Web Browser Grand Prix: The Top Five, Tested And Ranked

[Guest]

I just read Matt Cutts talking about this review - I pretty much projected the winner when Cutts had a semi on over the review. Nice review though (barring Matts unsightly erection)!

 

[brjtexas]

Great article - really enjoyed the wide range of performance tests. One test I'd like to see in the mix for the next run is the performance benefit each browser sees from multi-core CPU's. One test to consider is moonbat, a threaded version of sunspider that uses HTML5 web workers, but would like to see other tests that expose the amount of threading in the various browsers. We're clearly heading into a future where performance comes from SMP multi-core scaling rather than ever increasing clock rates, so it would be very useful to see how well the browsers take advantage of such advances in computer architecture.

 

[schnitter]

I'd like to use chrome, but I hate not being able to:
- shift enter for .NET addition on address
- right click a picture and select "View Image"
- give no warning upon closing multiple tabs

I forgot what else I couldnt live without going from FF to Chrome.

In due time, I'll make a switch.

 

[industrial_zman]

how bout IE with Chromeframe testing and see if it can be on par with the others.

 

[Guest]

Wikipedia claims "The Mem Usage column on the Processes tab [in Windows Task Manager] is actually the process' working set. A process has little or no control over its working set, which turns this column useless to determine how much memory a process is consuming."

How do you defend using this metric in your benchmark?

 

[mrcairo]

In my limited research based on my normal everyday usage, Chrome by far handles memory allocation better than any other Browser I've used. Used Firefox for the last few years and after using Chrome for the last month I'll never go back. It's faster and handles everything I've needed to do.
Very glad I decided to give it a try.

Not sure how Firefox beat it out in your tests on Memory, that just hasn't been the case with my experience.

Regards

 

[Guest]

I use Chrome from time to time (well, actually, Iron, but that's another story) and appreciate what it does. Speed isn't everything, though. Features, web compatibility and customizability count even more with me, and Chrome is woefully lacking in both those areas. "Out of the box", Opera has the best combination of speed, web compatibility and customizability, IMO, and wins my approval over the feature-poor (but simple--something some folks appreciate more) and marginally customizable Chrome. Firefox, Safari and Internet Exploder aren't even on my plate for consideration. A browser that doesn't do mouse gestures, for example, right "out of the box" is just too crippled for words, IMO, and FF, Safari and IE (as well as Chrome) are all lacking such essential (well, to me ) features without add-ons--where add ons are even possible for some of the features I've become used to as "built in" in Opera. Nice that Chrome has emulate Opera's Speed Dial feature, though in a crippled sort of fashion, and that a similar feature can be added on in Firefox with an extension. Still, why have either crippled or add-ons when Opera has it built in, is my view.

BTW, did anyone doing the testing wonder if the memory usage Opera exhibited on the multiple tab loads was related to the speed of loading those multiple tabs? I really don't care if Opera takes up 150MB of memory to load the 30-40 tabs I may load at startup, as long as they load quickly and smoothely. What's a few hundred megabytes of memory usage on a machine with 8GB, anyway? (OK, so your test machine had half as much).

In the final analysis, FF, Opera, Chrome and Safari each have user groups that can benefit from their features and functionality, and as long as folks use ANY of them to migrate from Internet Exploder, the world will be a better place.

 

[clinamen_1]

I have strange behavior with FF 3.5. I cannot run two systems on a router using 3.5 or3.6. 3.0 runs fine. The OS is XP32, SP3.

With the later FF versions, which ever computer happens to be running FF first, it takes over. The second computer times out on all tabs. Eventually, some tabs do load, but do not refresh. Once they do, the first computer starts to misbehave.

This is also true if only one computer is running 3.5 or 3.6, but both running 3.0 is fine.

I have switched routers, talked to my IP (Qwest) who spent an hour and a half on line with me running tests.

I filed a bug with FF who referred me to a similar bug filed earlier, but so far no answers.

Maybe I'll try Chrome. It's already on my Win7 OS but I haven't used it.

 

[Guest]

I couldn't help notice that you left out that Opera doesn't work on Ebay, IMDB, Abc.com, or Netflix.com. Or so many others. Which is pretty hilarious.

 

[SlipUp]

The post you replied to mentioned three things and you replied to only one of them , speed? And you got that one wrong as the article clearly stated that FireFox won the page load test hands down. By the way last time I looked both referred to two things not three.

I disagree, Firefox is noticeably slower. The memory use thing is a farce. Firefox's memory handling is terrible.

 

[SlipUp]

the release of Opera 10.5 was hurried because of the ballot screen deadline. It's not good, but I can understand that, at least, they have a good reason (better than with 9.50). That could lead to some strange problems.

Haven't seen any problems. Also, they didn't rush 9.50, nor did they rush 10.50.

- The new JS engine (Carakan) is more RAM consuming.

Not necessarily. Opera dynamically adapts memory usage.

