RSSApple Safari 4 beta raises the bar for speed, compliance
By Scott M. Fulton, III | Published February 25, 2009, 11:11 PM
We've been hearing quite a lot from every browser manufacturer, including Microsoft and Mozilla, about the incredible speed increases that eventually, pretty soon now, right around the corner, will be realized the moment one of them bites the bullet and installs a new, faster JavaScript interpreter. Well, consider the bullet officially bitten. Betanews tests of Apple's new beta of Safari for Windows, using a freshly cleaned Windows Vista SP1 virtual machine "white box," demonstrates significant speed improvements even over previous Safari versions, which were already quite fast.
What's more, the Safari 4 beta (reviewed by Tim Conneally Wednesday afternoon) scored a perfect 100% on the Web Standards Project's Acid3 test -- the industry's most respected indicator of whether a browser renders text the way the engineers of HTML, CSS, and JavaScript intend it to.
While Safari is officially the world's #3 Web browser, according to the latest data from analytics and research firm NetApplications, that's mainly because of two reasons: It's the Web browser for the iPhone, and it's the principal browser for the Mac. Its Windows user base is relatively negligible. But that might change if Apple were suddenly to emerge with a value proposition that distinguishes it from the rest of the pack, even from Firefox. If the code of Safari Beta 4 stays stable through its testing period, Apple may very well have one.
In our initial tests this afternoon, we pitted the latest Safari 4 beta for Windows against the latest production versions (not betas) of the other four major Windows-based browsers: Internet Explorer 7 (fully updated), Firefox 3.06, Opera 9.63, and the most recently updated Google Chrome. We gave each browser the same battery of tests: a CSS rendering benchmark designed by the British Web design training firm HowToCreate; a JavaScript speed test by Irish Web developer Sean Patrick Kane; and the familiar Acid3 standards test (which measures compliance, not speed).
A virtual machine is a handicap in itself for benchmarking purposes, but what's important is that its an equivalent handicap for each of the contenders. So what matters is the relative speed of the browsers with respect to one another; the proportion of how much faster one is than the other will probably hold reasonably true in any production environment. The host system in our test uses a 2.4 GHz Intel Core 2 Quad Q6600 with 3 GB of DRAM.
While Opera 9.63 put in a good CSS rendering score of 10 microseconds (ms), after a 250 ms load time (the interval spent in preparing JavaScript to run), Safari 4 blew right past Opera with a 7 ms render time and a 54 ms load time. This compared to 26 ms render time and 177 ms load time for Google Chrome, 61 ms render time and 400 ms load time for Firefox 3.0.6, and a lumbering 131 ms render time and 555 ms load time for IE7. That means IE7 was 17 times slower than Safari at CSS rendering.
In the S.P. Kane test, which uses a battery of common JavaScript tasks executed in sequence, Safari also blew past its competitors with a score of 174 ms, versus 348 ms for Chrome, 381 ms for Opera, 499 ms for Firefox, and a thumb-twiddling 1533 ms (a second and a half) for IE7. If you can imagine the sound of CBS' famous 60 Minutes stopwatch in your mind, Safari can complete a battery of JavaScript instructions in the interval between two of those ticks, while you could count eight of them before IE7 got its work done.
While Firefox 3.1 is due to include the TraceMonkey JavaScript interpreter, whose initial internal tests are said to reveal orders of magnitude greater speed than with the 3.0 series, no public beta has been released yet with TraceMonkey included. Now may be a good time.
Finally, in the all-important Acid3 test, yes, Safari 4 turned in a perfect 100% score. This compared to an 85% for Opera, 79% for Chrome, 71% for Firefox, and an appreciably pitiful 12% for IE7.
There is a clear-cut reason -- at least one in the works -- for Windows users to have a browser made by Apple. But as with other Apple software (QuickTime and iTunes), regardless of whether the user chooses not to have Apple's automatic updater during the setup routine, once again, it gets installed anyway.
Apple Software Update showed up in our list of uninstallable programs, even though we explicitly requested it be omitted. Just as an experiment, we wanted to see whether Apple's setup routine would recognize Software Update was already present during a separate Safari installation process -- in other words, would it still give you the option to not have automatic updating, even if you didn't request it in the first place, if it's already installed. Indeed, Apple's setup program omits the option to disinclude Apple Software Update if it's installed -- that's not to say it "greys" the option, it leaves it off. So Apple clearly knows what it's doing, and why.
