Originally posted on my blog here: Original Post

I currently use K-Meleon as my default web browser. But, with so many browsers now available for Windows, I wanted to find out which one would be best for my machine.

So, I installed the latest versions of Firefox, Seamonkey, Opera, Safari, Google Chrome, Lunascape, Konqueror and Flock. I used the default configuration for each browser, with no extra add-ons or plug-ins.

Out of curiosity, I looked into running Galeon and Firefox under Cygwin, but it turns out that Mozilla and Cygwin do not get along.

I used the normal configuration for my computer, without disabling any anti-virus software or any other daemons I generally have running. The only other application I had running other than the browser was Notepad (for jotting down notes).

Before running the tests, I ran my computer through an anti-virus scan and a maintenance scan to clean up any registry errors or junk temporary files that might slow things down.

I measured each browser using two different benchmarks.

First, I ran them through the free online benchmark service at http://service.futuremark.com/peacekeeper/.

Then, I launched each browser through a little program called AppTimer to measure how long they took to launch. I launched each browser six times. I did not count the first result, as it was always longer because the browsers need to set default settings the first time they are run. The chart below lists the average of the last five launch times for each browser.

First, here's some basic stats about my computer:

Processor: AMD Athlon XP 2600+ 1.91 GHz
RAM: 2GB
Video Card: ATI Radeon X1650
OS: Windows XP

Next, the benchmark scores for each browser (higher numbers are better):



Finally, how long each one takes to launch, in seconds (lower numbers are better):



Lunascape won the benchmark test, and launched pretty quickly, but I thought the default options were pretty ugly. After exploring the browser's options for a bit, I found it's highly customizable and now I've got it just the way I like it.

I was very surprised by K-Meleon's poor results. Any chance that I ran the benchmark software incorrectly? Do these numbers look generally correct to everybody?

Quote

I was very surprised by K-Meleon's poor results. Any chance that I ran the benchmark software incorrectly? Do these numbers look generally correct to everybody?

Mozilla 1.8.1.19

@ roystonlodge

1.)
Such tests have a single shortcoming.
They are little to no informative how fast on the average, a browser during usual surfing indeed is .
Keep in mind that this is only a JavaScript test.

You can make a much more convincing and reliably test by your one by loading sites you frequently visit with different browsers and by measuring the time each browser needs to completely render the page.

2.)
Depending on the complexity of the settings each browser offers for tweaking there will be a very huge or little performance difference according to your settings.
E.g you can speed up Opera about 30% modifying/desabling default settings.
Can as well apply for startup time.
Therefore testing with default settings doesn't neccessary reveal the abilities of the browser

3.)
You can make tests without disabling any anti-virus software assumed you have perfect knowledge about how your AV works and configuring it accordingly.
Supposing you have a browser with runtime packed components and another one without runtime packed components, an AV with a capable unpacking engine will decompress first the components in order to analyze the files which will take longer compared with components which are not compressed.

As a conclusion IMO such benchmarks are firstly (and accordingly) made for (subtly) marketting



Edited 3 time(s). Last edit at 03/30/2009 06:51PM by Yogi.

yeah.. just like nico has mentioned, these benchmarks are more or less for engines comparison. in a previous benchmark site posted by nico.. i tested both km1.5.2 and km 1.6 with 1.9.1 against firefox 3.0.1beta2(gecko 1.9.1) and firefox 2.0(gecko 1.8).

respectively both kmeleon versions exceeded their firefox counterparts by aprox 200 points. so in the end what makes km appear slower than those other browsers is the arguably 'older' engine. the benchmarks mostly test js ability which is indeed slower in gecko 1.8 than other engines using very fast js renderers by comparison.
i'm not sure though about the launch time tests because i know on my pc, on warm km launches faster than lunascape and at par with chrome.. on cold, chrome is quicker but the difference is quite small.

there's a very important issue that all benchmarks ignore or incapable of doing.. even those reviews done by techies always neglect; it's the endurance of the browser over long periods of usage especially when having many tabs open or heavy pages. this is the real test that no one dares to mention on benchmarking sites.. the average user probably spends at least half an hour per session with maybe 4 or 5 tabs at the same time.. a heavy surfing user will have an extended usage of maybe 2-3 hours per session.. i've done those tests with all the current known browsers(except for chrome) and only 2 browsers come out walking, even ie which is a system embed; starts acting up after long sessions.. the memory usage starts growing rapidly, the browser eventually becomes unstable and requires a restart and in some severe cases, only a system reboot or memory compressing will restore everything to norm. this is where kmeleon really shines.. so far, i can not find any browser that can handle extended usage with all sorts of crappy websites like kmeleon and opera do. those 2 are very well behaved and very well coded applications that know how to stay standing in the end.. they have a good relation with the system and endure whatever you toss at them. that's why, no matter how fast a benchmark sez this browser is and how good that browser scores.. the everyday real browsing is much more than that and that's why i think kmeleon and opera are the best browsers for the windows platform.

