Web Browser Grand Prix

An anonymous reader writes "After seeing Opera's claim to 'Fastest Browser on Earth' after their most recent release, Tom's Hardware put Apple Safari 4.04, Google Chrome 4.0, Microsoft Internet Explorer 8, Mozilla Firefox 3.6, and Opera 10.50 through a gauntlet of speed tests and time trials to find out which Web browser is truly the fastest. How does your favorite land in the rankings?"

Comments

Slashdotted

[Akido37]

The site is Slashdotted so hard, the link was removed from the summary to give the poor guys a break.

Re:Slashdotted

[number17]

No, it's just slow because you are using IE.

Link

[mingot]

http://www.tomshardware.com/reviews/firefox-chrome-opera,2558.html

Re:Link

[Snowblindeye]

A lot of these speed tests always compare javascript performance, which I have to say matters less for me on a day to day usage than other things.

At the end of the article (10 pages later), they do break it out into categories. The winner of the 'page load' category is: Firefox.

I care about other things as well, startup times for example (won by Opera), but if I had to pick one most important category for me, it's page load times. YMMV, obviously.

Shortcut to summary: http://www.tomshardware.com/reviews/firefox-chrome-opera,2558-10.html

Re:Link

[verbalcontract]

And the conclusions:

Category / Test: Overall Winner

Startup Times: Opera

Memory Usage: Firefox

Page Load Times: Firefox

HTML: Safari

CSS: Safari

Tables: Safari

JavaScript: Chrome

PeaceKeeper: Opera

Acid3: Chrome

DOM: Chrome

Flash: Opera

Java: Opera

SilverLight: Firefox / Internet Explorer

Re:Link

[0100010001010011]

Memory Usage Recount.

He just did static 'load page look at memory usage' tests. Which is fine. If you only load 10 tabs of pages and never visit anything else.

Firefox constantly eats memory on my MacBook. If I have both Firefox and Photoshop open, Firefox consistently eats more memory than Photoshop. Things will grind to a halt until I kill Firefox.

It was enough to get me to jump ship to Chromium, where aside from the occasional Flash Plug-crash, doesn't require being reset every hour.

Re:Link

[Dumnezeu]

I care about other things as well, startup times for example (won by Opera), but if I had to pick one most important category for me, it's page load times. YMMV, obviously.

I care about security and safety, so I just avoid IE. I care about privacy so I avoid Chrome. I care about bloatness so I avoid Opera. I care about functionality so I choose Firefox. I think it's the lesser of all evils. Correct me if I am wrong.

Re:Link

[ChronoReverse]

I'm curious as to which version you're using, whether you used a clean profile, and which plugins you're using.

My own testing of Firefox doesn't ever show the massive memory leaks often claimed.

Re:Link

[sweatyboatman]

how many add-ons were you running with Firefox?
how many are you running with Chromium?

Re:Link

[Anonymous Coward]

Windows doesn't even have fork(), you newb.

Re:Link

[amicusNYCL]

How come addons are a defense for Firefox eating a lot of memory? Does anyone run Firefox with no addons? Everyone always uses addons as the reason why they love Firefox so much, so why shouldn't addons be included in benchmarks? Every benchmark of Firefox should include the 3 most popular addons installed and running. Currently, that looks like Adblock Plus, Video DownloadHelper, and Personas Plus.

Real-world usage of Firefox includes addons, so should benchmarks. A benchmark of Firefox with no addons is useless data for anyone who actually does use them.

Re:Link

[Enderandrew]

Firefox has an intentional feature where they keep fully rendered pages in memory so they can reload faster when you hit back. They also keep full tab sessions in memory after you close tabs. You can turn these features off if you don't like them.

That being said, I leave Firefox open for days, if not weeks. I run tons of tabs, Greasemonkey scripts, extensions, etc. I haven't seen memory leaks since the Firefox 2.0 days.

I keep considered switching to Chrome, but Greasemonkey scripts still don't work properly, and I can't stop ads from loading. (Chrome adblock solutions render the ad even there is malicious code, but hides it from showing)

Re:Link

[WinterSolstice]

Not to mention, the lack of some of these addons (no xmarks bookmark sync, flashblock barely works) actually keep me from using Chrome.

I just gave it another whirl, and on my machine it is a bit faster... but I'd have to port all my passwords by hand. Between 7 machines.

No thanks.

Re:Link

[Dracker]

Firefox with even just a couple extensions is WAY more bloated than Opera.

Re:Link

[element-o.p.]

The other day, I turned JavaScript off on my browser (I had a reason...maybe testing or annoyed by ads on a page...I don't remember exactly), but forgot to turn it back on after I was done with whatever it was that I was doing. A little later, I opened FF again, and wondered why so much of the content I expected to see in my browser was missing.

As you said, YMMV, but I would say that JavaScript execution time is pretty much every bit as important as page load unless you have limited your web browsing to pages created back in the '90s.

Re:Link

[blind biker]

That link is right smack in the submission!
This must be the most redundant post in Slashdot history, and yet it got modded up.

Re:Link

[clone53421]

It was added. The /. editors occasionally do edit.

Re:Link

[curunir]

The memory comparison tests were flawed enough to keep me from taking any of their results seriously. While there was very little mention of how memory usage was determined, what little there was indicated that he used the task manager and, for Chrome, added up the totals for each Chrome process.

This is a well covered mistake that has been pointed out since the first tests that showed Chrome being a memory hog. And while I won't get into it, the simplistic method the review seemed to use shows a complete and utter lack of understanding of the concept of shared memory. So many people have made this mistake that the Chrome developers even created a page you can load in Chrome that will show you an accurate total of the memory used not only in Chrome but in any other browser you have running. But I would have expected someone working for a technical publication who's about to publish an article partly on the subject to do the limited research necessary to learn why such a simplistic measuring tool such as the task manager is completely unsuitable for tests like the ones they performed.

Suffice it to say if the memory usage comparison was as naive as it appears to be, I've got very little confidence in any other metrics they gathered or any conclusions they reached.

Re:Link

[bmk67]

How come addons are a defense for Firefox eating a lot of memory?

Because it's quite possibly the addon, and not Firefox that's eating the memory?

Personally, I think it would be interesting to see some benchmarks done on Firefox with and without some popular addons.

Re:Link

[element-o.p.]

How come addons are a defense for Firefox eating a lot of memory?

Unless Chrome, Opera, Safari, etc. have a similar function built in, you aren't comparing apples to apples otherwise. If you start including addons, it would be possibly for a biased tester (MS perhaps? Google? Apple?) to pit FF with the most heavy-weight addons (s)he can find against another browser in a minimalist configuration. It might be reasonable to consider the memory usage, page load times, etc., for each of the browsers in a "typical" configuration, but then you have to decide what "typical" means. For example, I don't use Video DownloadHelper, and I've never heard of Personas Plus, but then again, I'm a bit of a Luddite, despite (because, perhaps?) being employed in IT. Consequently, benchmarks with those addons have little to no bearing on *my* typical usage.

FWIW, and in part to answer your question, "Does anyone run Firefox with no addons?", the only FF addon I typically use is DOM Inspector, because from time to time, I create web pages that include JavaScript. Having said that, I would still agree that Adblock Plus should be included in the "typical" addons list.

Re:Link

[element-o.p.]

Nuts...sed "s/possibly/possible/" in the second sentence. Missed that in the preview.

Re:Link

[blind biker]

Embarrassing.

Re:Link

[EvanED]

If it does things intelligently, which I'm sure it does, it makes sure as much as possible of the browser is initialized off of the process is forked. I'm pretty sure Win32 is the same as Linux w.r.t forks - those pages are mapped COW and don't actually impact the memory usage.

And I'm *positive* you're wrong. Unless you mess with the Native NT API's process create (this is, I think, officially undocumented), Windows process creation works entirely differently from Unix's. You don't get fork() at all, just something like the fork/exec combination.

Re:Link

[0100010001010011]

About the same.

LastPass.
AdBlockPlus on Firefox, GlimmerBlocker for Chrome/Safari.
XMarks on both.
Greasemonkey on Firefox, GlimmerBlocker for Chrome/Safari
WebDeveloper on Firefox, Built in functionality (which is loads better) on Chrome/Safari.

Apples to Apples, same functionality, 10x the memory.

Re:Link

[ottothecow]

I am currently using firefox with no addons. I installed it on my work computer not long after I started (and discovered that unlike my previous workplace, nobody was going to come find me every few months and ask why I was running unauthorized software). Since then, I have not needed any addons.

On my home desktop I use use a set of addons that have grown over the years. Download statusbar, some image zoom thing, some exif properties thing, maybe a mouse gesture addon that I don't actually use, maybe a few other things. No adblock or flashblock on that computer right now (though I think I have some hosts file edits to block those inline text ads...I am just too good at ignoring ads to bother).

On my netbook, I have download statusbar (or I think I added it after my most recent ubuntu upgrade), adblock plus, flashblock and maybe something else. The pair of blockers are more there for speed reasons as firefox performs much better on a low powered system when it is loading less stuff (duh)...

From this, it looks like the only time I use one of the listed top addons is on an underpowered system to ostensibly improve performance...using adblock on a page render test seems like cheating...

