-=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- (c) WidthPadding Industries 1987 0|144|0 -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=- -=+=-
SoCoder -> Link Home -> Videos


 
Spare
Created : 25 June 2010
 

HTML 5 browser speedtest



https://www.youtube.com/watch?v=OIZUdZdFzOo
In this video you can see how IE, Chrome, Firefox and Opera perform on hardware acceleration in HTML 5.

 

Comments


Friday, 25 June 2010, 06:18
JL235
He concluded that FireFox was the HTML 5 champion, but I disagree. Under high stresses it was only marginally faster then IE 9 (just 1fps difference) yet under low stresses IE 9 was consistently at 60fps whilst FF was wandering between 50 to 60. Most apps won't stress the browser, so consistentcy is better!

I'm really surprised tho at the real improvement with IE. Now both JavaScript and rendering are comparable with other browsers. If only they would fix up the UI then it'd be sold on me. Although I suspect all of the issues with Opera (and especially Chrome) will be sorted by the time those versions come out of development.
Friday, 25 June 2010, 06:41
Jayenkai
That's the main thing, though, isn't it. They're REALLY trying here, but for some smegged up reason the big "IE9" is still really really barebones!
Quite why they can't do what every other browser's doing (Slap a quick address bar, add a little back+forward button, job done) is beyond me.
If they released this, right now, people might give a damn.. (But, you know, with a giant "unsecure/incomplete!!!" sticker on it!)
Friday, 25 June 2010, 08:13
CodersRule
Remember also, that this is Microsoft's HTML5 test. I'm not saying it's not accurate, but they might have fudged it a bit to make it work well with IE.
Friday, 25 June 2010, 08:14
Stealth
This is a terrible test. It's forcing all the browsers to compete on the same machine for limited resources. They really aren't isolating the experiment variable (the browser) here.

Another big factor is that if you look at the URL of the browsers, it says "ie.microsoft.com". This test is hosted by them, so of course it's going to run great on IE 9.

It's all a marketing game.
Friday, 25 June 2010, 08:16
Jayenkai
So, make your own independent test, and get it out there.. You'll probably get a fair number of visitors, for something like that!
Friday, 25 June 2010, 08:17
CodersRule
A link to the test if you're curious is here:
ie.microsoft.com/testdrive/Performance/FishIE%20tank/Default.html

Also; I trust Mozilla's tests, and I see these:
https://wiki.mozilla.org/Platform/GFX/Direct2DDemo
people.mozilla.com/~vladimir/demos/photos.svg (particularly noticable when sizing up the photos)
srufaculty.sru.edu/david.dailey/svg/balloon.svg
srufaculty.sru.edu/david.dailey/svg/clipdrag12.svg
maps.google.com/(zoom in and out once, to make sure that images are cached, and then see the smooth zoom in/out animation in the D2D build)

Friday, 25 June 2010, 08:48
JL235
Jayenkai That's the main thing, though, isn't it. They're REALLY trying here, but for some smegged up reason the big "IE9" is still really really barebones!
Quite why they can't do what every other browser's doing (Slap a quick address bar, add a little back+forward button, job done) is beyond me.
If they released this, right now, people might give a damn.. (But, you know, with a giant "unsecure/incomplete!!!" sticker on it!)
The IE9 previews are previewing Trident, not the rest of IE. I'm hoping this means they are secretly working on a huge IE UI overhaul, perhaps using the ribbon?

Stealth This is a terrible test. It's forcing all the browsers to compete on the same machine for limited resources. They really aren't isolating the experiment variable (the browser) here.
and as the user showed IE 9 and FireFox was using less those limited resources (namely the CPU) then the others, yet were faster. All of the browsers were limited, there was no bias given to any of them.

On my PC I'll never run just a browser with no other apps open; and even then within the browser will be Flash/Java/Silverlight/etc using more resources. The browser will always be competing to some degree. So the test seems entirely fair to me.
Friday, 25 June 2010, 08:50
CodersRule
On my PC I'll never run just a browser with no other apps open;


That's true; but to do a comparison, you need to be able to actually compare them. You ain't comparing if they're competing for resources.
Friday, 25 June 2010, 10:07
JL235
But as long as no browser is given higher priority or any other bias then they are all evenly competing for resources. In which case the test is fair.
Friday, 25 June 2010, 10:16
Stealth
That's really hard to prove.
Friday, 25 June 2010, 10:38
JL235
Now your just being paranoid.
Friday, 25 June 2010, 11:11
Stealth
Webkit is a screaming fast rendering engine. I seriously doubt that Trident has not only fixed it's original speed issues, but superseded Webkit.

What I do think, is that Microsoft got Trident to do a particular task very well (more so than other engines), and made a test demoing that. I'm interested in the overall speed of the rendering engine, not controlled tests. I wouldn't be so hard on Microsoft, except that they've done this in the past. They make the new versions of IE out to groundbreaking (remember, IE 8 was suppose to pass the acid 3 test?).

Microsoft blew my trust. However, if by some crazy engineering achievement they did create a faster engine than webkit, I'll eat my words.
Friday, 25 June 2010, 11:50
JL235
This isn't Trident vs WebKit, it's Trident vs WebKit vs Gecko vs Presto. I expect that yes the MS test used was built to show IE 9 in a magnificant limelight. However under stress IE 9 didn't come out on top, FireFox did. Opera also gave far better performance then Chrome. If WebKit is so much better then why did Chrome easily come last?

If IE 9 was shown on it's own to be lighting fast whilst the others were really slow then I'd agree the test could be biased. But that's not what happened and with three browsers beating WebKit I find it hard to believe it's faster.
Friday, 25 June 2010, 12:14
Stealth
Webkit is the fastest rendering engine available. It clearly fails at this task, but across the board it's the clear winner.