AMD Releases UVD Video Decode Support For R600 GPUs

[russofris]

So is the end of the big lie? The lie that releasing UVD1 code or specs would somehow violate the MPEGLA license agreement, leak HDCP keys, etc... Or is the lie somehow still going?

If it's over, what's the next big lie?

[chrisr]

The Next Big Lie

If it's over, what's the next big lie?

I think the next "Big Lie" is that users will actually stop moaning, if only AMD would implement "Feature X" in the Open Source drivers.

[russofris]

I think the next "Big Lie" is that users will actually stop moaning, if only AMD would implement "Feature X" in the Open Source drivers.

As long as we're not told that the release of Feature X has been blocked by the legal department due to a scam perpetuated by content rights holders in order to give AMD's US-based competitors a decade long lead in the market.

I'm sure the lawyers will create some new imaginary conflict they can leach money from for another decade.

[pali]

Christian K?nig at ML (http://lists.freedesktop.org/archive...st/066771.html) wrote:

"Let me know if it works for you, cause we don't really have any hardware any more to test it."

[AnonymousCoward]

Thanks Christian and AMD!
Will test it as soon as it lands in Fedora, because I have no time atm to compile.
Great that this is eventuelly implemented. For me I can tell, I am perfectly happy with the radeon driver now.

[kbios]

According to http://lists.freedesktop.org/archive...st/066556.html there is a chance vdpau gl interop (and thus xbmc/kodi) can work on rv770, but not on older chips (anything else including rs880, sigh). Still, great work!

[chrisr]

Hmm, now that's interesting.

According to http://lists.freedesktop.org/archive...st/066556.html there is a chance vdpau gl interop (and thus xbmc/kodi) can work on rv770, but not on older chips (anything else including rs880, sigh). Still, great work!

So UVD 2.0 does support field-based output? Or is field-based output part of the GPU's OpenGL hardware?

[twriter]

When they said "best effort", I think that means "We provide these patches, it may be work in 95 % of the time but if there is bugs, don't expect us to fix them because we don't have time to work on this. Regards".

Best effort simply means we have to prioritize use of our resources. Roughly speaking, our priorities are:

1. specific customer engagements, i.e. working with customers who are building products based on AMD hardware and the open source drivers;
2. newer hardware;
3. older hardware.

I don't think this is terribly surprising. We still get the occasional bug report for R100/R200 and we do investigate. We don't abandon support for older hardware but the majoriy of our development and testing is focused on newer generations (note the "s").

Our open source developers put a lot of effort into supporting the community. Case in point, we got permission for this release on Friday evening, about 6:00 PM Eastern and Christian released it Sunday morning, less than 48hrs later. Thanks Christian!

Tim

[kbios]

So UVD 2.0 does support field-based output? Or is field-based output part of the GPU's OpenGL hardware?

I don't know for sure, but RS880 has UVD 2.0 too, so if it was just that it should work like RV880...

[agd5f]

So UVD 2.0 does support field-based output? Or is field-based output part of the GPU's OpenGL hardware?

Field vs. frame output is a mechanism of the UVD block. The 3D hardware is irrelevant.