Submitted by If the latest reports on IE9 preview are accurate, then this is a very encouraging development: “The fourth and final Platform Preview for Internet Explorer 9 was released today, along with a raft of new HTML5 demos and tests.” Microsoft have reportedly further improved IE9’s performance and standards compliance.“Microsoft is touting the new browser’s substantially improved performance, both relative to its predecessor and, thanks to its hardware acceleration, relative to other HTML5-capable browsers. The new version leapfrogs Safari 5 in the SunSpider benchmark, though still marginally trails behind Chrome 5 and 6, and Opera 10.60. Still, the mere fact that Internet Explorer 9 is competitive is a vast improvement. The browser’s standards conformance has been improved, too. It now earns a score of 95/100 in the Acid3 test.”
This is excellent news. Well done to Microsoft and the IE9 team, since it looks as though when it is released IE9 will at long last catch up significant ground on other browsers.
When it comes out IE will actually score (slightly) better than Firefox in the Acid 3 test. They’ve actually seemed to be focusing on that to.
I bet they’ll probably have SVG Fonts working before them to. Something Robert O’Callahan (Mozilla dev) didn’t want to implement because he liked WOFF better.
I still would rather they use Webkit and do frequent releases while using autoupdate so there isn’t still a bunch of people with IE9 10 years from now.
“Support for SVG Fonts in the web development and font communities has been declining for some time. There’s already been discussion without objection of dropping SVG fonts from the Acid3 test. The community has put forth a proposal in the SVG Working Group to give SVG Fonts optional status.
Instead, developers can use the Web Open Font Format (WOFF, supported in IE9 Platform Preview 3 as well as other browsers) for both HTML and SVG content. It works well in conjunction with the CSS3 Fonts module and has broad support from leading font vendors (e.g. here, “a majority of font makers have already settled on WOFF or services like Typekit as their format of choiceâ€). WOFF fonts are a better long-term solution for many reasons discussed previously.
Similarly, support for SMIL animation of SVG in the web development community is far from strong. The leader of the SVG standardization effort wrote that not supporting SMIL in its current state is probably best “since the SVG WG intends to coordinate with the CSS WG to make some changes to animation and to extend filters.†There’s already work started to reconcile CSS3 animations and SVG. Developers interested in animating SVG can use JavaScript, as the samples in the test drive site do today, with consistent results.”
http://blogs.msdn.com/b/ie/archive/2010/08/04/html5-modernized-four…
If we are going to compare a future release of IE (IE9) with Firefox, then we should compare it with the release of Firefox that will come out at about the same time (which will be Firefox 4).
This IE9 preview scores 95 on the acid3 test. Currently, the Firefox 4 preview scores 97.
http://ourlan.homelinux.net/qdig/?Qwd=./KDE4_desktop&Qif=minefield_…
So I’m not sure exactly how you came up with the notion that IE9 would score better than Firefox, but I’m not sure that you are correct.
However, having said all that, a score of 95 for acid3 tests is pretty good compatibility with standards, and Microsoft should be commended for it.
Edited 2010-08-05 22:53 UTC
(disclaimer: Mozilla employee here. Doesn’t make me an expert but might bias me)
Totally agree with the parent answer, IE9’s score should be compared to FF4’s score, 97%.
The remaining 3% is just SVG fonts.
The reason why Firefox doesn’t support SVG fonts is that Mozilla people don’t implement broken specs just for the sake of passing the new cool Acid test. Many believe that SVG fonts are essentially useless, see
https://bugzilla.mozilla.org/show_bug.cgi?id=119490#c38
Quote: “By far the biggest reason to support SVG fonts is to pass Acid3. I don’t think it’s a good idea to add an otherwise-unnecessary feature just to pass a test.”
While other people at Mozilla point out that the SVG fonts spec has major technical shortcomings rendering it useless for many scripts, see
https://bugzilla.mozilla.org/show_bug.cgi?id=119490#c49
Quote: “Text shaping is the process of converting Unicode character into positioned glyphs. SVG fonts do not support complex text rendering which is an absolute requirement for supporting scripts like Arabic and Indic. SVG fonts are just a toy in that regard. You can’t do serious work with them.”
The Mozilla people are wrong, SVG Fonts are not a broken spec. They work fine in Inkscape and Webkit-based browsers (which will also support WOFF and not pick one over the other).
If the tools are already here then it’s important regardless of it’s shortcomings. This is exactly the kind of thing that makes people rely on Plugins over Browser API’s.
@Superb
That sucks! I use Inkscape d****t! And WOFF (AFAIK) is not available in it.