It would be easier to trust Apple to do a better job with its Web browser if it would have its software do what we lowly Windows users ask it to do.
Despite that, Apple's early victory could bode well for the company at a very opportune time: If the European Commission proceeds as it has threatened to do, Microsoft could be compelled, at least insofar as Europe is concerned, to offer Windows 7 customers the opportunity to install each of the major alternative browsers instead of Internet Explorer 8. If that's the case, Microsoft may find itself at least linking to, if not directly including, a major piece of Apple software with its operating system. And that means Apple Software Update could be installed by those who choose Safari...and that means they could wake up one day, as many already have, and find something new and unexpected called MobileMe.
Chrome is considerably quicker on many tests performed. These results are a bit...."off" or there's something externally influencing them somehow. Several websites prove this.
Safari is pretty though.
Score: 0
|Guys it's clear, a new King of browsers has emerged. Safari 4 rips every other browser out there a new one. If you thought Safari 3 was scary fast, Safari 4 is terrifying fast. While Top Sites and Tabs at the top aren't new, Apple just has a way of implementing this so much better.
Isn't it just amazing at how everything Apple touches it turns to gold? How does this company constantly innovate and deliver top quality after top quality product? Safari 4 is all the rage. Here is my short YouTube review: http://www.youtube.com/watch?v=jJk9hpbJ3cA
Score: -9
|Uh, ms is the unit for milliseconds, not microseconds. But it's no big deal - you're only off by a factor of 1000 :-)
(Seriously, do you guys proofread at all?)
Score: 2
|paul thurrott says it best on his site, with his review of safari 4 :)
Score: 3
|I'm sure a guy who makes a living by reviewing and praising all things Microsoft is a totally legit and unbiased source for reviewing Apple's Safari 4 browser which just by the way, Apple happens to be Microsoft's biggest direct and most vocal competitor? O_o
What else would you expect a Wintard to say?
Score: -6
|It's ugly as crap on Windows. Why can't Apple make good looking software for Windows?
Score: 8
|This guys sums up my feelings very well.
http://www.winsupersite.com/alt/safari4.asp
Score: 5
|Agreed. Summed it up pretty well.
And those tests were pointless as stated by others.
Score: 3
|LOL. Like anything actually looks good on Windoze. And I suppose IE7 or IE8's bloated and handicap large buttons is "good looking software"? frankwick, your just too funny.
Score: -5
|Guys, don't take it so negatively.
Yes, the test setup is not ideal, but it should send a message to other competitors: "beat me". We, as consumers will only win.
Don't forget that the speed and compliance to the standards is only one side of the coin. There are many other factors that you have to consider: security, extensibility, stability, etc... I am wondering how Safari competes in those categories.
As a web developer, I really hope that all the new versions of the browsers will have much greater level of compatibility than they currently have. Multi-browser support is a huge project time killer.
Score: 1
|"Multi-browser support is a huge project time killer."
You got that right.
How many times have you designed a site, thought "wonderful, I'm finished" looked at it in IE and wanted to tear your face off in rage?
Score: 0
|Lol. All of my sites broke in Safari 4... =*(
Score: 0
|Yeah i know. But since Microsoft has a monopoly, you have to spend days porting your standards compliant web site to the horribly broken Microsoft browser. This will never change. Microsoft is moving on....they want to turn the web into one giant proprietary, DRM infected, Silverturd
Score: 0
|Sure it's fast, but speed isn't the only thing that makes some browser a good browser.
As far as usability and speed together are concerned, Opera and Firefox are still the leaders.
I just can't iamgine browsing without AdBlock, bookmarks synchronization and mouse gestures. GMail Notifier is just a cherry on top...
As for the everything else, classic Apple BS PR crap served as best thing after sliced white bread.
Even though more than half of stuff is nothing special or worth mentioning. Uuuu 150 new features.
Sorry, i could only notice like 3 or 4 of them compared to Safari 3...
Score: 1
|I'm afraid you need to read more carefully (I thought the same at first). It's 150 features. Most of them aren't new. It doesn't call them all new features.
However, it does list a few of them twice, which is blatant cheating to make it a round number.
