RSSFirefox 3.5 Beta 4: Mozilla delivers the speed, as Beta 5 gets under way
By Scott M. Fulton, III | Published April 28, 2009, 11:41 AM
There are now (once again) three simultaneous development tracks for Mozilla's Web browser, as the first public beta of Firefox to be numbered 3.5 has officially hit the streets; the first private Beta 5 of Firefox 3.5 is being distributed to Mozilla testers; and the latest Firefox 3.6 Alpha continues to make headway.
It would appear Mozilla developers learned a lot from last week's code-frozen version of 3.5 Beta 4, as Betanews tests indicate the organization kicked things up a gear. The last code-frozen version before the public build produced a composite performance index score of 9.19 -- that's 919% the performance of Microsoft Internet Explorer 7 (not IE8) in the same system. But after refreshing our test virtual machine with the public Beta 4, that index kicked it above the 10.0 mark to place at 10.44.
How come the final build is faster? Much faster error handling, as shown by the browsers' scores in the Celtic Kane JavaScript test, which indicates that some of the temporary error handling code was likely removed; and generally better performance in all stages of the SunSpider benchmark.
The public Beta 4 actually posted a slightly better performance score than the first nightly build of Beta 5, which posted a 10.33. Beta 5's Acid3 standards compliance score holds even with Beta 4 at 93%, and its Celtic Kane and HowToCreate.uk CSS rendering test scores hold up well against Beta 4. But the SunSpider scores show a general slowdown in most departments by 20 - 25%, which indicates that Beta 5 will be mostly focusing on features and usability rather than functionality.
And though the latest nightly build of Firefox 3.6 Alpha 1 ("Minefield") gained a tick in speed with a 10.25 index score -- compared to last Thursday's 10.14 -- it's been bested by 3.5 Beta 4, thanks in large measure to its improved JavaScript error handling. Despite our early fears that the new TraceMonkey JavaScript interpreter wouldn't perform in the same league as Google Chrome and Apple's Safari 4 beta, Mozilla has been able to dial up the speed considerably, posting scores that are now 33% than in our first tests of Firefox 3.1 Beta 3.
Still, a score above 10.0 puts Firefox in a very competitive position against Chrome, currently the second fastest browser in our tests, but only by a small margin against the Safari 4 beta. Both Safari's and Chrome's index scores of 14.39 and 13.07, respectively [correction], are helped by the fact that they post 100% compliance scores in the Acid3 test.
While all this is going on, Mozilla has now publicly announced the release of Firefox 3.0.10, which may now be automatically downloaded using older versions. We'd been noticing some poor performance since last week's release of 3.0.9, whose reign didn't last so much as one week. Since yesterday, we managed to trace the source of our hanging problems -- thanks in large measure to Mark Russinovich's Process Monitor -- to a corrupted bookmarks file, specifically places.sqlite. Deleting the file and having Firefox rebuild it appears to have solved this problem for now, though we cannot say for certain yet whether Firefox 3.0 or one of our add-ons is responsible for the corruption.
What those numbers tell me is that your test setup is broken. I can assure you that no "error handling" code was removed (or added, or modified) between code freeze and ship. Code freeze means just tha in this caset: code freeze. No more changes. The time lag between that and release is QA and build (signing, setting up the update binary diffs, and so forth). Unless you're using a different definition of "code freeze" than Mozilla itself was using, of course. I'd be interested in the changeset IDs for the two builds you tested (available via about:buildconfig).
Score: 0
|Scott,
Just a note - you list Safari's and Chrome's scores respectively - but backwards. Unless I am reading the image backwards, Safari has the 14.39 score and Chrome has the 13.07 score....
Score: 0
|Been using it for a day and a bit now, it's never crashed once, bloomin' fast, impressed. So it's goodbye Chrome, and hello Firefox 3.5 Beta 4 my new default. This will be my third default browser so far this week, I may be fickle, but I do know a good browser when I see one, and this is a good browser.
Score: 0
|Why in this performance test were not included the finals versions of the browsers Safari 3.2.2 and Opera 9.64?
Score: 2
|hah. It blocks the .net framework assistant that Microsoft added a while back. WTG Mozilla!
Score: 0
|Yes, WTG Mozilla! You've forced me to use a different browser when I want to watch movies on netflix. I think I'll just switch browsers to one that works.
Score: 0
|@mjm01010101 Indeed. Rather foolish.
Score: 0
|Good. I never installed it yet Microsoft background installed their insecure Java clone and disabled the ability to uninstall it; i had to hack the registry to remove it. Typical Microsoft...
Score: 0
|try ie tab add on, it's nifty
Score: 0
|@fatty: I could have sworn Microsoft have stopped making or distributing their old Java VM.