Re:Link

[DMUTPeregrine]

I have quite a few extensions*, 512MiB memory cache capacity set, and rarely see Fx using more than 550MiB of RAM. I don't have to reset it every hour, or even every day. *23 to be precise: Adblock Plus, All-in-One Sidebar, AutoPager, BetterPrivacy, DownThemAll!, FireGestures, Gmail Manager, Greasemonkey, Image Zoom, Leet Key, Morning Coffee, NoScript, Nuke Anything Enhanced, Password Hasher, PDF Download, RSS Ticker (CPU hog, that), Session Manager, Stylish, TACO, Tree Style Tab, Update Notifier, XUL Profiler, Youtube Comment Snob. Firefox 3.5.8, running on a Kubuntu 9.10 system. There could be Mac specific bugs, or some extension or plugin problem. But Firefox has gotten much, much better about actually freeing up RAM with the newer versions.

Re:Link

[rtaylor]

It isn't bloat if they are features you want. It is only bloat when they are features somebody else wanted.

Re:Link

[obarthelemy]

very frankly, as long as its in the same ballpark, speed don't matter.

I find Tom's article ridiculous, for at least 2 reasons:

1- they focus on performance, and disregard features completely. That's their choice, but it's an idiotic one

2- they compare perfs in wildly different configs: a fully usable Opera (with its integrated mouse gestures, adblock, noscript, synch...) vs a unusable barebones Firebox with 0 addons.

Re:Link

[hibiki_r]

The fact that a lot of content involves using javascript doesn't really mean that javascript performance is every bit as important as page load: It's not a matter of quantity of content, but of the amount of computation required to render it.

If the javascript is doing an XMLHttpRequest to back to the server, and needs a tenth of a second to do it, while it takes a second do page load, is the javascript performance that important?

Picking a browser for its javascript performance is a bit like selecting a language to serve dynamic web pages by looking at language performance. To serve a web page that takes most of its content from a database, we must use a programming language, but for the most part, the language we use doesn't really affect performance much, because the time it takes to fetch the data makes even the slowest production languages be a small blip in the performance for most of us. We don't end up writing all of our pages in C.

Re:Link

[bunratty]

Because out of thousands of addons, some are written by poor programmers and have horrible memory leaks. If you use one of those few addons with a horrible memory leak, Firefox will use incredibly large amounts of memory within a short while. As long as you don't use one of the buggy addons, you should be fine. You're probably correct that some memory benchmarks should be done with a handful of very popular addons, but those are the ones that don't exhibit the memory problems.

Re:Link

[houghi]

At the end of the article (10 pages later), they do break it out into categories. The winner of the 'page load' category is: Firefox.

Now what would slow down my browsing experience more? The browser or the 10+ pages the article is spread over?

Re:Link

[element-o.p.]

We don't end up writing all of our pages in C.

I know a guy who does. He's a bit of a freak, though, so that doesn't invalidate your point :)

Re:Link

[gmthor]

Windows doos have POSIX compatibility, so yes it does have fork().
But it's probably more efficient to use the Windows equivalent.

Re:Link

[MaXimillion]

Which is why I use Firefox. Not too many features included by default, but by far the most addons, plugins and scripts to give it all the functionality I want.

Re:Link

[Goaway]

Out of the browsers available, Chrome and IE8 are probably the securest right now, thanks to proper sandboxing. Chrome is also the least bloated.

Re:Link

[Goaway]

So, what, you're complaining that Firefox didn't lose hard enough?

Re:Link

[alexo]

That being said, I leave Firefox open for days, if not weeks. I run tons of tabs, Greasemonkey scripts, extensions, etc. I haven't seen memory leaks since the Firefox 2.0 days.

I have to restart FF every couple of days (and sometimes several times a day) because memory usage goes off the scale.

When I start it, with 25-30 tabs open from the last session, it uses 300-400 MB.
After I use it for a while -- opening and closing new tabs, downloading stuff, etc. -- its memory usage begins to grow until it reaches about 1.5GB, at which point FF hangs hard and needs to be killed.

Sometimes I get the memory usage growth even when not actively using FF, just by leaving it open for the night.

I suspect that the culprit is a plug-in, an add-on, a GM script or a combination of several of them, but I'm having trouble isolating the case because some of the stuff I do with FF requires certain add-ons/scripts so turning them off will not constitute a clean experiment (for example, downloading images from WebShots).

I'd like to see your list of add-ons (Nightly Tester Tools can generate it) and GM scripts to help rule things out.

Re:Link

[alexo]

Because it's quite possibly the addon, and not Firefox that's eating the memory?

Are there tools that can point out which add-ons allocated the memory?

Re:Link

[Enderandrew]

I generally only really run with Adblock Plus, Greasemonkey, and Nightly Tester Tools. The only GM scripts I run with are for Facebook.

If I hazard a guess as to your situation, I'd guess it is likely a single GM script.

Link to the story

[Sidn]

http://www.tomshardware.com/reviews/firefox-chrome-opera,2558.html

Re:A link to the article would be nice.

[Lunix Nutcase]

The link was in the original submission. ScuttleMonkey apparently is too much of an idiot to remember to have copied that along when posting.

Of course ...

[daveime]

I'm sure the loading times of all browsers would be faster if the "article" wasn't spread over 11 damn pages ...

Perhaps they could run comparative tests on ad-blocked and flash-blocked vs vanilla spam versions ?

And am I the only one who finds it fucking cynical in the extreme, to force you to surrender your email address just so you can use the printable version and skip the advertising crud ?

Re:Of course ...

[Lunix Nutcase]

And am I the only one who finds it fucking cynical in the extreme, to force you to surrender your email address just so you can use the printable version and skip the advertising crud ?

They only want to provide such a feature to members of the site. What's cynical about that?

Re:Of course ...

[daveime]

Because I am worth more than the 0.001 cents they will get for selling my email to spammers ?

Re:Of course ...

[Lunix Nutcase]

Because I am worth more than the 0.001 cents they will get for selling my email to spammers ?

Since when did they ever do such a thing? Secondly, even if they did what is cynical about that?

Re:Of course ...

[Sir_Lewk]

I stand corrected, there is apparently a mod here who doesn't know what it means either and thought me deserving of an overrated moderation (the most overrated of moderations...). I'll clear this up:

cynical - distrusting or disparaging the motives of others.

Tom's Hardware isn't being cynical, he is being cynical. This isn't just a minor usage error, he got it completely in reverse.

Re:Of course ...

[Mashdar]

Because I am worth more than the 0.001 cents they will get for selling my email to spammers ?

Since when did they ever do such a thing? Secondly, even if they did what is cynical about that?

lol.

Re:Of course ...

[element-o.p.]

Better yet, find their contact e-mail off the web page, and use that instead.

Re:Of course ...

[daveime]

Correct definition, reversed usage. What I meant was ...

THEY are being cynical, because they don't *really* believe people are going to print the article, people are simply using it as a way to bypass the advertising real estate and 11 page loads to read 11 paragraphs.

So they hide it behind a registration / login page, even if you manage to read the review without having to endure the ads, they'll still get their pound of flesh by selling your email addy anyway. It's not a "member benefit", it's a way of catching you on the way out if they haven't already caught you on the way in. Something like Jehovahs who camp outside your letterbox all Sunday, knowing you'll have to go out sometime.

Re:Of course ...

[Sir_Lewk]

Membership has benefits. This isn't cynicism, it is business. Either way you spin it though, you are pretty damned cynical.

But...but...

[Lunix Nutcase]

"After seeing Opera's claim to 'Fastest Browser on Earth' after their most recent release, Tom's Hardware put Apple Safari 4.04, Google Chrome 4.0, Microsoft Internet Explorer 8, Mozilla Firefox 3.6, and Opera 10.50 through a gauntlet of speed tests and time trials to find out which Web browser is truly the fastest. How does your favorite land in the rankings?"

I use Lynx you insensitive clod!

Chrome = teh winnar!

[ZERO1ZERO]

followed by Opera, then Safari, Firefox, and IE.

analysis and conclusions

I just installed Opera 10.5 and it's decently good enough for me to continue using it .

Re:Chrome = teh winnar!

[roju]

Although Firefox somehow wins the "Page Load Times" category, which seems more important to me than javascript benchmark speed.

Re:Chrome = teh winnar!

[Rockoon]

Chrome was the winner, but unfortunately one of the JavaScript tests was the Google benchmark. Thats a facepalm for toms hardware this time.

I would love to see these tests done with only independent benchmarks.

Re:Chrome = teh winnar!

[Lunix Nutcase]

And one of them was Apple's, another was Mozilla's and another was an independent 3rd party's test suite.

Re:Chrome = teh winnar!

[ircmaxell]

What's weird, is looking at their page load time benchmarks, it's only first for Tom's Hardware... IE takes 2 out of the 5 sites, followed by FF, Chrome and IE at 1 each. What interests me, is the only load time that Chrome is slower than the average is YouTube. Couple that with its shoddy flash performance, and to me it indicates some kind of a bug or regression either with Flash itself, or with the plugin system inside Chrome. I use Chrome as a primary browser on Kubuntu, and I must say it is slow, but definitely faster than FF...

What's interesting, is that it shows Chrome losing on the memory front for all but 1 tab... This has more to do with the 1 process per tab model than it does with "inefficiencies", and paints a false picture (considering each of the rest use a single process for all tabs). If people understand the benefits of 1 process per tab, I think it justifies the added memory usage. I mean we're only talking about 146mb for 10 tabs. It's not like we're talking gigabytes...