@lemur2
I was going by release date actually. Since Firefox 4 is delayed until next year and Internet Explorer will come out in September passing FF 3.6 on the Acid test.
But apparently IE won’t implement SVG Fonts either so they won’t ever score 100. So nevermind, I guess Firefox is saved from the embarrassment of being beat by IE of all browsers in standards compliance.
Not according to Mozilla:
https://wiki.mozilla.org/Firefox/4/Beta#Milestones
Beta release 7 is October 1, 2010.
Final freeze is October 15, 2010.
The plan seems to be for a late October, early November release date for Firefox 4.
That is this year. Firefox 4 is an entirely comparable release to IE9.
Edited 2010-08-06 01:57 UTC
Even so, for a short while it’ll pass Firefox. Come on, let them have at least that!
Sorry, but no. The IE9 release is not coming in September, that’s when they’re releasing the first beta. IIRC, IE8 spent > 6 months during its beta cycle, although I’m willing to believe IE9 might go faster what with the alpha previews they’ve been releasing.
I’d be really shocked if IE9 came out before FF4.0 did.
Edit: Although I have to admit that I’m already shocked to learn that IE9 is scoring 95 on Acid3. I never expected that to happen so soon, and it’s a promising sign from MS.
Edited 2010-08-06 03:50 UTC
They will be duly sent a cake!
http://www.openbuddha.com/2008/06/17/ie-sends-mozilla-a-new-cake-fo…
Oh. All this time I thought it was still coming out next year. Cool. Well it looks like I clearly jumped the gun here.
Still it would’ve been funny if Internet Explorer beat Firefox in a Acid 3 test because of some stand against SVG Fonts. But IE developers are doing the same thing now.
What’s the point in Web Standards organisations if everyone ignores the recommendations?
Now that is just mean. And I thought IE fanboy-ism went out of fashion around the time IE6 came out
Microsoft admitting in print and with charts just how badly one of their products, and a well-known one at that, SUCKED, compared to the competition.
I guess they’ve come a long way from the days when Bill Gates implied that Windows 98 was a superior OS to Sun Solaris.
Microsoft should dump the “Internet Explorer” name and go with something else. I like the promise of how much better IE9 sounds but honestly I don’t see myself using a browser offered by Microsoft for one reason.
Microsoft has proven it doesn’t understand the consumer cloud and the fact their user base is still so large it lets them dictate web standards which they can control to slow down things like CSS3 and HTML5 that compete with their products like Silverlight.
Microsoft doens’t have to “understand”, they just need to follow standars.
Yeah, Microsoft is really slowing down HTML5/CSS3 (both which have large portions which are not done, have ambiguous descriptions, and in the case of SVG Fonts, are useless completely .. ) by actually bringing HTML5/CSS3 into the 21st century with hardware acceleration.
HTML5 (using this term broadly but mainly Canvas and SVG) wasn’t even in the same ballpark as Silverlight performance wise before IE9 came into the picture. So saying that they retarded progress for the sake of protecting Silverlight (which is more than a web technology, and has a broader scope and application than a web technology) is stupid.
Before IE9 it was a race to the bottom to implement these features as quickly as possible, with no regard to the fact that they performed like shit.
Sigh!
Why do Windows people always claim that Windows is first and best with everything?
IE9 is not released yet. We are still in the timeframe that can be described as “before IE9”.
If we are taliking about hardware acceleration for browsers, Firefox 4 has this feature also. Firefox 4 is also not released yet. Firefox 4 hardware-accelerated-rendering is slightly faster than IE9 hardware-accelerated-rendering.
The truly pertinent point to note, however, is that hardware-accelerated-browser-rendering performance is, however, mostly determined by …
wait for it …
hardware.
Edited 2010-08-06 12:02 UTC
Because the Direct2D acceleration in Firefox 4 is a reactionary move to IE’s announcement. The guy doing it is really good, and I don’t mean to take away from what he’s done, but honestly, it started off as a science project.
No it is not. No idea where you get this from. I get 15 more frames per second with FirefoxB2 vs IE9 PP4.
The reason Firefox will never be as fast as straight up Direct2D is that there is a significant amount of abstraction in the Firefox graphics stack (only recently added because of the need to accelerate across all platforms .. but it perplexes me that they added this kind of composition despite the adverse effects it’d have on their existing D2D acceleration, the reasoning they gave is bullshit too, getting over limitations in cairo. Makes no sense to me abstract things an additional 2-3 levels instead of just contributing to cairo, but okay.
