[adadadad]

just checked 'html5test.com' . its really nice to see that bb10 scores gigantic 484+10 bonus points. Though, honestly I want bb10 browser to be the best, but has this score got anything to do with overall user experience? absolutely not.! Playbook browser also scores 411+9 bonus points over current OS version, which is higher than 'Opera Mobile' ,'Dolphin HD' , iOS safari....... While I must admit playbook browser is very good, but it is nowhere close to the browser mentioned above ( in terms of smoothness, performance....). there is no denying in this. it has got some issues like it closes unexpectedly (with 400+ MB of free ram available. similarly, I want a browser, in which I don't need to turn off javascript to attain smoothness. turning it off makes it barely usable. And I forgot to mention that damn checkerboarding. most annoying thing. And its not that I'm android , blackberry or iOS fanboy. I have a Playbook and a bold 9930. Since I own RIM products, I expect from it. and I want it to get better.

[togardergrosse]

I agreed with you.
The reason why PB browser isn't smooth probably because it's actually an app made with Adobe Air. Not a native code or something.

[diegonei]

The experience isn't bad at all. The issue right now seems to be JavaScript bogging down the loading speeds but besides that, the PlayBook Browser is a beast.

Just try loading a site with JavaScript disabled and you'll see what I mean.

About AIR... It doesn't help, but did you know the BB10 Browser is coded in HTML5? Divede by zero or what?..

[adadadad]

The experience isn't bad at all. The issue right now seems to be JavaScript bogging down the loading speeds but besides that, the PlayBook Browser is a beat.

Just try loading a site with JavaScript disabled and you'll see what I mean.

About AIR... It doesn't help, but did you know the BB10 Browser is coded in HTML5? Divede by zero or what?..

ya I've noticed that turning off javascript makes it smooth. alongside, it makes it useless too. what is javascript for, if I need to turn it off??? other browsers are super smooth even on heavy javascript pages. I just hope, BB10 browser lives up to all the expectations and hype. and I want it to happen. a loyal bb user.

[adadadad]

I agreed with you.
The reason why PB browser isn't smooth probably because it's actually an app made with Adobe Air. Not a native code or something.

perhaps you are right. but no one cares, how an app is made. performance is all, they care about.

[Kyle27]

My Dev PlayBook running OS 10.0.9.388 on top of OS 2.1 scores 484 + 11 bonus points.

[adadadad]

My Dev PlayBook running OS 10.0.9.388 on top of OS 2.1 scores 484 + 11 bonus points.

its the html5 score of bb10.

[diegonei]

Well, we're telling you why the browser has the issues it does. You're free to do whatever you want with the info (ranting is always an option). At least now you know that performance and HTML5 compatibility are nowhere related.

As for other browsers on other platforms doing better, some can handle JS better (desktop ones for sure). Mobile... That will hit and miss as far as I know. Mum's Defy doesn't do much better speed-wise.

But yeah, should get better with BB10.

[notfanboy]

It's one thing to have a high compliance score, but what's more important is whether the feature is stable and performs well. I stumbled on this interesting site: Microsoft HTML5 Test Drive, which contains a ton of performance benchmarks and cool looking demos.

Site Map

I was only able to try a few of these tests, but the PB browser didn't perform very well on a couple.

FishIE Tank - what's the frame rate with 100 fish?
HTML5 Fish Bowl - the page loaded, but this test wouldn't start for some reason
Chalkboard HTML5 Benchmark - tried this a couple of times, and it crashed the browser both times

Could be something wrong with the Playbook I was testing on. Can someone try these out?

It's worth noting that while the iPad2 only scores 344 points on the html5test.com website, it performs very well on the Microsoft tests.

[samab]

It's one thing to have a high compliance score, but what's more important is whether the feature is stable and performs well. I stumbled on this interesting site: Microsoft HTML5 Test Drive, which contains a ton of performance benchmarks and cool looking demos.

It's worth noting that while the iPad2 only scores 344 points on the html5test.com website, it performs very well on the Microsoft tests.

That's because those tests are benchmarking the GPU's, not really benchmarking the browser itself.