Something I would like to see a lot more than this kind of benchmark is interface speed testing. How long does it take to switch between tabs? How long does it take to scroll to the bottom of a 1mb page. Etc... That's why I like Chrome better than the rest. Not because it's page load is faster, but because everything is faster (at least based on my non-scientific tests)... I'd also like to see some cross-platform testing. Sure, Chrome may win out on Windows. But what about on Ubuntu? What about on Mac? What about on Fedora? etc... This comes to an even harder point when you look at the shoddy flash implementation on Linux. The difference between 1st and 2nd may be a lot bigger than it is on Windows...

Re:Chrome = teh winnar!

[ircmaxell]

Remove that test from the results, and Chrome still wins. But look at the results of that test. Chrome wins, yes. But not by a HUGE margin (the difference between second and third is larger than 1st and 2nd). At least it's not as bad as the Dromaeo test (Where Opera is out in front by so far, it seems more like a bug in the test than a win for Opera)...

Re:Chrome = teh winnar!

[roju]

An important question about the memory test: how does Windows Task Manager account for shared memory or clean copy-on-write segments? If each Chrome process shares 90% of its memory, then those test results are going to be highly misleading.

Re:Chrome = teh winnar!

[Steve Max]

Another one was Mozilla's, and on that test Firefox got its ass handed to it by Opera (and Chrome/Safari, by a smaller margin). It's not Toms Hardware's fault if the people interested enough in Javascript performance to write a benchmark are usually interested in Javascript performance because they write Javascript engines.

Re:Chrome = teh winnar!

[Rockoon]

The fact that Mozilla's suite doesnt make Mozilla look good in no way means that we should be using Google's V8 suite to compare Chrome vs other browsers. I would argue that Firefox's results on Mozilla's own suite (good or bad) are meaningless, and Chromes results on Google's suite (good or bad) are meaningless, etc..

Google's suite is great for comparing Firefox, Opera, Explorer, and Safari.
Mozilla's suite is great for comparing Chrome, Opera, Explorer, and Safari.
Apple's suite is great for comparing Firefox, Chrome, Opera, and Explorer.

Its like using an nVidia benchmark to compare nVidia GPU's with Intel's and AMD's. Nobody would take such a benchmark seriously.

As far as Mozilla tanking on their own suite, the Open nature of the Mozilla Foundation as a whole probably has a lot to do with that. After looking at some of their tests, it looks to me like the source of their benchmarks is always based on some 3rd party, that Mozilla did not author the tests themselves (but certainly had selection powers) but instead have simply consolidated many disparate tests from many authors into a single framework.

Re:Chrome = teh winnar!

[BZ]

Turns out, there are no good independent benchmarks, because writing a decent browser benchmark is _hard_. The two sets of independent tests actually used in the article are broken beyond belief (don't measure what they think they're measuring, are easily gamed, etc).

In practice, even the dependent benchmarks aren't very good (Dromaeo measures its harness overhead more than anything else in a lot of cases, V8 tests are designed around the V8 engine, Sunspider has known bugs that cause some tests to produce random results and cause others to do different amounts of work in different browsers, that sort of thing). But they're somewhat better than the independent ones....

Re:Chrome = teh winnar!

[BZ]

Did you look at the "Mozilla" benchmark? It's basically V8 + Sunspider + some other stuff. Very heavily weighted towards the V8 + Sunspider scores. So that number is basically just the average of the Sunspider and V8 numbers.

And the Opera number there is due to the fact that the tester stopped and restarted the test, which resets the timers involved... Over here Opera does well on that test, but nowhere close to that far ahead of the others.

In all, this is about par for the course for benchmarking articles out there: someone who doesn't understand the (not particularly good and hard to use well) tools misusing them badly, then reporting the results.

Re:Chrome = teh winnar!

[BikeHelmet]

Sunspider is a bigger facepalm. When it first came out, it re-downloaded scripts every time it used them, and factored that into the scores. I recall trying it on ADSL, and getting amazingly better results than anyone on dialup or slow broadband.

It seems to be okay now, but back then it was a noob benchmark.

Re:Chrome = teh winnar!

[bingoUV]

Not a very popular configuration, but since you asked:
I run an Ubuntu 8.04 VM(VirtualBox) on Fedora 12 host. I use both chrome and firefox browsers in the VM, and tab switch times on Chrome are much higher than on firefox. Chrome is much less responsive in this respect than firefox.

Re:Chrome = teh winnar!

[hkmwbz]

Chrome is claimed to be the winner only because the author uses a strange way to count scores. If you count the average score, Opera wins.

Re:Chrome = teh winnar!

[hkmwbz]

The problem is that the V8 benchmark is created to make Chrome look good. So it skips stuff Chrome is slow at. At least Safari "loses" at Sunspider, which makes it a bit more credible than the test Google made to make it appear like Chrome is much faster than the rest using trickery...

Re:Chrome = teh winnar!

[hkmwbz]

As I said in another comment: The problem is that the V8 benchmark is created to make Chrome look good. So it skips stuff Chrome is slow at. At least Safari "loses" at Sunspider, which makes it a bit more credible than the test Google made to make it appear like Chrome is much faster than the rest using trickery...

Re:Chrome = teh winnar!

[clone53421]

Really? That’s strange, let’s double check.

We’re going to award 5 points for every first-place score, 4 points for second, 3 for third, 2 for fourth, and we’ll give 1 point for last place. If you have a better ranking system, feel free to explain what it is and why it’s fairer than this. In the meantime...

Chrome: 7 firsts, 9 seconds, 2 thirds, 4 fourths, 3 fifths. Total: 88 points
Opera: 7 firsts, 5 seconds, 6 thirds, 3 fourths, 4 fifths. Total: 83 points
Safari: 5 firsts, 4 seconds, 12 thirds, 2 fourths, 2 fifths. Total: 83 points
Firefox: 4 firsts, 5 seconds, 4 thirds, 10 fourths, 2 fifths. Total: 74 points
IE: 2 firsts, 3 seconds, 3 thirds, 3 fourths, 14 fifths. Total: 51 points

Now, you were saying?

IE and Facebook...

[clone53421]

For whatever reason, Microsoft's browser loads the Facebook homepage with extreme haste. Firefox, Chrome, and Opera take second, third, and fourth (respectively). Safari takes almost twice as long as the second-place finisher Firefox, and more than four times as long as IE.

Probably because Facebook cuts out a lot of the functionality that IE wouldn’t support anyway?

Re:IE and Facebook...

[Korin43]

Or because it's easier to render a page if you don't care about bugs. I'm guessing IE's score on the ACID tests and speed rending certain pages are related.

Re:If you want a fast web browser...

[bakawolf]

I wager it fails the ACID tests.

Re:If you want a fast web browser...

[Virak]

And netcat > /dev/null wins over even that, but for some strange reason people are interested in browsers that have more than the bare minimum of standards support.

You newbie

[MrEricSir]

Lynx is for newbies. Real men telnet to port 80 and type in the HTTP headers manually, then parse the response in their minds.

Re:You newbie

[clone53421]

They took telnet out of 64-bit versions of Windows. :(

Re:You newbie

[PIBM]

Why is there a 'then'. It should be done instantly, else you ain't a real one!

Re:You newbie

[iamapizza]

Telnet is for boys. Real men hold the cable up to their eyes and blink in the voltages representing the packets requesting the URL. The response is irrelevant.

Re:You newbie

[NevarMore]

The response is irrelevant.

Thats an awfully obscure way of saying "who reads the articles"! :)

Re:You newbie

[StikyPad]

Mere child's play. Real men telnet on port 443.

Re:You newbie

[HaZardman27]

I'm pretty sure it's there, it's just disabled by default in Vista and 7. You need to the add or remove Windows features window to install it.

Re:You newbie

[Enderandrew]

You just, but RMS apparently does something similar to this. We fires up page requests from the command line, and has the results return in a mail client as pure text.

Re:You newbie

[Enderandrew]

But there are 64-bit builds of Putty.

Re:You newbie

[element-o.p.]

1) That's (part of the reason) why I use Linux. I need a toolkit that is useful in my job, and telnet is one of those required tools.

2) putty will do telnet.

Re:You newbie

[Hurricane78]

No. Real men don’t hide behind encryption like sissies.
And real man don’t use telnet, but just pipe the network stream right onto the screen, while writing packets with only the num pad (with the non-numeric keys on it configured as A-F).

If we want encryption, we do it in our heads. ^^

Re:You newbie

[element-o.p.]

First, reread GPP: "They took telnet out of 64-bit versions of Windows :(".

Second, this was an example of a tool that is (sometimes) missing in Windows. IIRC, and I could be mistaken, it is missing from XP Home. GPP says it is missing in 64-bit Windows. Maybe so, maybe no, but even when it is present, it is significantly lobotomized by default. On 2K and XP, for example, when telnetting to port 25 or port 110 to send/receive e-mail manually (something I do regularly as an ISP sys admin), it does not echo back what you type by default, and it also royally screws with the formatting when you use the delete or backspace key (can't remember which; it's been a while). I recall seeing somewhere that there is a flag that you can set to alter this behaviour, but since I am already familiar with Linux, I'd just rather use it.