Only after a certain threshold is passed. There is certainly a great deal of room for optimization using any graphics API. If what you say were true, every game would run at a smooth 60 FPS just because the hardware is identical. Obviously there are other factors at play, and this is the case in Firefox Beta4.
However, all of this is not relevant. My point was that IE9 cannot be seen as holding back the very same web standards it modernized by accelerating them into levels which can move past a few frames per second.
It started a long while ago.
https://bugzilla.mozilla.org/show_bug.cgi?id=527707
http://www.basschouten.com/blog1.php/2010/01/18/layers-cross-platfo…
http://blog.mozilla.com/joe/2010/05/25/hardware-accelerating-firefo…
Mozilla “Bug” 527707: Investigate Direct2D rendering backend. Reported: 2009-11-10. Well before IE9 announced it. Who exactly is reacting to whom?
http://www.neowin.net/news/html5-speed-test-firefox-37-narrowly-bea…
“In a four-way HTML5 speed test between IE9, Firefox 3.7, Chrome 6, and an unspecified version of Opera (probably the latest), Microsoft’s latest platform preview held its own. In fact, it was only narrowly beaten out by Firefox (by about 5%).”
Oh dear.
My point was that it wasn’t IE9 that did the hardware acceleration in the first place. IE9 using hardware acceleration was … a reactionary move to Mozilla’s developments in this area starting circa November 2009.
Edited 2010-08-06 13:34 UTC
That’s interesting, because IE9 was shown off with Direct2D acceleration November 18th (requiring an obvious incubation period which meant that the D2D stuff was landed much earlier than that)
I saw this, and the test is flawed in that GPU resources on Vista/7 are scheduled when they’re in use by multiple sources. You’re not getting a true comparison until you compare them alone.
Additionally, they both share a similar bottleneck regarding the amount of Fish on screen. At lower numbers I’ve gotten twice the FPS of Firefox using IE.
This is also replicated across other tests (Amazon Shelf, the Dice test is not even a contest for IE9)
And your point has been thoroughly debunked.
I don’t think so.
Direct2D hardware acceleration for Firefox was last (because it is Windows only). Hardware acceleration for Firefox was first implemented as OpenGL, so that it would work on all platforms that Firefox supports.
The Direct2D stuff is Johnny-come-lately.
Yep, Cairo was always meant to be used for a cross platform acceleration platform. The problem was that windows lacked a decent acceleration API until Direct2D came out in Windows 7, and once that happened Firefox started looking into it almost immediately. No surprise there.
Edited 2010-08-08 05:32 UTC
Exactly. Only Windows of all platforms has poor support for OpenGL.
Regardless, the innovation of support for hardware acceleration of rendering by a browser, regardless of the graphics API, was first implemented in Firefox.
Except that the Glitz cairo backend was only ever enabled on Firefox for Windows. Not to mention, it’s unsupported, and unmaintained, and has no support in recent versions of Cairo, and has been marked for removal from the Mozilla source tree.
Also, it’s not so much hardware accelerated as it is doing software rasterization, and uploading the final texture to OpenGL. You get OpenGL API manipulation of the texture, but the drawing itself is not hardware accelerated. Only the post processing effects are.
Direct2D offers things like hardware tesselation, font rendering acceleration, etc. things that are next to impossible to do with any degree of accuracy on OpenGL.
Cairo does not have an OpenGL backend, this is precisely why Layers was devised.
Don’t let facts get in the way of your mindless IE bashing, despite the fact that you’ve been proven wrong already with your claim that IE was beat to market by Direct2D, and that IE performs worse than Firefox does.
If you bothered to do some research, no I don’t mean skimming your favorite search engine for the first few hits, but actual research, reading of mailing list, and if you actually followed this stuff, then you’d know just how wrong your assertions are.
Retained Layers replaces their illfated compositor in Firefox, which was meant to remedy the fact that Cairo has no OpenGL renderer. Let me repeat, cairo has no OpenGL renderer.
Direct2D in Cairo (the excellent work of Bas), is the only competent hardware accelerated cairo backend.
Jesus christ, this is not hard.
Doesn’t Cairo accelerate through XRender by default on Linux, and CoreAnimation (or something) on OSX? Perhaps those are not quite as good as D2D (I really don’t know the details) but they are certainly far better than the old GDI on windows and provide at least a degree of hardware acceleration.
So the point remains, Firefox already was trying to use hw acceleration on platforms where it could. As soon as Windows 7 was released with a good API to do so in Windows, Firefox attempted to do the same there. So i’m not sure where you get off claiming they are only following in IE’s footsteps.