Score: 0
|@Paul - .net is a Java clone, or more specifically J2EE clone. When the old Embrace, Extend, Extinguish play did not work on Java, they created .net as a clone, without the multi-platform support.
Score: 0
|...if you completely ignore the fact that .Net is more functional, more extensible, easier to work with/develop for, and performs exponentially better than Java....
Of course, fatty *always* ignores the facts.
Score: -12
|PC_Tool makes a complete fool of himself/herself yet again....
"..if you completely ignore the fact that .Net is more functional, more extensible, easier to work with/develop for, and performs exponentially better than Java...."
1) .Net is more functional - how do you quantify this? There are a *TON* more frameworks and libraries for Java than all the proprietary .net languages put together.
2) more extensible - Do you even know what that word means? Obviously not. Considering i can download the source to Java, i would say Java is _more_ extensible than .net
3) easier to work with/develop for - depends on the developer. Take a Java developer and drop them in the proprietary .net world and they will have trouble. Take a .net developer and drop them in the Java world and they will have trouble as it is all new to them.
4) and performs exponentially better than Java - if you knew anything about software development you would laugh at that just like everyone else did.
"Of course, PC_Tool does not understand the difference between stupid and clueless...making a fool of himself/herself repeatedly"
fixed that for you...
Score: -1
|That might work on folks who don't know C from perl, but anyone who has developed for more than one language knows you can do far more with fewer resources, less code, with much easier to use tools than you ever could or will be able to with Java.
Hell, most programmers wouldn't even consider the two comparable...two completely different playing fields.
Hacking java source...
*laughing*
Good luck with that. Now try getting whatever you code with that hacked source to work *anywhere* else, genius.
Your stupidity knows no bounds...as usual. Stick to moderation spamming, man...keeps you busy so you don't end up exposing how incredibly ignorant you are.
Score: -14
|Everyone is laughing...here is some more rope....
Score: -1
|Hey, you got that 14th account set up! Congratz on being an almost complete imbecile! (still gotta work on totally disconnecting yourself from reality. Try imagining that Bill Gates is your father...that should really get the drool frothed for ya....)
Score: -1
|"Hey, you got that 14th account set up! " , Can't face the reality you have no idea what you are talking about and even all the paid Microsoft shills which astroturf internet forums can not keep your score up on this one. Everyone is laughing at you.
Score: 1
|"Can't face the reality you have no idea what you are talking about"
Gotta lay off the crack, man. The only one talking out of his a** here is *you*.
"even all the paid Microsoft shills which astroturf internet forums can not keep your score up"
What paid shills? You mean anyone who exposes your ignorance? Those aren't shills, their simply folks with brains. Which also explains why they don't waste their time spam-modding posts of people they don't agree with.
But I'll play some more with your idiotic "j2ee is better than Net" BS...
.Net can be written for using any of 25 supported languages. How many does Java support?
On a test application that both Sun and MSFT participated in,the .Net application was just over 2000 lines of code, the Jave version was over 14,000. The .Net version took 2 weeks, the Java 10. The price/Performance cost ratio for the .Net version: $316, Java: $1305. Maximum Pages served per second: .Net 1400, Java 600, max concurrent users: .Net 6000, Java 4000, max transactions per second: .Net 117, Java 59.
http://www.promoteware.c.../ArticleView.aspx?id=10
Now, accuse the site of being run by a bunch of MSFT shills and be sure to use all 14 accounts to mod this post down as it contains cited facts...which we all know, scares the hell out of you.
In reality, .Net isn't even the best for most uses. It all depends on what you're doing. But hey... you stick with your hacked Java source for everything...let's see how far that gets ya. ;-)
Score: 3
|Hi, I tested acid3 test via http://acid3.acidtests.org/ and Google chrome showed as 79/100 and not as 100/100. My google chrome version is 1.0.154.59 and it shows as up to date. Which version of Google chrome shows 100/100.
Anyway Acid3 test results were impressive with 3.5 Beta4 showed 93/100 compared 3.0.0.10 showed as 71/100. I like the beta version hope they will release the final version soon.
Score: 0
|2.0.174.0 (the latest beta build) is 100/100 but fails the linktest so doesn't pass Acid3.
Score: 0
|Chrome's nightly build of 2.0.177.0 (14573) passed the acid3 test (100/100)
Score: 0
|beta5pre string is only a temporary bump due to the finish of beta4. There will be probably no beta 5, rather, RC1 should be what we are about to see.
Score: 1
|I have been using the pre-betas since 3.1 beta 2 and I can get all of my extensions working just using Nightly tester tools - and I have 54 installed....
Granted there are some that simply will not work, but that is because of a major re-write behind the scenes for extensions developed for Fx 2.x....
Score: 0
|