For perhaps a better example, I believe it was Fyodor who complained that SP2 on XP (again, IIRC) broke the way Windows handles raw sockets, which significantly limited the way nmap works. I believe there have since been workarounds (not being a Windows user, I haven't bothered to follow the issue very closely), but that right there is the crux of the problem -- in my experience, I find that I have to work around Windows. OTOH, I can simply *use* Linux. Maybe that's because I use Linux more often, and therefore it seems intuitive to me, whereas Windows doesn't. Maybe someone who is really good at Windows would have to work around Linux. Whatever; it's all good. Linux is the most effective toolkit for me. If it's not for you, then more power to you.

In any case, good job making an uninformed decision about how informed my decisions are on the basis of two sentences I wrote on a forum [:rolleyes:]

Re:You newbie

[element-o.p.]

That's a seriously ignorant comment.

Were I using telnet to gain a command shell on a remote device, you might have point (maybe). However, if you think that that's all telnet is good for, then you are showing just how little you understand about system administration.

For giggles, try this:
$ telnet slashdot.org 80 Trying 216.34.181.45... Connected to slashdot.org. Escape character is '^]'. GET http://slashdot.org/ <...snip...> $

If you can't see how that could possibly be useful for network testing, then I won't bother to continue arguing with you. I've got better things to do.

Re:You newbie

[rliden]

No they didn't.

Windows 7 x64 Home Premium has: Telnet Client, Telnet Server, RIP Listener, and TFTP Client. They are turned off by default for security reasons.

I don't know why they weren't in whatever version you saw, but they haven't been removed.

Re:You newbie

[clone53421]

C:\Users\Laptop>telnet
'telnet' is not recognized as an internal or external command,
operable program or batch file.

...was about the extent of my testing, other than looking through the start menu to see if they created something new to take its place.

But, yeah, anyway... found the option to turn it back on, so I’m good now.

Re:You newbie

[gparent]

I don't care what he said, telnet is still in 64-bit Windows. You're just as uninformed if you simply use his non-fact as part of the basis for not using Windows. There's plenty of reasons not to use Windows, and plenty of reasons not to use Linux, but lack of a telnet client that in fact exists, is not one of them. Also nmap works just fine on WinXP. More FUD, yay!

Re:You newbie

[arth1]

I've had to do that exactly once, in the childhood of SSL.
Nowadays, I use openssl as a telnet substitute against secure web sites:

openssl s_client -connect hostname:443
CONNECTED(00000003) ... [certificate stuff] ...
GET / HTTP/1.1
Host: hostname
Connection: close

It works almost as well as telnet against a non-secure port. Only almost, because you can't go into the background with CTRL-] and do all the useful things telnet lets you. Still very useful for troubleshooting, though.

Re:You newbie

[clone53421]

Yeah... I was mistaken. Calm down.

Re:You newbie

[plague3106]

Is it helpful? Sure. Are there tools better suited to the job to help with something like that? Absolutely. If you're trying to track down an unresponsive web server, I somehow doubt the only thing you'd do is that command. Also, seems to make much more sense to use a brower... same result in much less effort using the program designed to talk on port 80.

Re:You newbie

[element-o.p.]

Except that a web browser doesn't show you the raw contents of what is being transferred back and forth. Okay, wget or lynx can, but why spend several minutes on the man pages trying to figure out which flags to use, when a simple telnet would already have showed you what you need to see. TCPDump and wireshark will give you the raw data, too, but that's sometimes an elephant gun when you are troubleshooting a mosquito sized problem. My entire point is that there is a multitude of tools that a knowledgeable admin can draw upon. Telnet is one of them. It's not always the appropriate tool, but sometimes it is, and I miss it when it isn't there, or when it is but it's a limited version.

Re:You newbie

[plague3106]

Yes, you do things the old school way. There's a program (i can't recall the name of it at the moment) which will monitor the http traffic (or other kinds of traffic as well) between your web browser and server. It will display all the http status codes, cookie information, everything in a nice neat package. But see it runs on Windows, so I guess you instantely bleet that its useless, nevermind that it does everything your tools do.

But go ahead and do things the hard, long drawn out way. I'll have already figured things out and moved on though.

If you want a fast web browser...

[John Hasler]

...block all ads with Privoxy and shut off Javacrap.

Crack Monkeys!

[voodoo cheesecake]

I don't care how many research monkeys on cocaine say otherwise, I'm sticking with Firefox!

Re:If you want a fast web browser...

[bennomatic]

And I'd guess that certain DOM tricks, like drag-and-drop, don't work too well.

Javashit

[sexconker]

Can we please stop measuring browser speed with javascript? Javascript is shit 99% of the time.

And before you get any ideas, flash is shit 99.9% of the time.

How about browser makers focus on making popular sites suck less ass? Imagine if IE9/FF4 came with official mods for sites. Load facebook.com? Get it without the bullshit!

You can obviously already do this with great control with plugins for various browsers, but for it to make any difference for the average user it has to be built in, officially supported, and transparent.

If facebook doesn't like it, fine, let them get into an arms race with the browsers, just like how advertisers are starting to fight back against adblockers (who are also fighting back).

I'm sick of the shitty shitty shit on the web.
Measuring how fast a browser can wade through that shit is pointless. How about you measure how well a browser power washes that shit off of the site before serving it up to me?

IGN.com as viewed by
- Default IE8 reference
- IE9's shitripper
- FF4's shitripper
- FF3.6 + ABP + NS
- Chrome 2's shitripper
- etc.

That's a comparison I'd like to see. And if that comparison got attention, maybe, just maybe, sites would be designed with less shit.

is Safari startup time really surprising?

[nxtw]

Besides the obligatory browser code, Safari on Windows uses a lot of libraries that only get used by Safari - CoreFoundation, CoreGraphics, CFNetwork, the Objective-C runtime, and its own GUI (a limited Win32 port of Cocoa?). It also uses libraries that could be shared and/or duplicate builtin Windows functionality - such as sqlite3, zlib, libxml2, libxslt, and pthreads. (I imagine it uses its own SSL implementation too.)

The IE startup time seems higher than it should, because it uses the most Win32 functionality. It uses threading, SSL, XML, etc. from Win32.

Re:is Safari startup time really surprising?

[Dragoniz3r]

I'm not sure it ultimately matters why it loads slowly. Safari is a great browser, and I don't think anyone is trying to take that away from it by saying that it loads slowly, but in the final analysis, you probably want a browser that loads as quickly as possible. In fact, I think the only browser that there was a negative attitude toward at all was IE (with good reason, imo). For the most part, they just presented their findings and let it go at that.

Re:is Safari startup time really surprising?

[generalhavok]

On my Mac, Safari starts up a lot faster than IE.

Re:is Safari startup time really surprising?

[nxtw]

In software engineering circles, this is known as "bad design". Apple is good at many things, but writing software for non-Apple operating systems is not one of them.

Other browsers have the advantage of not using Objective-C. I suppose other Windows browsers are simply compiled with the Microsoft C++ runtime?

Re:is Safari startup time really surprising?

[nxtw]

Startup time on the Mac is even less important. On OS X I leave Safari open all the time, but with no windows opened when I don't use it. On other systems, closing the last window will usually quit the application.

Re:is Safari startup time really surprising?

[generalhavok]

I believe that much of the core functionality of Safari is part of OS X. On the Mac, most applications continue to run, even if you close the windows. This gets to be a problem where I work, where we have our graphics designers on Mac systems. They frequently complain about their systems being slow. That's because they'll run Photoshop, and close the windows, and not realize it's still running, and then launch InDesign, and Illustrator, and close those windows, and leave the applications still running, and then they'll open up Office to read some documents, and close the windows, and not realize Word or Excel is still running. Then of course, they have Safari running and Mail running, and usually iTunes running for their music.. Then they wonder why their systems are slow... So sometimes, that feature isn't much of a feature. You have to remember to actually go to the menu and close the application once you are done. At work, I use Macs, at home, I use Linux :)

Re:is Safari startup time really surprising?

[Goaway]

And re-writing an entire codebase from scratch for a secondary platform you don't care much for is good design?

Re:Fuck. That. Multipage shit

[insufflate10mg]

Too lazy to move your finger to click a mouse to continue reading? Sad...

Favorite Browser

[Anonymous Coward]

How does your favorite land in the rankings?

If it's your favorite browser, what does it matter how fast it is?

Re:Favorite Browser

[generalhavok]

If it's your favorite browser, what does it matter how fast it is?

A browser is a lot like a girlfriend. My favorite browser is my favorite because I know how it works, the little tricks and shortcuts, and I've been with it for a while and it's true to me. It's kind of like having a girlfriend. Sure, she may not be the prettiest or fastest or smartest girl around, but she's there for me, and I know how she works, and I know all the little tricks and shortcuts to make her work well for me.

Page load times...

[cplusplus]

...was won by Firefox, according to the summary at the end. Isn't that what the average user cares most about? How fast a page loads?

Re:Page load times...

[shutdown -p now]

I wonder how reliable the JS script they've used to determine when page has finished loading really is. Could it be that browsers that report higher scores in the test are simply more truthful about what they finished loading? (e.g. do all of them correctly account for plugins?).

There's one other thing. Historically, the usual trick employed by browsers is to delay rendering the page until it is partially loaded, so as to not constantly re-render. This speeds up the overall page load, but starting to render faster may well show the important parts of the page (those that user cares about) earlier, and if the renderer is fast enough, re-rendering the page repeatedly as it is being downloaded may look "smoother" from user's perspective, and be more usable.

