Web Browser Grand Prix 2: Running The Linux Circuit

[adamovera]

@toyotatabedzrock
I've seen that blog post, about 100 times already. Note the disclaimer on the top that says the contents do not represent Opera Software. I have been (still am) in correspondence with Opera regarding both of these issues, but have no formal statements to relate as of now. Dromeao has since been fixed and Opera 10.60 gets about 212 in Win7, far more realistic than the 2,000 it received before Mozilla applied the fixes (The Linux scores were never that high, and Opera tells me that Mac earned normal scores as well).

@DSpider
We use Ubuntu as our default Linux distro because it is the dominant desktop distro. Compiling from source on a source-based distro would probably be much faster than ANY stock binary distro. But how many people will that article jive with? There is never a lack of article topics in Linux, but there is the issue of readership. Something tells me that a Gentoo article won't go over too well. Linux in general is already a fraction of the audience of Windows-based articles, going with anything but Ubuntu, or perhaps Fedora or openSUSE, would probably doom the article before it even publishes. Sorry, I didn't make the market share situation, but I have to be aware of it.

 

[user 18]

I prefer firefox, have not used IE since version 6, LOL @ 8-tab limit. I have recently found that Chrome has it's place for slower to load sites, but that I prefer the UI of firefox 3.6.x. Never tried Opera, maybe I should at some point in the future

 

[Guest]

what we can see here is, that all that windows extra memory for the os, if the program is properly programmed, pay off with less memory occupation for the various programs, because they share more common components than it's done on linux; and, beside that, i find funny the fact that one of the things linux fan always brag about is the fact that they're memory usage is lower on linux... but, by their definition, isn't a bigger occupation equals to slower perfomance ? And yet, that's exactly what we are seeing here: better windows performance, which by my opinion is partially caused by a better handling of low-number smp node on the windows part(if you optimize the software to run on 16-24 server processor it won't run the same on 2-4 cores no matter what)

Instead, people should think this: processing power and memory is limited, but not time; but if you can increase the amount of memory you can use at any given time there is a strong possibility that you can achieve higher performance level, because you don't have to continually cycle the data you need or gather it again, which saves time and clock cycles to process the data(not to talk about loading from the hard drive, an eternity to a cpu..)

give it some thoughts, i'd also like to know if i'm wrong

 

[Guest]

You said, you are using Gnome System-Monitor to benchmark memory usage. How exactly do you do this? You should know, that there are many things that can get you an completely false impression on memory usage with some of them are:

- The buffer! The memory manager in Linux is using free memory to buffer application memory and data from the disks.
- The preloaded libraries have an effect on memory consumption. In an Gnome environment like Ubuntu, applications using Gnome-/GTK- technologies can use preloaded ressources which boosts startup time and lower memory usage. Firefox (GTK) has an advantage here.
- If you add the per-process values of the processes used that are shown in the systems monitor, the multi-process browsers like Chrome will have a huge disadvantage, as they are sharing many ressources between each other. Your system monitor can not follow them and will count the same shared ressource, even if only once loaded, for every single process.


When testing browsers, I would use a lightweight distribution like Arch Linux or Gentoo, install a base system, then install X.Org and the browser, boot up to X.Org, start "free" with piping the output to a file, start the browser up as you want to test them (load a page, open tabs) and use "free" again to log memory usage. The value with the buffers excluded from the second run, minus the same value of the first run should be the memory allocated by your browser and all of its loaded ressources. Reboot and proceed with the next browser.

When considering real usage effects (typically, users do not work on a pure X.Org environment), you would have to test with different desktop environments, and, as Ubuntu is focused mostly on Gnome, even additional distributions, perhaps Fedora as a second Gnome Distribution and Mandriva and OpenSUSE for KDE testing.

 

[Guest]

You guys forgot Maxthon 3... Only the fastest browser in the world...

 

[Casper42]

I am curious how the chart on the last page was decided?

The HTML 5 section lists Chrome as weak, but when I backed up my browser 2 pages to the section with a HTML5/CSS3/ACID3 testing it looks pretty damn strong to me.

Just wondering what I am missing that cause you to rate it as weak?

 

[apcullen]

This is an excellent article overall.

However, I would argue that, for the end user, the only really important benchmarks are memory usage and page load times.

Firefox looks absolutely awful in many of the synthetic benchmarks, however it's page load times are decent. I think this article may expose more flaws in the benchmarks than it does in the browsers.

 

[luke904]

[citation][nom]Sihastru[/nom]Opera still can't render pages properly, still can't print content properly, and we waste our time with senseless tests of imperceptible speed.[/citation]

it doesnt render pages right SOMETIMES because its a standards compliant browser and it makes little to no effort to correct websites that are not compliant... thats what makes it fast.

i love opera and i use it for everything, i cant remember the last time i had to fire up chrome or firefox for something. sure somethings things are misaligned but seriously who gives a sh**....

as for printing.. never had any problems so im guessing its a PEBKAC.. if you dont understand then its defiantly one... lol
im kidding.. maybe it really isnt that good at printing.. NOT..lol



[citation][nom]Tamz_msc[/nom]The article that I was waiting for.How the tables have turned!Conclusion:Firefox is quite capable in both Linux and Windows.I'm using Firefox 4 beta and I find it pretty quick.[/citation]

odd conclusion considering the final placing...