So if anything, it seems to me like Windows was following OSX and Linux by finally providing a decent 2D acceleration API, and then the IE team followed Firefox by actually attempting to use it in a browser.
Edited 2010-08-08 23:58 UTC
Your final point is not quite accurate. direct2d is a good acceleration api. Xrender is not. The sheer fact that many (most?) common usages barring font rendering are slower on xrender than software.
XRender is not even on the same planet as Direct2D. I’m unaware that they offload remotely as much to the GPU. Full disclosure though, I don’t know much about it. What I have been able to find, doesn’t seem to say nice things about it. Words I’ve heard thrown around are “terrible and “disaster”
The Quartz situation I do know more about, OSX by default gets free 2D compositing, this is not an immediate mode hw accelerated api. The hw accelerated stuff is opt-in.
In fact, if you look at Mozilla’s bug tracker, you can see them complaining about a lack of QuartzGL (aka Quartz2D Extreme .. not to be confused with Quartz2D which Core Graphics layers ontop of), and mentioning that it is something that they would like to do in the future.
I don’t think this point was made, all things considered. You have XRender, but there’s really no comparison. Unless you’re getting any kind of smooth framerate on IE’s test suites.
Maybe that was too hasty of me, I’ll concede that point. The overreaching point remains though, their hardware accelerated situation on anything other than Windows is terrible. Even their Windows situation could use substantial improvement. IE flies in this regard.
No. Just no. Microsoft has always had hardware accelerated drawing APIs, they just have not either been public or good. Look at GDI, DirectUI, and now Direct2D.
The Linux situation always has been, and will be for the foreseeable future, terrible.
The OSX situation seems to be disabled by default because the implementation is so bug ridden.
Flies in the face of that whole, IE used GDI prior to using Direct2D, which was hardware accelerated on Windows XP and Windows 7.
It’s the same situation as XRender though, GDI is. It’s hardware accelerated, if you want to call it that, but they’re both pieces of shit.
Man do you ever have it backwards. It is way more important to actually support the standards than it is to wring some extra performance out of them. In that respect, IE9 is finally catching up to the status quo after years and years of being an embarrassment. Canvas is really not that important to users compared to all the other standards they screw up.
So one irrelevant technology wasn’t as fast as another irrelevant technology? Who cares. Everyone doing the kinds of things that required this kind of functionality is using flash anyway.
Hilarious, given the only browser actually performing like shit was IE. Everyone else was doing just fine.
You can do both, considering they are different parts of the pipeline. While everyone laser focused on Javascript microbenchmarks, the total story of performance was largely forgotten.
I think they caught up and surpassed in a lot of respects, and have some to go yet in others. Every engine has rendering quirks. Don’t get me started on Gecko or Webkit deviations from standards.
Hell, the ambiguity of those standards led to the same implementation looking different across Gecko, WebKit, and IE. Who’s right?
Canvas is only one aspect. There’s also SVG, and the animating of properties on elements in general. Any kind of modern animation is dreadfully slow on any non-hw accelerated rendering engine.
Look at the test suite for non canvas examples of how pitiful performance on other browsers is.
Explains Chrome getting 1FPS and Firefox (pre D2D) getting 1-5FPS. Open your eyes, and see the progress that has taken place over the last few months, across not only IE, but other browsers, and largely thanks to IE.
I am just waiting for the day ie15 comes out and people still will complain that the latest whizbang html5 thingy also has to run on their by then 15 year old ie6
I don’t care so much about sunspider or any other synthetic test. Way more important is the UI responsiveness, an area where IE has been hideous so far. I have some low-powered Atom hardware at work, and with IE8 it takes several seconds just to open a new tab. Chrome and Firefox are essentially instant. So hopefully they’ve worked on their UI responsiveness and not just the rendering engine.
Does it really matter at this point if browser x can score 10% better on a synthetic javascript benchmark than browser y?
Google Docs is as heavy as AJAX sites get and it works just fine in IE8. Sure it loads faster in Chrome but the difference is minimal.
This is getting to be like a hard drive benchmark that measures how fast 1mb can be read. ZOMG new fangled hard drive can load 1mb 10% faster so it is the hard drive king! Of course to the user the difference is imperceptible but that doesn’t make for good headlines.
Well, maybe it doesn’t matter right now, but if this javascript/SVG performance allows HTML5 to replace flash in the future, it will matters.
Maybe one day we’ll be able to write most of the flash games using standard technology.
Of course we aren’t quite there yet.