I know that this setting is configurable for Opera, though I don't recall what the default one is. I think it's also configurable for Firefox. IE always has a pretty significant delay there, and I believe it's hardcoded. No idea about Chrome & Safari. Anyway, my point is that, if this setting varies by default, timing of complete page loads can give quite differing results which are not reflective of actual user experience.

Re:Page load times...

[shutdown -p now]

Also regarding memory usage... Opera has configurable memory cache setting. I've just checked what mine is - I don't recall changing it after installing 10.50 - and it's 200Mb. Looking at the tests they've done, none of them loaded enough tabs to fill that, so it is reasonable to assume that Opera just felt free to cache everything on loaded pages, and not clear the cache, since the limit isn't exceeded.

I wonder how the same test would go if they opened, say, 30 tabs...

Re:Page load times...

[VoiceInTheDesert]

Somewhat true, but I find firefoxe's memory usage and startup time to be really irritating. That and it's gotten cluttered. Chrome's sleekness was the first big reason I switched over. Only the URL bar and bookmark bar take up screen space. The rest of it is dynamially hidden (status bar) or compartimentalized (the settings and configure menus) into smaller areas so as to not take up as much space.

Page loading is well and good, but when it comes down to it, browser's physical size, resource footprint and startup time are more noticible to me.

Re:Page load times...

[Colonel Korn]

Somewhat true, but I find firefoxe's memory usage and startup time to be really irritating. That and it's gotten cluttered. Chrome's sleekness was the first big reason I switched over. Only the URL bar and bookmark bar take up screen space. The rest of it is dynamially hidden (status bar) or compartimentalized (the settings and configure menus) into smaller areas so as to not take up as much space.

Page loading is well and good, but when it comes down to it, browser's physical size, resource footprint and startup time are more noticible to me.

Firefox won the memory usage title in TFA, too. Also, I find that a properly customized Firefox has one of the smallest browser UIs. There's a title bar, then a single bar that includes all menus, shortcut icons, bookmarks, URL, and a search bar (and it's not cluttered if you've got at least 1280 pixels of width), and then either a list of tabs or a web page.

What I'd really like to see is a scrolling benchmark. If I have a page with a bunch of Slashdot comments and I want to scroll through it at a given rate, which browser shows me the smoothest motion? It's not just aesthetic. Smooth motion means you're missing less as you move and it allows more precise stopping. In my experience Safari and early Chrome versions were very bad at scrolling, but more recent Chrome is about as fast as Firefox. Opera won my personal comparison, but it was only a subjective test.

Re:Page load times...

[izomiac]

Startup time is the most visible, and essential for when you want to quickly check a single website (e.g. googling something really quick). Javascript speed is the limiting factor for web apps, flash speed for gaming.

Page load time is important, but dwarfed by network latency and speed in non-pathological cases, so I'd actually guess it's among the least important for end-users. Also, while there was a 20% difference between fastest and slowest, that's only about 1/26th of a second so it's approaching the limits of human perception. That said, ignoring 40 ms here, 50 ms there will lead to users finding a program laggy but not being able to specifically point out what's slow.

Re:Page load times...

[clone53421]

Startup time is the most visible, and essential for when you want to quickly check a single website (e.g. googling something really quick).

Funny... for me, startup time is the least visible, and not essential for when I want to quickly check a single website (because it’s almost always already running).

Re:Page load times...

[element-o.p.]

Start-up time isn't so important to me, because I typically have a web browser open when I am using my computer. If I only start my browser once or twice a day, that's not such a big deal. On the other hand, that does have the effect of making memory usage more important.

Re:Page load times...

[ChronoReverse]

Not to mention the capability to use the extension Tree Style Tabs. This alone would keep me on Firefox let alone how it also has things like Noscript (while Opera has per site javascript settings, it's not even close to the finegrain control Noscript provides in a simple manner).

Re:Page load times...

[houghi]

Sure page load time is important, but is the difference enough to influence your choice of browser?
Graig list has a difference of 40 milliseconds. Not something a user would notice or be bothered with.
Facebook has a difference of 300 milliseconds.

When adding up all the numbers of the pages speed test, the time difference between 1st and second is 181.8 microseconds. Between second and third it is 386.6 and then 11.4 (Then a wopping 1054.4)
So the difference of all thosw pages between first and fourth is just above half a second with a total load time of 4.5 to 5 seconds.

And if you look at the load times for Tom Hardware Website itself, you start to notice that not the browsers are the real issue, but the websites. If you are so interested in browser speed, repair you site.

Re:Page load times...

[BikeHelmet]

Firefox re-renders while loading. That it scored best indicates quite a superior experience as far as page load times.

Re:Page load times...

[shutdown -p now]

All browsers re-render while loading. But the initial render pass is not done immediately after the browser sees the first tag in HTML body - it waits for a bit longer to render something that's actually meaningful. That delay was what I was talking about.

Re:Page load times...

[hkmwbz]

Opera would have easily won with 30 tabs open. Opera is really the only browser that easily handles a lot of tabs.

Re:Page load times...

[hkmwbz]

Could it be that browsers that report higher scores in the test are simply more truthful about what they finished loading?

So IE is more truthful, and everyone else is "lying"?

Re:Page load times...

[shutdown -p now]

There may be many factors at play, and this might be just one of them. I know that, subjectively, IE is abysmally slow (and it is also the only measurement that really matters for user experience). Opera is very different, though.

Can't be a real Grand Prix

[Anonymous Coward]

The American contenders turned up

Re:They all win

[AndrewNeo]

Well, except for Internet Explorer.

Functionality More Important Than Speed

[amustic]

Firefox may not be the fastest, but with its builtin function plus rich array of addons, it's the most useful.

Re:Fuck. That. Multipage shit

[thatblackguy]

It is making the user click more to watch more of their ads. It's a needless restriction and that's why it sucks.

Re:So?

[Joce640k]

Only one browser in the list has adblock/noscript/flashblock.

Without those the other browsers are automatically losers no matter how fast they start up.

Firefox

[diegocg]

As always, Firefox ate much less memory than competitors...specially against opera & chrome.

Re:Firefox

[garaged]

that and some plug-ins that still don't have correct counterparts on chrome and opera keeps me getting back to Firefox.

But to be honest, just the memory usage is enough to not want to change to other, I have 2 freaking GB of RAM and I have slow downs because of the memory usage of chrome and opera, thanks but I rather run a little slower javascript but keep my computer responsive.

Performance I care about is hard to measure

[nxtw]

I care about things like responsiveness. How long does it take to redisplay after switching tabs or adjusting zoom? Is the UI still responsive when another tab/window is busy? Are scrolling and window resizing smooth? Will the browser respond well if the internet connection is lost / the system wakes up from sleep, when using AJAX applications like Gmail/Google Reader? (I had problems with one browser behaving badly with Gmail/Google Reader if the pages were open before entering sleep mode.) Will the browser perform well over RDP, VNC, or NX?

Start-up time isn't very significant - I generally leave browsers running all the time. Memory usage isn't very significant unless the system is low on memory. Otherwise, I prefer that the browser uses as much memory as it can to cache things. Rendering/script delays are not noticeable on modern systems.

Re:Performance I care about is hard to measure

[shutdown -p now]

I care about things like responsiveness. How long does it take to redisplay after switching tabs or adjusting zoom? Is the UI still responsive when another tab/window is busy?

Speaking of responsiveness, one neat thing about Opera 10.50 - all tab-specific dialogs are modal to the tab, not to the entire browser window. This means that, if a tab loading in background displays a JS alert, it doesn't suddenly pop up in your face requiring immediate attention - instead, the tab will get a marker indicating that something changed - and you can freely switch back and forth between tabs without closing the dialog first.

Re:Performance I care about is hard to measure

[clone53421]

I really, really want a Firefox addon to do this. JS alert-bombs could kiss my ass.

At least per-tab processes, if Firefox would ever implement them, could be killed to shut down a particular tab without killing the entire browser.

Re:Performance I care about is hard to measure

[John Whitley]

Start-up time isn't very significant - I generally leave browsers running all the time.

Motion seconded. I find this metric pointless in isolation, since the actual browser startup time is dwarfed by reloading persisted tabs across sessions. User-facing startup time for me is effectively (app startup + time to load up needed tab(s)). Tab persistence is an absolutely must-have feature as far as I'm concerned -- the browser can crash, I can close the browser, or even shutdown the machine without "losing my place".

Re:Performance I care about is hard to measure

[bingoUV]

Memory usage isn't very significant unless the system is low on memory. Otherwise, I prefer that the browser uses as much memory as it can to cache things

I guess this decision should be left to the OS. The way it should be done is: instead of caching in memory, the browser should cache in filesystem. If the OS is generally experiencing an abundance of memory, the filesystem dump would remain in OS cache and this filesystem dump is as almost good as caching in memory. If OS is facing a memory-crunch, the filesystem caches will be discarded and browser doesn't take up scant memory for its own good.

Speed and little more

[gmuslera]

They included how well they ranked in the acid test, but most of the article was about raw speed. But for "best" there are more criteria to take into account. Features, availability of extensions (specially the ones you in particular need), OSs where it runs, security, matters at the moment of making a choice. But at least is a good clue that opera and chrome are usually the fastest ones, safari and firefox aren't so far, and IE is the worst choice is speed is an important factor.