 

[SlipUp]

This leaves me wondering how Avant Browser and Seamonkey stack up. For the longest time I used Avant (being completely IE compatible, but ran faster at the time). However, I switched to Seamonkey because of its inclusion with Puppy Linux (meaning it was small and fast) and also from Mozilla. I'm on Windows 7 now, but my specs make it almost unbearable to run Firefox or IE on it.

Avant is just a skin for the IE engine, not a separate browser.

 

[SlipUp]

I couldn't help notice that you left out that Opera doesn't work on Ebay, IMDB, Abc.com, or Netflix.com. Or so many others. Which is pretty hilarious.

I have no idea what you are talking about. Opera works fine on all those pages on my computer.

With the exception of Netflix, which blocks Opera. You need to fool the site into thinking that you are using Firefox, and it works fine.

Chrome, on the other hand, crashes consistently on Gmail, and I can't use Firefox with Gmail at all

 

[zgubidan]

Good article, but I can not access the print button in Opera 10.50.
Please fix it.

 

[Guest]

I think it would be much more revealing to test memory by opening tabs and actually clicking through links. Many links. The idea is to simulate real web browser usage -- and nobody opens their browser to 1, 5 or 10 tabs, then does nothing with those tabs. In my experience, Firefox's memory issues are never about initial tab opening, but always about it ballooning in memory as you continue to surf the web, open new tabs, close old ones, etc. Opera also balloons, but Firefox was always much worse.

-- Pistos

 

[wildwell]

Really interesting article. Thanks!

 

[macumazan2010]

I've had memory usage issues with firefox. When I've closed multiple tabs and left just one up, firefox still uses around 200 MB of memory. I assume this is a caching issue since firefox is no longer rendering the pages.

 

[scooby-doo]

Hi there, although I am an avid user of chrome I would say these sorts of stats are relatively meaningless

1) Page Load time -- since first you have to have DNS resolving -- so who says that the DNS servers resolve the addresses of the pages to be loaded in exactly the same way / time.

2) The traffic on the actual websites themselves may be totally different for each test

3) Your own ISP / Network connections could also be affected by the amount of traffic on it -- such as other users when using Broadband connections.


This type of comparison is a bit like applying extreme statistical analysis to TWO variables or working out the standard deviation of 3 observations.

A more meaningful test is would be Security and accuracy of rendering the target websites.

These days just choose the browser you like -- there won't be a huge amount of difference in the performance of them as an end user - especially if you are connected to a workplace LAN where the network can get very busy at "popular times" such as around Lunch time etc etc.

 

[evilcokemachine]

Considering that you state that you expect most of your readers to be running modern hardware with lots of memory, and considering that this test claims to be a test of the speed at which a browser can display a webpage/webpages, I think that the analysis that you do in the "memory usage" section is misleading.

The fact that a browser uses more memory than another to display those pages is not itself bad, nor indeed is it an indicator of slow browser performance or page rendering. Using more memory can actually increase the speed of browsing, certainly with respect to webpage rendering.

The only circumstance wherein memory usage might be detrimental to browsing speed would be when all or a significant portion of the system memory were already being used. Taking into account what you stated at the beginning of the article regarding your assumptions of your readers' hardware, I think that the conclusions that you draw about the various browsers' usage of memory are inaccurate.

That is, because Firefox uses the least memory, for example, when loading five tabs, while Opera uses the most, does not necessarily indicate that the user's browsing experience will be slower in Opera than in Firefox. In fact, if the reason for Opera's higher memory usage is efficient caching &c and not leaks or general inefficiency and the like, then Opera may well offer a faster browsing experience than Firefox in this case.

 

[grege]

@Manwell999

In Chrome all you have to do is open a blank tab and the favorites menus appear. There are two, a frequently used favorites bar and a traditional list.

 

[Guest]

Alas, they cooked the books a bit on this one! One benchmark with Java showed IE to be slower, so they naturally INCLUDED A WHOLE PAGE OF JAVA BENCHMARKS! Why else but to make IE look poor? Why not a whole page of Silverlight benchmarks to make Chrome suck?

 

[ozzydt]

I've using Firefox and Opera simultaneously, I prefer Opera but it still having some compatibility issues and that's Firefox for, never tried Chrome, I'm fine with my pair of browsers.

 

[randomizer]

Alas, they cooked the books a bit on this one! One benchmark with Java showed IE to be slower, so they naturally INCLUDED A WHOLE PAGE OF JAVA BENCHMARKS! Why else but to make IE look poor? Why not a whole page of Silverlight benchmarks to make Chrome suck?

What are you talking about? There's only one benchmark on Java in the whole review, same as with Silverlight.

 

[arnweb]

This is not a fair test. You're running 32-bit application on a 64-bit OS. What about WOW performance issues. It's like running Win-3.1(16-bit) or DOS (16-bit) application on Windows 95 or higher. Also, how many users use Win 7 64-bit? (I guess less than 10%, currently)

 

[JohnnyLucky]

WOW! 14 pages worth of comments. Is THG headed for a new record?