You’re more bottlenecked by the actual speed of Canvas, SVG, et all in that situation. Which IE9 is excellent at due to hardware acceleration.
I still think a combinations of all these web technologies are awkward to use, but different strokes for different folks.
Microsoft really has done something incredible starting with IE8. They did more in two releases to turn their browser around than others have done in many releases. It’s a monumental effort, and one they deserve to be commended for.
Of course, it’s what everyone was saying was the case since when IE6 was out, it was the most standards compliant browser. IE6’s issue was that Microsoft ignored the browser, and it lived way too long.
That’s it. There was no neglecting of standards, it was simply them taking their eye off of the ball. Now look at them, they’ve done things that no one thought they would, simply because no one bothers to read up on history.
And to think, a few months ago some idiots were calling on them to ditch Trident for WebKit.
Wow, you mean after years of doing absolutely nothing they finally threw resources at the problem and managed to sort of catch up to other browsers in some aspects? Yes, very impressive.
And we’d be far better off for it. One less rendering engine to support. It’s not like IE has any unique features.
Yes, exactly. It is impressive if you consider the incredible gap they had to close, and the amount of time in which they did it.
Yeah, maybe you don’t see the problem in having a lack of competition in the rendering engine space, but others do.
of course you’re going to be developing faster especially if you’re following the wake of those who came before.
Particularly pertinent as other implementations are wide open to all. No saying that microsoft is copying code but I bet that they are getting inspiration from gecko and webkit…
Well, the problem is the ambiguity of the standards. Along with implementing said standards, they’ve submitted back hundreds of test cases and worked with the W3C to help define some of the undefined behaviors.
So it’s not like their job was any easy.
It’s not about whether IE8 can render the sites that exist. It’s about the sites that don’t exist because IE8 can’t render them.
My point isn’t specific to IE8.
I find lauding Javascript results from other browsers to be just as silly.
We already hit diminishing returns a few iterations ago. There is not going to be an AJAX website that works fine in Chrome 5 but not 3. Just because browser X can process javascript 10% faster does not mean there is an advantage for users and developers.
The last month number — I studied — told about 60% of the users still use XP. So the question is: Does it run on XP ?
pica
No, it’s Vista SP2 / Windows 7 only. It uses Direct2D and DirectWrite for all rendering, which is how it gets hardware rendering acceleration, but those APIs only work on Windows Vista / Windows 7.
So, IE8 is the last version that will be available on Windows XP.
…and got 91/100 with IE9 latest preview.
I’ve tried to submit the comment bellow to that blog entry, didn’t pass the microsoft censorship.
PLEASE MICROSOFT SUPPORT ISO STANDARDS
In this case ISO-8859-16, which was ratified in 2001.
Here are the results of a Romanian pangram web page encoded as ISO-8859-16 http://bit.ly/b6YRTo :
Internet Explorer Technical Preview 4 – http://bit.ly/b4hs6J
Internet Explorer 8 – http://bit.ly/a3fyKl
Firefox 3.6.8 – http://bit.ly/aDk120
Opera 10.60 – http://bit.ly/a1qQ30
Safari 5.0.1 – http://bit.ly/bmIwvY
Google Chrome 5.0.375 – http://bit.ly/9UDE6f
I’ve tested also the full ISO-8859-16 charset web page
http://www.secarica.ro/test/iso-8859-16.htm
Internet Explorer Technical Preview 4 – http://bit.ly/9iuBvM
Internet Explorer 8 – http://bit.ly/9w1kRg
Firefox 3.6.8 – http://bit.ly/9puJ1n
Opera 10.60 – http://bit.ly/dA2E3a
Safari 5.0.1 – http://bit.ly/98nwvY
Google Chrome 5.0.375 – http://bit.ly/dD9eJm
ISO-8859-16 is supported by ALL the above web browsers
except Internet Explorer.
I have reported this issue through Microsoft’s Connect
service and the bug was closed because it was “by design”. In order to see (and vote) the bug one needs to follow the steps:
1. go to https://connect.microsoft.com/directory/
2. create microsoft connect account, even though you have a live account
3. click “apply” for “Internet Explorer Feedback Program”
4. go to https://connect.microsoft.com/IE/feedback/details/554861/iso-8859-16…
5. (optional) vote by clicking “I can too”
ISO-8859-16 is important because “Romanian Top Level Domain”
(http://www.rotld.ro/engleza/index_en.htm) refuses to allow proper Romanian web sites names (including the five special Romanian characters) until all major browser vendors (operating systems) have support for
iso-8859-16.
PLEASE MICROSOFT SUPPORT ISO STANDARDS