The main debatable test was the specific sites benchmarks one, as it could had measured in good part how much tuned for specific browsers are those sites, but if are the kind of sites you visit more, probably could notice the difference (at least, until that sites acknowledge that worth optimizing for webkit or gecko too).

Interesting... IE sucks... except when it counts.

[dtolman]

IE did best or near best in the web browsing events most users will care about - page load time sfor popular sites like yahoo, facebook, or youtube.

So how does a web browser that apparently sucks at so many theoretical benchmarks, crush the competition in real world load times? Apparently it doesn't matter what you do, if major websites tailor themselves to you.

can someone tell me why

[circletimessquare]

google ripped off simon for its chrome icon?

http://en.wikipedia.org/wiki/Simon_(game)

whenever i see that chrome icon, i want to start pressing the panels before i forget the sequence

Re:So?

[BitZtream]

You are misinformed, I presume you are refering to Firefox, however Chrome and IE both have extensions to do roughly the same thing.

Just because you aren't aware of things outside your viewport of the universe doesn't mean they don't exist.

How much of all these numbers are practical ?

[parallel_prankster]

How do these numbers help me choose a browser ? When on any test - speed or memory - most browsers performed differently on different benchmarks from the same category ? I wish they also explained whether being second in a particular memory test is that bad. Besides, I use a lot of extensions in Mozilla that blocks flash, adblocks etc, how do those affect memory consumption and speed ? I wish there was some more details about these results, especially how the numbers translate to daily use.

won't switch to Chrome yet

[c_jonescc]

I really like Chrome, and according to Tom's numbers it would probably provide a superior browsing experience, aesthetics aside. Yet, I can't make the switch.

I'm addicted to mouse gestures for all my surfing. I switched to Opera way back when, solely for the gestures, and liked it so much I even sent them $20 (paying for a browser!). I switched to Firefox when I learned about the 'All-in-One Gestures' add on.

I'd really like to switch to Chrome, but simply cannot until I find a way to deal with my deep seated gesturing habit. Right-clicking, or moving the mouse arrow to the top left of the screen both seem tedious (which feels really lazy to say), when all I want to do is go back or open a link in a new tab.

Re:won't switch to Chrome yet

[clone53421]

Keyboard shortcuts are far more useful and easy than mouse gestures, IMHO.

Re:won't switch to Chrome yet

[konadelux]

well I do 99% of my surfing with my laptop trackpad, so I'm not sure how well this works, but maybe give it a shot...

Smooth Gestures

Re:won't switch to Chrome yet

[c_jonescc]

oops - I should have reinvestigated this before I posted. There is a gesture add on now.

I don't like keyboard work when casual surfing because I usually end up sitting way back from the keyboard and have a cuppa coffee in my hand.

Re:won't switch to Chrome yet

[gregmac]

I was in the same boat, just found smooth gestures - it works quite well, though I also had to install smooth gestures new tab because otherwise it doesn't work on the "new" tab that shows thumbnails etc.

I've started using Chrome on my laptop now, mostly because Firefox inexplicably started taking up many hundreds of megs of RAM and becoming very slow after a day or two, to the point I had to restart. I have some extensions installed (firebug, firecookie, web developer, gestures) but nothing that I don't have installed on other systems.

Anyways, it's noticeably faster than FF, and so far I like it. No real complaints for my laptop surfing so far (which is reading news, email, and random research/documentation/surfing).

Re:won't switch to Chrome yet

[clone53421]

I’ve actually encountered an unpleasant website that screwed up Firefox permanently. By “unpleasant”, I mean spawning hundreds of new tabs, frozen-browser, rickroll/GNAA-style unpleasant. By “screwed up”, I mean that afterward, every 5 seconds Firefox pegged the CPU at 100% for a second. By “permanently”, I mean the abnormal CPU activity occurred whenever Firefox was running even after restarting both Firefox and the entire computer.

A complete uninstall and reinstall of Firefox brought everything back to normal again. Your experience would probably be no different.

Re:Fuck. That. Multipage shit

[BitZtream]

If you pay for a subscription perhaps you can turn off the ads ... otherwise you are getting the article in exchange for viewing ads.

You of course have the option to not view the ads or click continue in exchange they have the option to not deliver it to you. Simple really, but thanks for making sure everyone knows how you feel, its very important that we continue to get this worthless contributions to the discussion, slashdot could not exist as we know it without them.

Re:So?

[PRMan]

Do tell. Since I've never found a per-site whitelister like NoScript on anything else.

Conclusion

[shutdown -p now]

On the whole, there's really a single practical conclusion from those tests that is useful to a user:

Any browser is fast enough, so long as it's not IE.

The winner, hands down, is ...

[PPH]

... wget.

Real geeks read straight html.

Re:The winner, hands down, is ...

[stinerman]

Pshaw. I telnet to port 80. You and your fancy wget!

Re:The winner, hands down, is ...

[Khashishi]

Bah, I lick the ethernet jack to read high and low voltages, and rub wool on my head to set them. You and your fancy telnet!

Re:The winner, hands down, is ...

[element-o.p.]

Real geeks read straight html.

...in binary.

Re:The winner, hands down, is ...

[RAMMS EIN]

The problem is that straight HTML has become relatively rare. First, it was tag soup with rare pieces of information in between, but at least that could be filtered to find the information. Nowadays, the information often isn't even there until you execute some JavaScript code. Now, I _can_ execute JavaScript code in my mind, but having that interact with the server is a bit too much of a challenge. That leaves me, every text-only browser I know of, every web crawler I know of, and probably loads of other programs in the cold. And often users of browsers other than the latest Firefox or MSIE, too.

As an old-school webmaster, I am sad to see it has come to this. Flashy, interactive pages are nice, but HTML+JavaScript+HTTP is not and has never been the right way to do it. Meanwhile, the information-to-noise ratio has gone down the drain. At least more and more people are breaking away from XMLs grasp and moving to more efficient representations, such as JSON, these days.

Re:Fuck. That. Multipage shit

[clone53421]

I didn’t read any of the ads, until I was reading the comments and finally decided to load it once with adblock turned off to find out why people were complaining about Tom’s taking forever to load.

Re:They all win

[clone53421]

Even Internet Explorer.

Re:If you want a fast web browser...

[thomst]

...block all ads with Privoxy and shut off Javacrap.

And then browse with blazing speed ... the 3 web sites that remain partially functional without Javastuff, that is.

Chrome memory usage

[l00sr]

Once again, calculating Chrome's memory usage is not as simple as summing the memory usage of all its processes, because shared libraries are only loaded once. It's unclear as to whether these benchmarks took this into account. More info here.

I love Chrome, but...

[VoiceInTheDesert]

It still has a lot of compatibility issues. Sites like Youtube, ESPN and others still have bugs in formating and video playback with Flash. I love the speed and it's my primary browser at this point, but if the bugs aren't fixed and a better ad-block isn't added soon, I will be disappointed.

Oblig. Princess Bride quote

[HaZardman27]

You keep using that word. I do not think it means, what you think it means.

Re:Oblig. Princess Bride quote

[freemywrld]

I do not think it means, what you think it means.

I do not think that comma, belongs where you put that comma.

Re:Oblig. Princess Bride quote

[gparent]

That's exactly his point.

Re:Oblig. Princess Bride quote

[HaZardman27]

I realize that; I meant to reply to the GP rather than the parent.

Re:A link to the article would be nice.

[SQLGuru]

It's not like we read the article anyway. I came here looking for the summary posting saying which browser Tom declared the current winner. I must not have read far enough down because it wasn't in the First Post (or the second - aka the first REAL post)......

Trying to read the article with the winner

[griffinme]

I was trying to read the article with Chrome, the eventual winner, and these incredibly annoying ads took up a big chunk of the top left corner. This is with the AdBlock extension for Chrome. The ad was bad enough that it covered several words of the first sentences on each page. I thought that a nice experiment would be to load the page on Firefox. Hey, whadyaknow, the annoying ad is gone in Firefox. To me that says Firefox won the two most important categories. Mem usage and getting rid of annoying ads.

Re:So?

[clone53421]

For very rough values of “roughly the same”.

Anything but IE, and you got quality

[SmallFurryCreature]

Really, all the other browsers are good.

Rough comparison:

• Opera is the most rounded for the tabbed/mouse gesture browser. It takes a major crisis like your hardisk dying for Opera to loose the session, it always keeps the pages you visited when you closed the browser (or it crashed, it does happen sometimes), and that can be very handy if you been tracking some obscure problem and don't want to loose your place.
• Firefox is the most versatile. It can do anything you want it to and then some. For web development it is beyond compare, the best browser ever. The performance can be a bit slow at times and not all the features are terribly well integrated. Its tabbed browsing for instance looses its sessions if the download window was open when you close the browser. "Oops, wasn't the download window your session? No you lame git, it was not." Same with mouse gestures, they work, but not as smooth as Opera. But then, opera doesn't have all those extensions.
• Chrome is fast and with Youtube HTML5 support it makes the site very smooth. But it is also bare bones. I use it as the default "emergency" browser. With all my firefox dev extensions and about a gazillion open pages in Opera, if I need a browser fast or a program insists on opening a page in the default browser, then Chrome is the one. But as said, it has virtually no extensions (yet) and lacks stuff that Opera has offered for years now by default and you can easily enable in Firefox.
• Safari. Well, I got a mac, but frankly, it might or might not be good, but just how many browsers do I need? It is okay, but like many Apple products, it is rather hard on the 'you will use it as we want it' and lacks the features of all the others. Chrome without the google...

