Firefox Beta Scores 93 On Acid3 Test 282
CodeShark writes "Mozilla released their latest Firefox 3.X beta today (3.5b4), and increased their score on the Acid 3 test to 93 [on my XP laptop], with tests 70, 71, and tests 75-79 being the final challenges. Curiously though, the current release of the top Acid3 performer — Safari — still not only rates higher (I got scores of 99 once and 100 most of the time) but is usually faster by a little (1.1 sec avg. vs. 1.4 over ten runs apiece) but only because the new Firefox beta was all over the map — frequently better by 25% (.85sec) or tanking badly with rendering times in the 2.5 — 3 second range, and both suffer performance hits on one test (#69)."
Opera 10 as well (Score:5, Informative)
Opera 10alpha is also a 100/100 on the acid 3 since dec 12, 2008
http://www.opera.com/docs/history/
Acid2 already looks fine in Fx 3.0.10 (Score:5, Informative)
How does it rate on Acid 1 & 2, and have the other browsers worked on reaching 100% on the previous tests also, or did they give up on previous tests when the next one was released?
Acid2 already looks fine in the latest general release version of Mozilla Firefox.
Re:Meh. (Score:1, Informative)
96/100 with svg.smil.enabled set to true (Score:5, Informative)
Re:Just a random pet-peeve that came up here -- (Score:3, Informative)
Garbage collection (Score:5, Informative)
Presumably the test should take about the same time to run each time, right?
One of the 100 tests is JavaScript garbage collection. A garbage collector that uses tracing without reference counting isn't necessarily guaranteed to finish in a given amount of time.
Re:Why the variation? (Score:5, Informative)
Under system load, or browser load (such as extra stuff being done in the rendering thread whilst the test is running), a browser may not always pass this test. Whilst its an OK test, there will be no way to reliably pass it 100% of the time, and as CPU's become faster and more efficient, its likely browsers will pass eventually regardless of if they optimise their code or not.
Its also one example of why the ACID tests are quite overrated.
93 on Xp, 91 on RHEL 5 (Score:2, Informative)
Re:Safari and Chrome bound to get better? (Score:5, Informative)
Were adblock and flashblock available for Safari or Chrome (and I believe this is in development for Chrome), and were Chrome available as a Mac version, I would stop using Firefox overnight.
Adblock has been available for Safari for years now. You can get it here:
http://safariadblock.sourceforge.net/ [sourceforge.net]
A Flash block addon for Safari is also available:
http://hetima.com/safari/stand-e.html [hetima.com]
While Safari doesn't have the same ease of plug-in support as Firefox, there's enough for most people who want to make the switch.
Re:Safari and Chrome bound to get better? (Score:3, Informative)
While Safari doesn't have the same ease of plug-in support as Firefox...
It sounds like you're actually thinking of the Firefox "extension" or "add-on" API. Both Safari and Firefox support plugins [apple.com]. Extensions and plugins are not the same thing. This seems to be a common mistake.
Acid1 and Acid2 results (Score:3, Informative)
You can see how well all browsers perform on Acid1 by watching the Acid1 browsershots [browsershots.org].
You can see how well all browsers perform on Acid2 by watching the Acid2 browsershots [browsershots.org] or the Acid2 Wikipedia article [wikipedia.org].
Re:Why the variation? (Score:3, Informative)
I never understood why did they include speed in a browser test?
Because if you can't do it quickly it isn't functional. It's just like specifying video has to play at a given, acceptable frame rate to pass a test that confirms something can play said video. Playing it jerkily in an unwatchable way is not good enough
...but it would mean that the best browsers would fail on a slower computer, and the worst would pass on a faster one. This is not objective.
Which is why the ACID tests each specify reference hardware, like most respectable test suites do. That is objective. Just because you don't have or use that reference hardware and run the test more informally does not mean it is a flaw with the test instead of your procedure.
Not to mention that setting a threshold for speed is impossible. Who says how fast is fast?
The people who write the test pick a minimum acceptable rate and the feature is supposed to be able to function in a timely dependent manner. If it can't it doesn't work with the spec.
Right, which is why the latter is not sufficient for the assigned purpose, just as a Web browser that can render a background color, but it takes four hundred years, isn't in conformance with that specification for all practical purposes. When you're dealing with static elements, slow is pretty relative. When you're dealing with animations, time is a easily defined and crucial part of the spec.