet another Trident-based browser from China. With some custom memory saving technique.

gosurf431.jpg.xs.jpg

Home Page : http://gosurfbrowser.com/?&ln=en

Download size : 2.15M

Blank Page : 4.0M

One Page : 27.3M

Six Pages : 48.2M

Memory Leak Bug : Yes

17) Acoo Browser 1.84.640

Another Trident-based multi-tabbed browser.

acoo_browser258.jpg.xs.jpg

Home Page : http://www.acoobrowser.com/

Download size : 1.71M

Blank Page : 17.1M

One Page : 30.0M

Six Pages : 95.5M

Memory Leak Bug : Yes

18) EnjoyIE 3.0 build 080401

An interesting Trident shell from China that integrated a preview function similar to the popular Coolris Preview. In the Chinese version, the search box also has inline real-time translation function.

enjoyie400.jpg.xs.jpg

with preview in action :

enjoyie1414.jpg.xs.jpg

Home Page : http://www.enjoyie.com/en/

Download size : 737K

Blank Page : 7.4M

One Page : 24.4M

Six Pages : 101.1M

Memory Leak Bug : Yes

19) Tencent Traveler 3.8 (Chinese-only)

A Chinese Trident-based browser from Tencent, the maker of the memory leak bug test, which predictably passes the test. Has some very aggressive memory saving technique and custom garbage collector.

tencent_traveler548.jpg.xs.jpg

Home Page : http://im.qq.com/tt/

Download size : 2.75M

Blank Page : 3.4M

One Page : 10.3M

Six Pages : 22.6M

Memory Leak Bug : No

20) MiniIE V1.5.668 (Chinese-only)

A minimalist Trident-based browser from China, claimed to be the smallest (GUI) browser with the fastest startup time. Has a funny nickname Naked Run Browser. Implements some extremely aggressive custom garbage collector. The only other Trident-based browser that passes the memory leak bug test.

miniie529.jpg.xs.jpg

Home Page : http://www.miniie.com/

Download size : 2.36M

Blank Page : 1.8M

One Page : 2.6M

Six Pages : 69.6M

Memory Leak Bug : No

Share this post


Link to post
Share on other sites

No Firefox or Opera? :s

Share this post


Link to post
Share on other sites
No Firefox or Opera? :s

Note how he mentions having tested 33 browsers, and only 20 appear in the post... guess he hit the post size limit.

To the original poster: Great job, and I can't wait for the remaining test results. Your efforts might settle the eternal 'Which browser sucks the least?' debate.

Share this post


Link to post
Share on other sites

continued...

===== Gecko-based =====

21) Firefox 2.0.0.14

THE Gecko-based browser.

firefox2675.jpg.xs.jpg

Home Page : http://www.mozilla.com/en-US/firefox/

Download size : 5.8M

Blank Page : 21.1M

One Page : 29.9M

Six Pages : 80.9M

Memory Leak Bug : No

Special Notes : Apparently Firefox doesn't like those Chinese sites. The interface temporarily locks up when opening those Chinese sites, even when the CPU has plenty of idle time. Seems to be some problem related to the XUL interface instead of Gecko.

22) Firefox 3 beta 5

The next generation of Firefox browser.

firefox3725.jpg.xs.jpg

Home Page : http://www.mozilla.com/en-US/firefox/all-beta.html

Download size : 7.19M

Blank Page : 30.3M

One Page : 38.5M

Six Pages : 83.5M

Memory Leak Bug : No

Special Notes : Again Firefox doesn't like those Chinese sites. The interface temporarily locks up when opening those Chinese sites, even when the CPU has plenty of idle time. Seems to be some problem related to the XUL interface instead of Gecko.

23) SeaMonkey 1.1.9

THE Gecko-based Internet Suite. Supposed to be a "bloat" compared to Firefox, its Navigator mode surprisingly takes less memory than Firefox, even when they are running on exact the same version of Gecko engine.

seamonkey975.jpg.xs.jpg

Home Page : http://www.seamonkey-project.org/

Download size : 12.4M

Blank Page : 18.7M

One Page : 28.1M

Six Pages : 78.1M

Memory Leak Bug : No

Special Notes : Yet again SeaMonkey doesn't seem to like those Chinese sites. The interface temporarily locks up when opening those Chinese sites, even when the CPU has plenty of idle time. Seems indeed to be some problem related to the XUL interface.