Now all we need is for MS to build a browser.

Re:Anything but IE, and you got quality

[parallel_prankster]

IE is not just performance bad, it is also a security nightmare. The only practical result from these numbers is that all other browsers are mostly equal, so stick to what you have if you are comfortable with it.

Paranoia

[MooseMuffin]

I'll admit I'm a bit crazy, but I'm still not comfortable installing anything from google on my machine. Their services are great, but they need to stay up in the cloud and away from my stuff.

Re:So?

[Dracker]

Opera's got it built in!

Preferences - Advanced - Content. Disable Javascript there.
Visit a site you want whitelisted? Right click - Edit site preferences - Scripting - Enable Javascript.

Easy!

WTF?!

[Lisandro]

I can't help to find the testing biased. With lovely tidbits like...

"After reviewing the JavaScript benchmarks, we've decided that Tom's has no choice but to run all of them in the future. While I personally lean toward JSBenchmark, since it isn't affiliated with any browser, its results don't reflect the outcome in Dromaeo. Until the reason for Opera's devastating Mozilla score can be explained, I believe we'll have to run all of them to get the clearest picture. If you disagree, or have an opinion on a better way to benchmark JavaScript, sound off in the comments section below."

Or the conclusions, where out out of 13 categories, Safari won 3, Opera 4, Chrome 3, Firefox 3 and IE only one (shared with FF). Yet, they proclaimed Chrome as the winner. Lovely.

Yes, i'm an Opera fan, but i also like Firefox and Chrome a lot. I'd just like to read fair reviews.

Re:WTF?!

[clone53421]

After reviewing the JavaScript benchmarks, we've decided that Tom's has no choice but to run all of them in the future. While I personally lean toward JSBenchmark, since it isn't affiliated with any browser, its results don't reflect the outcome in Dromaeo. Until the reason for Opera's devastating Mozilla score can be explained, I believe we'll have to run all of them to get the clearest picture.

What’s biased about that?!

If all (or several) of the Javascript benchmarks gave almost exactly the same results, they’d have been able to whittle down the list of benchmarks to test. Since there was a wide variation between the different results for the different benchmarks, they decided to keep all of the benchmarks for future testing, at least until we know why the different tests give different results. That makes perfectly good sense.

Or the conclusions, where out out of 13 categories, Safari won 3, Opera 4, Chrome 3, Firefox 3 and IE only one (shared with FF). Yet, they proclaimed Chrome as the winner. Lovely.

Maybe you should have read the very next paragraph under the list that you cite?

While it may appear that Opera had a better showing than Chrome, it does not reflect how many times that browser was edged out by Chrome when neither placed first. It also does not show that Safari remained smack in the middle throughout or that Firefox had a stranglehold on fourth.

Re:WTF?!

[Lisandro]

I did. Why bother taking numbers if you're just going to give a subjective result at the end?

Re:WTF?!

[clone53421]

It wasn’t subjective. It was based on the full table of results, not the summarized table of first-place results. It’s right there on the results page.

If you award 5 points for each 1st place, 4 for 2nd, 3 for 3rd, 2 for 4th, and 1 point for 5th place, Chrome wins with 88 points, Opera and Safari tie at 83, Firefox comes in with 74, and IE trails behind with 51.

Re:WTF?!

[BZ]

> I can't help to find the testing biased.

It's not biased. Just incompetent. You can tell by the fact that they think the JSBenchmark numbers actually mean anything (which implies they've never read the source for that benchmark), for example.

Re:WTF?!

[clone53421]

I’ve never read the source for that benchmark. Care to enlighten me?

Re:WTF?!

[BZ]

Ah, interesting. The source has actually changed significantly from the last time I looked at this. Or the article author is on crack about what benchmark it is.

I assume they ran http://jsbenchmark.celtickane.com/Run.aspx which actually looks like a legitimate benchmark. Celtic Kane used to push a dom/css/etc benchmark which basically did things like "measure speed of CSS animation" by setting style.x and style.y and the like.

Re:WTF?!

[BZ]

Of course the new benchmark still does things like "run this test for 100ms, take the number of runs, multiply by some magic number (different for each test), then add up all the results and declare it the total score". Which means that while the individual test results might mean something the total score doesn't, much. It's possible to be fast on one thing, really slow on everything else, and trivially win on total score, since it's a straight sum.

tl;dr version

[vivin]

Chrome: 88
Opera: 83
Safari: 83
Firefox: 74
IE: 51

(Scores calculated by using chart on last page)

Firefox works fastest on the iPad

[WillAffleckUW]

But only when I avoid Microsoft-centric sites and sites that try to load insecure scripts.

Re:So?

[Anonymous Coward]

I'm not sure if you are being sarcastic...but if you aren't, the good part of noscript is the ability to block all the javascript that is tracking/adds while allowing the script that allows forms to be filled out and videos to play.

Did anyone actually look at the tests?

[Anonymous Coward]

If you look at the results, they are all pretty similar, except IE. While Opera and Chrome barely win a few more times than Safari and Firefox, the reality is that the results are all largely similar and no one product really is much different than the others in performance. If the tests were weighted differently, or if the analysis used standard deviation instead of place, you'd find no real difference in any of these, again, except IE, which clearly did not fare well in these tests. Just my 2 cents--I don't hate any of them and even IE has people that like it, despite the fact that it is slower.

Re:So?

[Anonymous Coward]

Your right. Opera winds hands down... Oh? You where talking about a plug in that is only supported by Firefox and not built-in? Then all of them support t don't they? Or does Safari not have plug-ins?

Something smells fishy ...

[richtaur]

Internet Explorer is fastest only on Facebook and Yahoo!'s websites (both very popular websites and partners of Microsoft)? This is probably just a coincidence, but I've heard rumors before of MS inserting special code for, say, the Acid3 Test. Probably just paranoia ...

Re:Something smells fishy ...

[Mojo66]

Together with IEs bad performance in synthetic benchmarks, I'd conclude that said sites are heavily optimized for IE.

Speed is a useless test. Try real functionality!

[bradbury]

And by "real" functionality, I mean full support for open standards for video & audio w/o Flash, support for really large long term sessions (dozens of windows, hundreds of tabs, running for weeks) without consuming all the CPU (which Mozilla based browsers tend to do) or memory (which Chrome tends to do) on your machine, robust Javascript, Flash and Ad blocking (to prevent outsiders from using your computer as part of the cloud for their own purposes) and most of all "Green-ness" -- do long running browser sessions, particularly large ones, effectively "degrade" all of the tabs, windows and Javascripts (which have no "current" use because nobody is actually *looking* at them) so their CPU use is minimized and they are paged out? [Thus allowing ones OS to minimize power & memory use and hopefully minimize any data transmission charges (for browsers being used over 3G/4G networks)).

And for the people who don't understand the emphasis in retaining large sessions, you have probably never tried to restart a large session (which can take 10+ minutes of maxed DSL bandwidth and 15-20 minutes of maxed CPU time -- the browser being largely unresponsive during that period). You may have also have never tried to restart an old session only to find that the pages in which you were interested no longer existed. (Yes there are ways around these problems but they require plugins and/or large personal save page storage areas.

Can't see my current favourite there!

[Lemming Mark]

My current favourite is rekonq (http://rekonq.sourceforge.net/), a KDE-native WebKit-based browser. The version I'm compiling from git (they're releasing 0.4.0 soon and it's shaping up well) is looking very nice indeed. If you have the dependencies it needs (recent version of QT needed for plugin support. I'm running with QT 4.6 and KDE 4.4) it's very nice. KDE's web shortcuts work, integrates with KGet, Click To Flash built in, slim UI. I've got nspluginwrapper on this system (my 64-bit Fedora installed it by default) and it isolates the browser from plugin crashes (and I can kill the plugins if they use lots of CPU).

Re:WebKit For The Win

[cadeon]

Way to Apple-ize an opens source project. Yay.

It was started by the KDE team. Sure, Apple grabbed it and did a bunch of work with it, but that does not mean saying it's good is sucking Steve's member.

The browsers that have picked it up show that it's good. Those are of course Safari and Chrome, which is part of the reason why their numbers are awfully close in the comparisons. But it doesn't stop there, WebKit has also been picked up by Epiphany, iCab, OmniWeb, and Uzbl to name a few. WebKit is also likely the most popular rendering engines on mobiles, being on the iPhone, Blackberry and Symbian.

Yes, Google's JS engine is very different and nice. Yes, there are many differences between the browsers above. Even so, one must admit the renderer has a lot to do with the overall responsiveness and quality. And, well, WebKit is the best renderer. It's being adopted by other desktop browsers, being used on many different mobiles, and is consistently very good in the accuracy and speed tests.

Finally, I don't particularly care for Apple. I have one, it's nice, but I'm using my Ubuntu box with Chrome today. So... go do whatever successful trolls do and have a wonderful day.

Re:So?