------

ive used opera on my computer with 4 gigs of ram and my old computer with 512MB
sure it uses a couple hundred megs of ram with several cpu intensive tabs open (as in not just google results and wikipedia pages) but on my old computer it was very memory conservative... THIS IS HOW WEB BROWSERS WHOULD WORK, WHY THE HELL WOULD YOU HAVE 4 GIGS OF RAM IF YOUR PROGRAMS DONT TAKE ADVANTAGE OF IT??? the only reason i even got 4 gigs was because a year ago its was cheap.. $52 for 4 gigs of 1066... 2 gigs would have been plenty for my xp system. just keep the bloatware off the startup list

programs should use memory as much as it wants until that memory is needed... opera seems to, in my experience, do that.


now i leave you to think about what i said and me to think about why i bothered

 

[luke904]

[citation][nom]Syndil[/nom]Been using Opera since version 3.2. In response to people who claim it does not render properly:Opera is fully W3C complaint. If a page does not render properly in Opera, it is due to a failure of the page designer, not Opera. That said, I have not encountered any reputable sites that do not display properly in Opera, other than those controlled directly by Microsoft. Which, I am sure, is deliberate.Furthermore, all the talk of memory management is a non-issue. Opera does use more memory, but it does so for a purpose. Unless your machine is hurting for more RAM, it should not affect its performance. In other words, if the RAM issue is actually causing a noticeable performance problem (not just racking up high numbers in your task manager), fix your machine, don't blame the browser.[/citation]


thank you for clearing things up with the firefox fanboys...
lol

 

[mgillespie]

The section on Memory Manager demonstrates how clueless the tester is. Hanging on to memory is fine. It's only a problem if it hangs on to it when something else needs it (something these tests did not test for).

 

[mgillespie]

Cherry picked tests. Still not testing Javascript compliance with Sputnik I see, as Opera also wins that by a mile, and we can't have it so that your beloved American browsers don't win, can we...

 

[bhaberle]

[citation][nom]adamovera[/nom]When it's final, I'll test it.[/citation]
I'll be looking forward to it. =)

 

[nebun]

chrome is fast at first then it bogs down a lot, especially after a week of use

 

[Guest]

I am runing Linux Mint 9 (Ubuntu 10.04-based distro) on my laptop. My Chromium/Firefox difference in SunSpider is not that huge:
Chromium 5.0.375.99 - 656 ms
Firefox 3.6.8 - 2820 ms

I can confirm that Chromium uses a lot more memory than Firefox, despite Chromium's lightweight appearance. Chromium does feel somewhat faster though (until I run out of free RAM and need to swap😀).

 

[Guest]

Hi, I will love to see that test on Mac. I know, same hardware may not be possible (but you could install Windows or Linux on a mac), Also, you should include Safari. It will be a nice Windows vs Mac vs Linux competition.

 

[nightcrawl3r]

I love the UI of chrome. It's very easy to access everything and the browser doesn't look messy(like IE)at all.

 

[gaborbarla]

I kind of disagree with startup times. I find that on all my systems firefox launches much slower than Opera or Chrome. In fact firefox (with adblock, weathercast and videodownloadhelper plugins) takes quite a few seconds to start up on my system(s) while opera starts almost instantly.
Again this might be due the awesome plugins that firefox supports that I always install immediately on first use. Still, it can be quite annoying to wait upwards of 5 seconds to launch a browser.

Also Opera and Firefox keeps asking for proxy password even though I click remember my password while Ie7 does not. Unsure about Chrome.

Ie7 does not allow tabs to be dragged to quicklaunch bar, a feature that I like, but the other browsers like Chrome and Opera don't even seem to have a quicklaunch bar as they have a launch tab. Something that is a matter of taste.

Chrome remembers page zoom specifically for each page, while others like to change all pages zooms at once.

Overall one must take into consideration special features and shortcomings of each browser, before a winner is decided. In different enviroments I use different browsers depending on convenience.

The one thing that we should all be greatful for is the abundance of options at the moment. Healthy competition is needed in an browser arena which seems to be the platform of choice for future computing.

 

[fkr]

[citation][nom]haplo602[/nom]how about testing the browsers on a low ram system ? say 1GB ?

i run a 1 gig ram machine with a celeron M 1.5. opera is the best by far. I can open 40 plus tabs without issue. i can also save these sessions and reopen all 40 tabs without issue. all flash games on kongregate run fine, except when there are a million characters on screen. this is a processor limitation. hulu works well as does utube.

 

[Silmarunya]

On my Windows PC I find Firefox to be A LOT slower than Chrome or Opera which are about tied. When I'm using Linux (which is becoming more and more often to be honest), the difference is so small I can't subjectively notice it. So I just use Firefox 4.0, as it looks good and is open source.

 

[jaksun5]

An aside, I run an old PM 1.6 notebook with 512Mb of RAM and Ubuntu with Firefox 3 rocks, never skips a beat. Before upgrading from XP my system would crawl by the time the standard slew of updates were run. I know that the main area I'm benefiting is from the OS (from XP to Jaunty) but the fact that it's skinny on memory is a great help!

I also have a P3 600 running Puppy Linux 4.2 and Seamonkey, I can get a good 3 tabs open before it craps itself!

 

[jaksun5]

I should mention that's with on 128Mb of RAM