24) Flock 1.1.2

The Social Web Browser, with lots and lots of extra functions, so it bounds to take some more resource as a price. But by how much? we'll see.

flock108.jpg.xs.jpg

Home Page : http://flock.com/

Download size : 10.6M

Blank Page : 36.3M

One Page : 51.9M

Six Pages : 94.4M

Memory Leak Bug : No

Special Notes : Same XUL interface, same story with those Chinese websites.

25) K-Meleon 1.1.5

The minimalist Geck-based Browser.

k-meleon124.jpg.xs.jpg

Home Page : http://kmeleon.sourceforge.net/

Download size : 5.19M

Blank Page : 16.4M

One Page : 25.4M

Six Pages : 72.2M

Memory Leak Bug : No

Special Notes : It doesn't have problem with those Chinese sites, and it's not based on the same XUL interface as Firefox.

26) Sleipnir 2.7.1 w/Gecko

The most popular alternative browser in Japan. With Gecko engine.

sleipnir_gecko787.jpg.xs.jpg

Home Page : http://www.fenrir.co.jp/en/sleipnir/

Download size : ~20M + 4.47M

Blank Page : 30.2M

One Page : 39.9M

Six Pages : 85.9M

Memory Leak Bug : No

27) Lunascape 4.6.1 w/Gecko

Another popular alternative browser in Japan, with a lot of extra toolbar and scrolling news stuff, apparently its custom garbage collector doesn't work with Gecko as well as with Trident. Or it just means Gecko has its own better garbage collector than Trident. With Gecko engine.

lunascape_gecko522.jpg.xs.jpg

Home Page : http://en.lunascape.tv/

Download size : ~20M + 8.10M

Blank Page : 9.9M

One Page : 43.8M

Six Pages : 92.2M

Memory Leak Bug : No

28) Sylera 3.0.20 (Japanese-only)

A Gecko-based browser from Japan.

sylera750.jpg.xs.jpg

Home Page : http://dombla.net/sylera/

Download size : 12.2M

Blank Page : 14.7M

One Page : 25.7M

Six Pages : 75.2M

Memory Leak Bug : No

29) Orca Browser 1.0 RC3

A Gecko-based browser from the same Chinese programmer behind Avant Browser.

orca_browser792.jpg.xs.jpg

Home Page : http://forum.avantbrowser.com/viewtopic.php?f=36&t=13104

Download size : 6.3M

Blank Page : 23.4M

One Page : 38.1M

Six Pages : 83.7M

Memory Leak Bug : No

===== WebKit-based =====

30) Safari 3.31

Apple's web browser, based on KHTML/WebKit, ported to Windows.

safari237.jpg.xs.jpg

Home Page : http://www.apple.com/safari/

Download size : 18.6M

Blank Page : 29.2M

One Page : 47.5M

Six Pages : 128.9M

Memory Leak Bug : No

Special Notes : It seems Safari is not happy with those Chinese sites, after opening the two Chinese sites, its CPU usage jumps to 50% (full CPU usage for one processor) and the whole browser becomes very un-responsive.

31) Safari w/WebKit Nightly r32574

Safari with the latest WebKit nightly

safari237.jpg.xs.jpg

Home Page : http://nightly.webkit.org/

Download size : 18.6M + 14.6M

Blank Page : 30.2M

One Page : 47.1M

Six Pages : 131.7M

Memory Leak Bug : No

Special Notes : Seems WebKit is not happy with those Chinese sites, after opening the two Chinese sites, its CPU usage jumps to 50% (full CPU usage for one processor) and the whole browser becomes very un-responsive.

===== Presto-based =====

32) Opera 9.27

My personal favorite browser :p Merlin engine.

opera254.jpg.xs.jpg

Home Page : http://www.opera.com/

Download size : 4.7M

Blank Page : 15.3M

One Page : 24.9M

Six Pages : 72.9M

Memory Leak Bug : No

33) Opera 9.5 beta 2

The next generation of Opera, with Kestrel engine.

opera254.jpg.xs.jpg

Home Page : http://www.opera.com/

Download size : 4.99M

Blank Page : 20.9M

One Page : 29.1M

Six Pages : 91.4M

Memory Leak Bug : No

=====

Comparison table and charts coming...