[Buelldozer]

They didn't test FireFox? Because Chrome 4.x has Adblock, flashblock and noscript functionality is available in the Dev channel. I'm sure it'll be along for "Joe User" shortly.

You're modded to +5 insightful but 2/3rds of your list is wrong!

P.S. No need to take my word for it, see for yourself by going to https://chrome.google.com/extensions and then search on adblock or flashblock.

Please

[killmenow]

Real men telnet to port 443 and do the SSL encryption in their heads.

Tested on Windows 7... now do the same tests

[gosand]

Now that they have this set of tests, do the same thing on WinXP and Vista.

But since I use Linux, I really don't care all that much about this test in the first place. Firefox for me, Opera as a backup, Konqueror bringing up the rear, and XP/IE7 on Virtualbox for those teeth-gritting (but getting more rare) occasions when it is absolutely the last option.

winner: users

[Onymous Coward]

We're just about at the point where's Microsoft's stranglehold on web protocols has been wrenched off. Standards compliance and all around browser functionality and performance are available from a range of great offerings.

I feel comfortable enough with the progress that I think I'll stop flogging Firefox as the wedge to get us to this point. I encourage you to try out whichever standards compliant browsers you're inclined to, and to use the one(s) you prefer for whatever reasons occur to you.

I might discourage using IE, however. There's no guarantee MS won't be able to regrab a browser monopoly and start corrupting protocols again.

Re:So?

[ChronoReverse]

Unfortunately, the way Chrome is designed doesn't yet allow for the adblocking the way it's done in Firefox. If there were a malicious ad, it's still rendered then removed from view with the Chrome solution for instance.

Re:A link to the article would be nice.

[Brian Gordon]

Google Chrome comes out on top and the writer seems to make a good case for it.

The most interesting conclusions seem to be:
-Firefox is the most memory efficient with multiple tabs (!)
-Opera uses a lot of memory
-No browser really has a performance advantage across multiple sites (for example Facebook is really optimized for IE for some reason)
-Even professional writers don't know how to use the word "faze"

Re:Interesting... IE sucks... except when it count

[clone53421]

Yeah. Interesting to see how quickly Facebook would load on Firefox if you spoofed the IE user agent string...

I might have to try that tonight.

Re:So?

[clone53421]

SERIOUSLY? You’re asking someone to set script filtering by editing an .ini?

That’s worse than making them edit their about:config.

Opera winner in my opinion

[BountyX]

I think if these tests were redone with optimized settings in Opera, it would show Opera slightly ahead of Chrome. For example, on page load, Opera is set to draw in one second intervals during load, by default. Obviously for simple sites that load under a second, Opera will preform poorly. Another example is memory usage. They called Opera memory hungry but that is ridiculous. Opera allows you to set the maximum memory limit and is capable of caching to memory. Any unused memory is allocated to cache. You can greatly adjust Opera's memory usage by disabling page caching (set to 1000 by default), lowering maximum memory usage, removing history (set to 5000 pages by default), removing tab thumbnails; however, while Opera's caching may hurt memory usage, it is amazing in the long run. The equvilent to having a local proxy in your browser. Anyways, the bottom line is Opera's default settings are NOT optimal and modifying securtiy, history, memory, caching, DNS prefetching, and the UI can improve its overall preformance. I cannot say the same for chrome where the only optimizations that can be made is DNS prefetching and some security stuff.

Re:Opera winner in my opinion

[ChronoReverse]

Pretty much every single factor you mentioned about Opera is also true of Firefox. You'd need to "optimize" Firefox as well if you decide to run a bench against an "optimized" Opera.

Re:Opera winner in my opinion

[BountyX]

Yes, I agree. Most testing I've seen on browser benchmarks make no effort in providing consistent configurations. I know its hard to control the configurations, but at least *try*. The most configurable browsers (opera and firefox) are modest in their stock deployments, catering more towards security and control rather than speed and optimizations. Chrome barley provides any options, so I presume its optimized by default.

Re:So?

[Joce640k]

Yawn. Call me when they have 3/3.

Re:So?

[Hurricane78]

How roughly? IE definitely has nothing even remotely resembling an ad blocker. And: Can you block a specific HTML element in a page in Chrome? I doubt so.

Also, what is the point of blocking tracking pixels / web beacons, when you are using the browser of the biggest of all centralized data krakens: Google.
If the Google toolbar tracks, so does Chrome.

Endless obsession

[Waccoon]

Enough about JavaScript speed! Could we focus on security, please?

All JS scripts run in a single address space. I'm a bit annoyed that any JavaScript coming from a 3rd party site (namely ads) have the same privileges as JS from the main site. There should be a way to sandbox JavaScript and DOM, or at least configure access privileges. For example, unless configured otherwise, anything in an iframe should be sandboxed, and there should always be some kind of end-user override.

It makes no difference whether you're a provider or an end-user. Since when can we trust advertisers not to abuse privileges? They already siphon usage statistics. When will they start keylogging? So much of the web is "2.0" that just turning JavaScript off is no longer an option.

Even the open source browsers don't seem to care. Are we really going to rely forever on 3rd party extensions to make up for security shortcomings in software products?

Re:Endless obsession

[clone53421]

Um, I don’t think you know what you’re talking about. Javascript running in 3rd party iframes can’t access the parent page.

All I care about...

[wzzzzrd]

...which is the best pr0n viewer?

Re:So?

[Buelldozer]

Call me when FF doesn't consume an obnoxious amount of RAM to open a few pages and doesn't have a screaming hissy about SSL certificates.

Different people, different needs.

Of course it's a lot more fun to act like a snooty teenager, isn't it?

Missing standard daily browsing test

[locopuyo]

What this article fails to test is standard daily browsing. It only testing fresh loads in a clear cache. I visit the same web pages frequently, just like everyone who browses the web. They need to test the speed of the browsers like this. How long does it take to load the same page a day after you went there? How long does it take for a page to load when you hit the back button? What they tested doesn't cover very much of real browser use. Another fun test would be how long it takes users to physically navigate to pages they visit. But it would be much harder to get scientific results like this.

Re:So?

[jwhitener]

Chrome has adblock/flashblock, but lacks noscript right now. That's huge in terms of managing popups/malware imo.

Re:So?

[clone53421]

Yes. Most users don’t want to.

Re:A link to the article would be nice.

[bigngamer92]

No browser really has a performance advantage across multiple sites (for example Facebook is really optimized for IE for some reason)

Because most Facebook users are clueless about Privacy, as well as browsers?

Re:In absolute terms, they're all slow.

[jolson74]

WTF are you talking about? JavaScript is a "hack" but PERL is a "first class language". Please.

If you want to complain about how the DOM implementation in most browsers is a horrible piece of bloat, we can talk. But asserting that simply swapping out one web page scripting syntax for another will:

• magically make things blazingly fast
• make "developing client-side web applications... much more tolerable"

demonstrates a complete lack of understanding.

Re:A link to the article would be nice.

[RAMMS EIN]

``-No browser really has a performance advantage across multiple sites (for example Facebook is really optimized for IE for some reason)''

I think they're doing that right. Facebook really is for the (non-tech-savvy) masses these days, and that's where you will find lots of IE users.

Also, if you want to support all browsers, it makes sense to focus your optimization efforts on IE - optimizing things for other browsers is largely taken care of by the people who make them. So if you optimized for, say, Firefox, you would only be widening the gap. Helping the slowest one along makes for a more consistent experience across browsers.

Really? 4.04?

[Lord Byron II]

Come on. Please tell me that Apple didn't know better than to release Safari 4.04. lol =)

Re:So?

[Goaway]

Popups yes, malware not so much, since Chrome isn't as insecure as Firefox.

Re:In absolute terms, they're all slow.

[Goaway]

Chrome's Javascript is much faster than Perl, Python, Ruby, Tcl or plain Lua. Only LuaJIT beats it.

Re:A link to the article would be nice.

[AmiMoJo]

As usual these results are not that useful. Tom's loves benchmarks but tends to ignore the real world implications.

By far the two biggest factors are your ISP's DNS servers and the web site's ability to generate and serve pages quickly. No browser is going to make Facebook or eBay fast because their servers are just plain slow. So really the fact that one browser renders pages a fraction of a second quicker than another is irrelevant.

Usability and features are going to help you surf faster, but changing to another browser wont. Well, except maybe when it comes to IE.

Chrome is generally very fast for Web 2.0 stuff, i.e. interactive sites. That's about the only major difference, but even then it's marginal as most sites are just not that demanding. If they were then they would be unusable in IE and maybe even Firefox.

It would be more interesting to know which browsers cope well with lower speed CPUs, less RAM and slower HDDs. I.e. which will be fastest on a network or older laptop. Just having low memory usage or fast rendering doesn't tell you much about performance in that kind of environment.

Re:So?

[hkmwbz]

SERIOUSLY? Youre asking someone to set script filtering by editing an .ini?

No, you can edit filters in the content blocker dialog.

Re:A link to the article would be nice.

[hkmwbz]

The memory testing is quite stupid, though. Opera might use more memory over a shorter period of time to make stuff faster, but it evens out over time.

Re:Personal Experience of epSos.de

[clone53421]

Opera has that magic button that can switch off animated gifs

Esc works in Firefox.

and flash, so it is faster by default with no need for third party ad-ons.

I’m pretty sure Flash is a 3rd-party add-on.