XBMC Gains Crystal HD 1080p Decoding Support

[Ant P.]

How is Intel's open decoding solution any different from AMD's open decoding solution? Both use shaders. And if you you are using the broadcom for decoding, then it doesn't matter whose GPU you use, so why would Intel be any better choice than AMD? The AMD chip would give you better 3D in addition.

But the Intel setup is more likely to provide battery life long enough to watch the entire video.

[gbeauche]

Nevertheless i?ll keep my fingers cross for VDPAU in fglrx :-)

That won't change anything. Use the VAAPI driver, that's the best solution you can get so far and for the time being...

[dosenpfand]

fglrx supports UVD via VA-API.

Yes i know , i was talking about a "decent" support, i mean that it should be at least as good as "Pure Video" and VDPAU, actually

But ok i know that Linux isnt the "target market" and its nice that Ati pays some people for developing a decent FLOSS driver.

Nevertheless i?ll keep my fingers cross for VDPAU in fglrx :-)

Gr??e

[agd5f]

I really understand why ATI cant release specs for the UVD part, at least for now, but it would have been really nice to get a decent H.264 decoding with the fglrx, and in the best case via VDPAU.

fglrx supports UVD via VA-API.

[dosenpfand]

I really understand why ATI cant release specs for the UVD part, at least for now, but it would have been really nice to get a decent H.264 decoding with the fglrx, and in the best case via VDPAU.

Annyways, not does it matter but the fact Ati released the specs, was a reason for me to become Ati customer again, alltough fglrx is still behind Nvidias binary Linux driver in some aspects.

Well this broadcomm chip could be nice when low power consumption is needed, or maybe to give an old PC HTPC qualitys, but every modern,semi modern Desktop-PC doesnt have that much trouble in decoding the diffrent HD Codecs, even if most stuff has to be done by the CPU.

Gr??e

[some-guy]

Right. However chip design starts years in advance of the release so lots of this stuff was started years ago. We are attempting to make future hardware friendlier to open source, but it may take a while to actually see the results.

So am I right in assuming that means 3-5 years minimum?

[agd5f]

Near as I can tell, the idea was to make Intel that open source HD decoding solution, or did I miss something? Seems rather pointless to buy an AMD w/UVD and pair it with Broadcom.

How is Intel's open decoding solution any different from AMD's open decoding solution? Both use shaders. And if you you are using the broadcom for decoding, then it doesn't matter whose GPU you use, so why would Intel be any better choice than AMD? The AMD chip would give you better 3D in addition.

[Kjella]

I'm not really sure how we lose business by not opening up UVD. If you plan to use the broadcom chip, you still need a GPU to drive your display. AFAIK, there aren't really any other open source options out there at the moment. OTOH, we would really lose business if opening UVD somehow allowed bluray, etc. to be hacked.

Near as I can tell, the idea was to make Intel that open source HD decoding solution, or did I miss something? Seems rather pointless to buy an AMD w/UVD and pair it with Broadcom.

Sadly, I'm sure AMD would be slammed by all their nsaty DRM agreements. AACS is broken, BD+ is broken, HDCP is broken, BluRay can barely get more broken but the movie business is in denial.

[agd5f]

> I'm not really sure how we lose business by not opening up UVD.
> If you plan to use the broadcom chip, you still need a GPU to drive your display.

ATI has been documenting 3D, but there are plenty of people that do
not need or want 3D.

You may not want OpenGL, but newer chips only have a 3D engine and it's used for everything (2D, 3D, compute, buffer moves, etc.), so you need it for everything except software-only framebuffer.

With the broadcom chip all I need is a framebuffer, correct?
And pretty much any GPU has a FLOSS driver that can do
framebuffer, correct?

If the broadcom chip only accelerates decode, you still need driver code to handle rendering (scaling, colorspace conversion) and post processing (filters, brightness, sharpness, gamma, etc.). In which case you still need the 3D engine for that part.

The obvious solution is to seperate the digital restrictions mangling
from the video decoding in UVD3, and document the video decoding.

Right. However chip design starts years in advance of the release so lots of this stuff was started years ago. We are attempting to make future hardware friendlier to open source, but it may take a while to actually see the results.

[Dieter]

> I'm not really sure how we lose business by not opening up UVD.
> If you plan to use the broadcom chip, you still need a GPU to drive your display.

ATI has been documenting 3D, but there are plenty of people that do
not need or want 3D.

With the broadcom chip all I need is a framebuffer, correct?
And pretty much any GPU has a FLOSS driver that can do
framebuffer, correct?

The obvious solution is to seperate the digital restrictions mangling
from the video decoding in UVD3, and document the video decoding.