The Playbook has a single core SGX540 vs. the iPad 2 has a dual core SGX543. Not only is the iPad 2's GPU core more advanced, it has twice as many GPU cores. You are talking about the iPad 2 having 2.5-3.0x more GPU power than the Playbook. Even if you put iOS on the Playbook --- you are still going to have a dismal frame rate for 100 fish in the fishie tank test.

[notfanboy]

Were you able to run the other two tests? Framerate aside, they should at least finish, no? Or at least not crash the browser.

[mikeo007]

That's because those tests are benchmarking the GPU's, not really benchmarking the browser itself.

The Playbook has a single core SGX540 vs. the iPad 2 has a dual core SGX543. Not only is the iPad 2's GPU core more advanced, it has twice as many GPU cores. You are talking about the iPad 2 having 2.5-3.0x more GPU power than the Playbook. Even if you put iOS on the Playbook --- you are still going to have a dismal frame rate for 100 fish in the fishie tank test.

I don't think this is true. My iPod touch has a worse GPU than the Playbook, yet it runs 20 fish at 30fps while the PB bounces around 2-3 fps.

[samab]

Were you able to run the other two tests? Framerate aside, they should at least finish, no? Or at least not crash the browser.

Desktop browsers like Chrome also crashed on those tests.

Opera 12 review - Performance and Verdict | ITProPortal.com

[notfanboy]

Desktop browsers like Chrome also crashed on those tests.

Opera 12 review - Performance and Verdict | ITProPortal.com

Those test ran fine on all the devices I have on hand at the moment: iPad/iOS5, Samsung Galaxy 3/Android 4, Chrome/Win 7, and Safari/MacOS Lion

The Chalkboard test was one of the tougher ones. It crashed the Playbook browser, and it wouldn't finish on Chrome/MacOS. Interestingly the mobile devices performed better than the desktops, but the SG3 smoked everyone on this test.

I guess the point I'm asking here is: how trustworthy is that html5test number when if a browser can't execute some actual html5 tests?

Edit: From the link you posted: "HTML5Test.com isn't the last word though. It doesn't actually determine whether the HTML5 functions are correctly implemented, it merely tests for their presence."

[samab]

I don't think this is true. My iPod touch has a worse GPU than the Playbook, yet it runs 20 fish at 30fps while the PB bounces around 2-3 fps.

Those test ran fine on all the devices I have on hand at the moment: iPad/iOS5, Samsung Galaxy 3/Android 4, Chrome/Win 7, and Safari/MacOS Lion

The Chalkboard test was one of the tougher ones. It crashed the Playbook browser, and it wouldn't finish on Chrome/MacOS. Interestingly the mobile devices performed better than the desktops, but the SG3 smoked everyone on this test.

I guess the point I'm asking here is: how trustworthy is that html5test number when if a browser can't execute some actual html5 tests?

There are a million methods to render the fish bowl test or the chalkboard test by code.

Microsoft doesn't implement webgl at all --- they think it is a security vulnerability. Playbook's browser flies closer to the sun than anyone in terms of webgl implementation (the Playbook even gives you a security warning).

So you have 2 extreme ways to write software codes to render the fish bowl or the chalkboard test If you write codes in the most extreme way to implement webgl in every aspect of the fishbowl rendering test or the chalkboard test --- the Playbook will be fast. But the test is written by Microsoft, so we are talking about the other extreme where not a single line of webgl code exists in the test --- the Playbook will be slow or even crash.

Chrome and the Playbook crash --- because there are newer and better ways to render the chalkboard if they want to --- via webgl (which Microsoft doesn't use).

[notfanboy]

Chrome and the Playbook crash --- because there are newer and better ways to render the chalkboard if they want to --- via webgl (which Microsoft doesn't use).

Sorry man, I don't buy this reasoning at all. The PB browser crashes on an HTML5 test, and this is fine because it could have done the test using another technology? The point is about HTML5 compliance, not about how many other ways a graphics effect can be implemented.

[mikeo007]

There are a million methods to render the fish bowl test or the chalkboard test by code.

Microsoft doesn't implement webgl at all --- they think it is a security vulnerability. Playbook's browser flies closer to the sun than anyone in terms of webgl implementation (the Playbook even gives you a security warning).

