[bbaddict30]

I did a comparison between the html5 features from 2.0 to 2.1 beta using html5test.com and here are the upgrades:
1. Ogg Vorbis audio support
2. Grouping content elements: Reversed attribute on ol element
3. Fields: dirName attribute
4. Form validation
5. XMLHttpRequest Level 2: Document Response Type
6. FileSystem api
7. Page visibility
8. Full Screen video support
9. Window.RequestAnimationFrame

I hope this is useful.

[xsacha]

See:
OS2.0 vs OS2.1 vs BB10

BB10 changes are quite significant including subtitles, x264, mpeg4 and webcam support.

[polytope]

See:
OS2.0 vs OS2.1 vs BB10

BB10 changes are quite significant including subtitles, x264, mpeg4 and webcam support.

These results are things that RIM can brag about.

[bundyo]

Just to note that not all of these newly supported APIs do actually work - for instance FileSystem and full screen don't. Ogg Vorbis support is also misreported:
HTML5 audio test

WebGL on the other hand is twice faster than in 2.0...

[rkennedy01]

So it looks like 2.1 added some functionality, but that we have a lot more to look forward to when BB10 hits the PB. Interesting...and exciting.

RCK

[diegonei]

BB10 results are more than impressive...

[Blackberry_boffin]

RIM has invested a lot of effort and resources into this browser.
It is the current top browser around (and in beta )
I think RIM needs to name the stock BB10 Browser now.
It is more than a significant feature in every way and all significant features need to be know by name, for a better marketing message.
Are you listening Thor?

[lnichols]

RIM has invested a lot of effort and resources into this browser.
It is the current top browser around (and in beta )
I think RIM needs to name the stock BB10 Browser now.
It is more than a significant feature in every way and all significant features need to be know by name, for a better marketing message.
Are you listening Thor?

Other than html5 test the 2.1 browser sucks and is highly unstable. I can't look at crackberry forums without having to close and reopen the browser at least once. Flash seems to kill it quicker than anything. And they better get rid of the checkerboarding with BB10.

[greatwiseone]

The checkerboarding is probably due to performance of javascript in the browser. I think RIM's porting V8 over to the Playbook browser...which should improve performance...

[Double_J75]

The checkerboarding is probably due to performance of javascript in the browser. I think RIM's porting V8 over to the Playbook browser...which should improve performance...

Checkboarding is very frustrating. I would rather have a fast browser with a decent score vs. High html5 score that is slow, and has too much checkerboarding.

The forum pages i go to still look the same regardless.

[xsacha]

The slowdown in the browser is just the javascript engine and the fact that NEON was disabled (due to silicon errata). Since V8 has a tonne of NEON code, it would be beneficial for RIM to use the workaround for the Playbook.

Note: Future devices (Playbook 4G, BB10 devices) won't have this issue.

[lnichols]

The slowdown in the browser is just the javascript engine and the fact that NEON was disabled (due to silicon errata). Since V8 has a tonne of NEON code, it would be beneficial for RIM to use the workaround for the Playbook.

Note: Future devices (Playbook 4G, BB10 devices) won't have this issue.

It's not just slow, it crashes constantly. So based on what you are saying are the current Playbooks going to not get as good support for BB10? I thought that all OMAP 4 chips had NEON support.

[Double_J75]

My browser doesn't crash often. As i mentioned it is slow, and has far too much checkerboarding. The speed does get much better if you disable java.

The other issue is when you click on a link and the browser doesn't respond. If you open a new tab you will just get untitled. Often i have to restart the browser.

[xsacha]

It's not just slow, it crashes constantly. So based on what you are saying are the current Playbooks going to not get as good support for BB10? I thought that all OMAP 4 chips had NEON support.

Technically they do, but the specific one in the Playbook has errata that causes crashes/freezing if you use NEON code -- which is why RIM has disabled NEON system-wide. There are workarounds that RIM can use.

[meltbox360]

The slowdown in the browser is just the javascript engine and the fact that NEON was disabled (due to silicon errata). Since V8 has a tonne of NEON code, it would be beneficial for RIM to use the workaround for the Playbook.

Note: Future devices (Playbook 4G, BB10 devices) won't have this issue.

xsacha are you saying that the ti omap in the playbook is a flawed design that had issues out of the foundry? I'm confused but I am pretty sure that the playbook has no issue running neon instructions but it is entirely possible I missed something.

[VerryBestr]

... V8 has a tonne of NEON code ... .

Would sure like to have more info and/or links for what you wrote.

This post on the ARM blog suggests that V8 does not use NEON. It is over a year old, though, and the situation could be different today, of course. In the comments, the author ("ARM.Martyn") says:

| Google's V8 on ARM: Five Times Better - ARM Community

VFPv3 is being used, though that doesn't imply use of NEON ... NEON provides 32-bit floating point support only, so it's not suited to JavaScript's 64-bit floating point numbers

From what i understand, NEON provides a way to do pipelined or vectorized single precision math -- this would be used, for example, in video codecs, signal processing applications, etc.

[xsacha]

Would sure like to have more info and/or links for what you wrote.

This post on the ARM blog suggests that V8 does not use NEON. It is over a year old, though, and the situation could be different today, of course. In the comments, the author ("ARM.Martyn") says:

| Google's V8 on ARM: Five Times Better - ARM Community

VFPv3 is being used, though that doesn't imply use of NEON ... NEON provides 32-bit floating point support only, so it's not suited to JavaScript's 64-bit floating point numbers

From what i understand, NEON provides a way to do pipelined or vectorized single precision math -- this would be used, for example, in video codecs, signal processing applications, etc.

You're right. I must have been confused with another component of the browser (I'll check up on this - I was just browsing source code and came across a bunch of NEON instructions in multiple files but perhaps it was webkit). V8 is just VFP3.

xsacha are you saying that the ti omap in the playbook is a flawed design that had issues out of the foundry? I'm confused but I am pretty sure that the playbook has no issue running neon instructions but it is entirely possible I missed something.

Yes. It does have an issue running NEON as it can cause freezes/crashes.

[lnichols]

Technically they do, but the specific one in the Playbook has errata that causes crashes/freezing if you use NEON code -- which is why RIM has disabled NEON system-wide. There are workarounds that RIM can use.

Sounds like TI should be replacing the chip in a recall if you have to work around a supported feature that doesn't work.

[xsacha]

Sounds like TI should be replacing the chip in a recall if you have to work around a supported feature that doesn't work.

Silicon errata isn't uncommon. Won't result in a recall :P.
Here's the silicon errata on our OMAP4430 Multimedia Device: http://focus.ti.com/pdfs/wtbu/swpz009D.pdf
Even more can be found in the OMAP4460 on Playbook 4G: http://focus.ti.com/pdfs/wtbu/SWPZ017A.pdf

Tegra 2 had issues too. There was a developer-sided code workaround required on that chip (like errata 657451).

[Double_J75]

I'm a gebroni, all I know is the browser needs to be much quicker than it currently is (with java enabled).

[Double_J75]

Looks like rim is hiring java devolpers. Hopefully it will be to speed up the browser!!

https://rim.taleo.net/careersection/...=en&job=286565

[Ninjatogo]

Looks like rim is hiring java devolpers. Hopefully it will be to speed up the browser!!

https://rim.taleo.net/careersection/...=en&job=286565

Wouldn't they need JavaScript developers to help with this?