More about tests
http://img3.imageshack.us/img3/8648/acid2.png
http://img207.imageshack.us/img207/8122/acid3.png
I know that now only the few browsers are able to pass the acid3 test, but I was surprised when K-Meleon (actually Gecko which K-Meleon uses) haven't passed the acid2 - when IE 8 was released someone said "finally, it passes the acid2", which means that almost all the browsers except IE were able to pass it long ago

acid tests have little significance in actual browsing.. 99.9% of websites are not 100% wwc compliant. in the completely messed-up alphabet coding soup the web is made of, those acid tests are pointless.. i prefer a browser that never crashes on me than passing those acid tests.

so kmeleon doesn't pass the acid tests but i know for sure kmeleon never took any drugs

Hi everybody:

Your benchmark could be interesting, but my experience indicates me another thing:

THOSE VALUES COULD BE INTERESTING, BUT PRACTICE IS WHAT GETS U BEST INFORMATION.... and practice indicates nowadays that K-meleon is the fastest browser for me.... Pages load faster and it´s stable......


USE K-MELEON CCF ME by Hao Jing

dj_creator4

http://pp4mnk.zxq.net

Quote
disrupted
i prefer a browser that never crashes

Hmmm... Lately K-Meleon has begun to crash on exit (1.5.1 and 1.5.2):
http://img13.imageshack.us/img13/117/kmelcrash.png
but I think lately updated Adblock Plus 1.0.1 do it (because I haven't done any other significant changes)

Quote
roystonlodge
I currently use K-Meleon as my default web browser.

But, with so many browsers now available for Windows, I wanted to find out which one would be best for my machine.

Do these numbers look generally correct to everybody?

The same here.

Yes, IMHO You were correctly measuring JavaScript performance of the compared browsers on Your machine. Two remarks: IE firers up fast because it is preloaded. & IMHO Lunsacape.exe starts first and then loads an engine. Here K-Meleon 1.6. with the same engine loads the first page faster. I test with the same local page.


K-Meleon 1.5.x is one of the fastest of the last generation of browsers and goes with lean resources. How else it was done with the best compiler and rendering engine of that time. The browsing is still fast compared with other more modern browsers but with the advent of new JavaScript engines K-Meleon 1.5.x lags behind with its Script performance.

Right now a competition for faster Script engine going on.
It started with Safari 3 & chrome 1...

Till about end January 2009 M-Meleon 1.6 beta with GRE 1.9.1 (scavenged from Lunascape 5 beta ) was prime for all my machines.

This changed in February 2009 when Chrome 2 and Iron 2 (chrome browser without data collectors) betas came out with the newest version of V8 Script engine.

In March 2009 steve posted a link to his browser performance tests into Development forum. He has similar results as You, and IMHO includes some newer versions.

He did not notice when I had posted & He did not use exactly the same GREs for his comparison of FFox with K-Meleon. I put k-m 1.6. components into the same engine e.g. for comparison with Lunascape or future FFoxes. Guess steve did not know which and how.


Apart from the new Script engines the speed of the new generation browser's comes from a newer faster MSVC8 compiler. The new browsers are specifically build for new CPU generations. Experimental K-Meleon builds with that are available and can be used e.g. with Lunascape's Gecko engine.

These K-Meleons are several time faster with JavaScript than 1.5.2 (provided You have a new machine/CPU with ample resources.



Edited 2 time(s). Last edit at 03/31/2009 08:03PM by guenter.

airspirt, this crash doesn't appear to be adblockplus related.. but if you haven't done any significant changes, then try to reinstall adblockplus again.. make sure to overwrite adblock service files in components, the chrome file and the kplugin file.

i'm using abp 1.0.1 without problems.