Score: 1
|"Finally, in the all-important Acid3 test, yes, Safari 4 turned in a perfect 100% score. This compared to an 85% for Opera, 79% for Chrome, 71% for Firefox, and an appreciably pitiful 12% for IE7."
*Shakes head* Comparing this test on a beta product with non-beta products is a waste of time.
Opera 10 gets a 100% too.
It's what Apple themselves seem to have done and it's outright ridiculous to the point of being rude.
Score: 2
|Opera was first to score 100 % Yes it would seem Safari is .30 ms faster than Opera, but who's going to notice besides Data on Star Trek??
Opera 10 =
Failed 0 tests.
Test 26 passed, but took 82ms (less than 30fps)
Test 69 passed, but took 24 attempts (less than perfect).
Total elapsed time: 1.32s
Safari 4
Failed 0 Tests
Test 03 passed, but took 42ms
Test 26 Passed, but took 35ms
Test 65 Passed, but took 127ms
Test 75 passed, but took 51ms
Test 79 passed but took 46 ms
total elapsed time 1.03 s
So technically Safari 4 is a smidgen faster than Opera 10. But how many tests are on there?? Opera is slow on 1 test and less than perfect on a second test. Safari is slow on 5 tests. Opera was slowed down on the bases on 1 test. Test 26 vs Safari. That and it seems to suck my memory big time. So does Opera if I have too many windows open, but on one window Safari will do it. :( Just another browser taking up space. Would prefer Chrome. Now thats fast. But because it's stripped down.
Score: 0
|as @Makaveli213 said, this test is flawed
Score: 2
|"In our initial tests this afternoon, we pitted the latest Safari 4 beta for Windows against the latest production versions (not betas) of the other four major Windows-based browsers: Internet Explorer 7 (fully updated), Firefox 3.06, Opera 9.63, and the most recently updated Google Chrome."
Now wait a minute. Why use the latest version of Chrome but not the latest version of IE, Firefox or Opera?
If you want the test to be fully fair download Chrome from their site and install it. Do not update it. That doesnt make it fair.
Then you would have to use IE8 RC1, Firefox 3.1 Beta 2 and Opera 10 Build 1285.
In which case your point about Acid3. Null. Opera 10 has had that since its first Alpha Release. Back in December of 2008.
All the updated browser include a update javascript engine. If you want to pit this Beta up against other browsers. Pit it up against the Beta versions. Not the final. Cause you are then testing the newest product against the release versions.
It would be like pitting Opera 10/Firefox 3/IE8 against Safari 3. It wouldnt be fair. Cause they are not the same.
Make your tests more reliable. Pitting a Beta product that is not final against a final product that is not as up to date as the Beta is not accurate results. It is just to boost the overall appearance of the Beta.
Score: 12
|Exactly, flawed tests return flawed (or biased) scores... Shame on Scott.
Scott, please rectify the tests using the browsers latest versions or we have to think your articles are really biased.
Score: 0
|More useless, featureless crap from Apple.
Score: 0
|Yep, and all that useless, featureless crap will soon be copied by Microsoft, like they try to do with everything else Apple makes.
Score: -4
|Actually, I see Apple put quicktabs into safari. Wasn't quicktabs an IE convention?
Score: 1
|Maybe if the ads on BetaNews didn't take forever to load, my browser would be considered fast too. Seriously! BetaNews - Whoever is serving your ads is slowing down the loading of the pages!!!
Score: 0
|What ads?
Score: 2
|Solution:
Firefox w/ Adblock Plus... or Privoxy.
Score: 0
|.... or Maxthon.
Score: 0
|stability >>>>> speed
Score: 0
|...
"...no public beta [of Firefox] has been released yet with TraceMonkey included."
TraceMonkey is available in FireFox 3.1 Beta 2 and is enabled by defaul. In 3.1b1 it was disabled but could be enabled through about:config.
And really, a comment page where I have to register? 1000 other people likely didn't point your error out because they don't care enough to go through the hoops. Get rid of this.
Score: 0
|Forums or comment pages that don't require registration are extremely rare, as you well know.
Score: 0
|Exactly... or BetaNews could go the "Anonymous Coward" route that has made other sites such a pleasant experience. ;-)
Score: 0
|