So you have 2 extreme ways to write software codes to render the fish bowl or the chalkboard test If you write codes in the most extreme way to implement webgl in every aspect of the fishbowl rendering test or the chalkboard test --- the Playbook will be fast. But the test is written by Microsoft, so we are talking about the other extreme where not a single line of webgl code exists in the test --- the Playbook will be slow or even crash.

Chrome and the Playbook crash --- because there are newer and better ways to render the chalkboard if they want to --- via webgl (which Microsoft doesn't use).

That's an interesting take on it. But then it would invalidate your previous point that the test solely tests the GPU, since without WebGL, there is no GPU offloading at all. Also, just because the tests are implemented differently, doesn't mean they're done wrong.

I'd be interested to see the fish test on a BB10 dev alpha, I suspect the performance will be much better.

[samab]

Sorry man, I don't buy this reasoning at all. The PB browser crashes on an HTML5 test, and this is fine because it could have done the test using another technology? The point is about HTML5 compliance, not about how many other ways a graphics effect can be implemented.

There is no such thing as what most people refer to HTML5 --- it is a COLLECTION of various emerging technologies. The actual HTML5 spec won't be out in another 4 years.

[samab]

That's an interesting take on it. But then it would invalidate your previous point that the test solely tests the GPU, since without WebGL, there is no GPU offloading at all. Also, just because the tests are implemented differently, doesn't mean they're done wrong.

I'd be interested to see the fish test on a BB10 dev alpha, I suspect the performance will be much better.

There are many different ways to render things with GPU hardware acceleration --- via different api's (to use webgl or not) or via 2D or via 3D. Microsoft doesn't use webgl but it has other ways to implement hardware acceleration within ie10.

You can play a movie on your PC and let it just run in 2D hardware accelerated mode in your GPU's 2D pipeline. Or you can play a movie on your PC and let it render in the GPU's 3D pipeline. If you have an old PC, you basically have 2 choices --- your graphics card has hardware accelerated dvd decoding in the 2D pipeline (but you can't do special effects like sepia on your movie). Or you can decode your dvd in software but have the ability to have hardware accelerated sepia effects render in your graphics card. In both cases, certain portion of the rendering is done via hardware acceleration in the graphics card.

[notfanboy]

I don't think this is true. My iPod touch has a worse GPU than the Playbook, yet it runs 20 fish at 30fps while the PB bounces around 2-3 fps.

Here are some results:

FishIE Tank at 100 fish
Macbook Air 59 fps
iPad 53 fps
SG3 45-50 fps

Chalkboard test
Macbook stopped test at 180 seconds
iPad 50 seconds
SG3 18 seconds

[samab]

Here is the fishie tank test --- rewritten with webgl by a software engineer from Mozilla.

Jeff Muizelaar: Drawing Sprites: Canvas 2D vs. WebGL

[notfanboy]

I suspect the OS7 browser would have a higher fps than PB on the fishie test. I'll try this out later.

[Andrew4life]

I tried running the Microsoft FishIE Benchmark and got a frame rate of around 3fps with 20 fish. Pretty bad.

I then tried to run the webGL version that was ported by over by this person: Jeff Muizelaar: Drawing Sprites: Canvas 2D vs. WebGL
FishIE Tank

And I got these errors.


It ran much faster rendering black boxes. lol

For the webgl version, firefox was able to run it at around 20fps(but only after I had manually turned on webgl), opera couldn't run it giving me a webgl initilization error, and IE couldn't run it either.

[diegonei]

The curious thing is... I remember taking the fish tank test before (2.0) and it performed really well. What on Earth happened that broke this down?

[samab]

I tried running the Microsoft FishIE Benchmark and got a frame rate of around 3fps with 20 fish. Pretty bad.

I then tried to run the webGL version that was ported by over by this person: Jeff Muizelaar: Drawing Sprites: Canvas 2D vs. WebGL
FishIE Tank

And I got these errors.


It ran much faster rendering black boxes. lol

For the webgl version, firefox was able to run it at around 20fps(but only after I had manually turned on webgl), opera couldn't run it giving me a webgl initilization error, and IE couldn't run it either.

As Jeff Muizelaar himself stated in the website I cited --- the black box is caused by the very large texture size of 8192x1024. Whereas the SGX540 GPU inside the Playbook has a maximum texture size of 2048x2048.

http://muizelaar.blogspot.ca/2011/02...11995656332010