Share this post


Link to post
Share on other sites

O.o You sure went out of your way for this one, must have taken ages just sourcing all the browsers.

Share this post


Link to post
Share on other sites

Thats quite a list there :p nice (Y)

still doesn explain why my firefox is slower than IE though, lol. probably the plugins im guessing.

Share this post


Link to post
Share on other sites

Quite good test. (Y)

You should also add another value to see how much memory is used after 10 minutes when th 6 tabs have been open.

Share this post


Link to post
Share on other sites

The comparison table :

browser0561.png.xs.jpg

The comparison chart 1 (Blank Page) :

browser1551.png.xs.jpg

The comparison chart 2 (One Page) :

browser2483.png.xs.jpg

The comparison chart 3 (Six Pages) :

browser3526.png.xs.jpg

Edited by wellofsouls

Share this post


Link to post
Share on other sites

Yap... What a List! Thanks for the info

Share this post


Link to post
Share on other sites

what a great job!

ai have used 15+ of the broswers listed.

I like The World best!

Share this post


Link to post
Share on other sites

Wow. Amazing work!

Share this post


Link to post
Share on other sites

Thumbs up for the efforts. (Y)

Thanks for sharing with us...:)

Go Opera.

Share this post


Link to post
Share on other sites

I think I might check out GOSURF. Alot of these borwsers I have never heard of. Thank you for the list.

Share this post


Link to post
Share on other sites
You should also add another value to see how much memory is used after 10 minutes when th 6 tabs have been open.

Yup, that's a good idea (Y) Maybe I'll do that in a couple of days. It'll take some time though :)

Share this post


Link to post
Share on other sites
Yup, that's a good idea (Y) Maybe I'll do that in a couple of days. It'll take some time though :)

I'll be waiting. :D Maybe time it takes to do a cold start as well.

Share this post


Link to post
Share on other sites

Good job on the extensive and exhaustive test, and I'm sorry to ask this, but, what's the point?

Memory usage is hyped up these days. It has no relevance anymore, unless you're talking about memory leaks or seriously bloated software. Browsers in general are light, compared to office and multimedia suites.

Also, some browsers (eg: Opera) allow you to set the amount of RAM to use, so it's a variable value. What good are numbers, in this case?

Finally, you didn't mention what memory were you measuring? Private bytes? Virtual Size? Working Set?

--

IMHO, It would have been more beneficial, if you instead measured the startup times, page loading, rendering, script execution, GDI/User Handles, etc

Share this post


Link to post
Share on other sites

SO much data and no charts? Find a better way to present this data. MS Excel for example can make charts.

Share this post


Link to post
Share on other sites
SO much data and no charts? Find a better way to present this data. MS Excel for example can make charts.

he has done charts, although he hasn't labeled x-axis so I'm not sure what it represents :o average memory usage?

Share this post


Link to post
Share on other sites
he has done charts, although he hasn't labeled x-axis so I'm not sure what it represents :o average memory usage?

x-axis represents the memory usage from the text.

Share this post


Link to post
Share on other sites

Good job on the extensive and exhaustive test, and I'm sorry to ask this, but, what's the point?

Memory usage is hyped up these days. It has no relevance anymore, unless you're talking about memory leaks or seriously bloated software. Browsers in general are light, compared to office and multimedia suites.

Also, some browsers (eg: Opera) allow you to set the amount of RAM to use, so it's a variable value. What good are numbers, in this case?

Finally, you didn't mention what memory were you measuring? Private bytes? Virtual Size? Working Set?

--

IMHO, It would have been more beneficial, if you instead measured the startup times, page loading, rendering, script execution, GDI/User Handles, etc

well, I'd say the point is about as much as any other synthetic benchmarks out there :) also I did this because personally I have seen less memory usage tests around, when there are many extensive and in-depth page loading, rendering, scripting speed, etc. tests all over the web, done by people with better equipment (I don't even have a stop-watch :p) and across wider platforms (my only access to a Mac is my boss' MacBook Air, and I can't do all I want on it), for example Here.

As for the hype, I'd say memory usage is about as much hyped up as all the other stuffs like page loading, rendering, scripting speed, etc. Only that compared to startup time, page loading, scripting speed, etc. the memory usage is relatively more constant and less affected by some random factors like how your ISP like a certain server at a certain time, or how many cookies are stored in the cache, etc. About the relevancy stuff, when we have browsers that can finish some complex computations on DNA sequences within half a second, stuff like the page loading and scripting speed is really no more relevant than memory usage for real life situations. In reality I think we care more about the features than whether one browser will render FaceBook a couple seconds faster than another, when it takes more than half a minute to download the stuff depending on your ISP :shiftyninja:

For Opera, I use the default automatic setting, so it only shows how it consumes memory when you give it total control of memory management (like all the other browsers). If you personally set the amount of memory to be used youself, it will ofcourse be different. And I just record the Mem Usage in the taskmgr, so I guess it's showing the Working Set. I can add the other numbers if you really want, but then you'll have to wait to when I have some more free time ;)

About the speed tests, I do plan to do a scripting speed test, as that doesn't need a stop watch :p And that my ISP can be quite random on deciding when it'll be more happy to one site than another, so I doubt I can do any valid test related to the page loading speed. However the thing is it seems quite hard to find a scripting speed test that IE and all those Trident browsers don't hang up on it :wacko: well, I guess that just means Trident sucks at running scripts. So far I'm considering to use the DHTML performance tests provided by Opera Here, and I've already got some interesting numbers with the different Trident shells, it seems some Trident shells can have quite better performance than IE.

All in all, the point of this test is definitely not trying to encourage you to choose a browser based solely on its memory usage. However I think some interesting points can be observed from the results, like SeaMonkey actually uses less memory than Firefox, and despite the claim of more efficient memory management, the next generation of Firefox and Opera still can use quite a bit more RAM than their predecessors (ofcourse it could just be because they are still beta's), and some Trident shell's custom memory saving technique and garbage collector can do interesting stuff to the memory usage. I'm sure others can observe more points of interest from the results. After all, that's what synthetic benchmarks are mostly about, IMHO. :)

Share this post


Link to post
Share on other sites

Just wanted to say good work! That's a pretty big list of browsers and the testing must have taken ages.

Share this post


Link to post
Share on other sites

Sorry, but I do not find this the "most comprehensive" nor do I find this to be the "most complete" browser memory usage test. I will give credit that it is the most ambitious browser test I have seen.

Specifically, my issues are related to this:

2) One Page

Open http://www.msn.com, IE's default home portal page, wait for the memory usage to stailize, record the value.

3) Six Pages

Open five other pages, wait for the memory usage to stabiize, record the value.

So for the one page portion, and the six page portion of the tests, all you did was wait for the memory usage to stabilize and record the value? The results are misleading and frankly provide little usefulness in terms of real world usage. What this test should have included was long term memory usage using one page/tab, or using multiple pages/tabs. In the real world, most people are not concerned with memory usage after the usage simply stabilizes. That typically happens in a matter of seconds for most browsers. Under real-world usage, multiple tabs are being constantly opened, closed, and new tabs being opened.

Regarding your claims about Firefox 3, your test does not debunk the claim by Mozilla that Firefox 3 is MUCH more efficient than Firefox 2. Your results for Firefox 2 are also misleading as it's well known that Firefox 2 has *many* memory leak problems, specifically over a long usage period of time. Firefox 3 over a LONG period of time and when running many tabs has been shown by several tests to be more efficient than a vast majority of browsers out there, Firefox 2 included.

If your test was done for several hours with many tabs (12+) being opened, closed and re-opened for each browser, we would have seen vast differences between many of the browsers in your list.

Share this post


Link to post
Share on other sites

Damn bro thanks for all the hard work, great list there and hopefully will get rid of the browser battle lol.

Share this post


Link to post
Share on other sites

Have you considered the virtual memory usage? Tencent Traveller swaps physical memory to virtual memory constantly every few seconds.

Have you configured the ad blocker of each browser to block the same contents? A single flash ad could use more than 10mb memory.

Also how do you test the "Memory Leak Bug"? Where do you find the "Custom Garbage Collector" thingy?

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!


Register a new account

Sign in

Already have an account? Sign in here.


Sign In Now
Sign in to follow this  
Followers 0

  • Recently Browsing   0 members

    No registered users viewing this page.