The "2900 XT Lacks UVD" Thread

[Arun]

Please use this thread to discuss the lack of the UVD and AVP video blocks in the R600, aka Radeon HD 2900 XT.

The previous short thread on this subject has been locked to create this official one, and the truckload of posts in the RV610/RV630 thread have been moved to a new thread and closed. As such, this is the only thread where this discussion can/should continue.

Please do so in a civil manner, however. This is not the "AMD is doomed, their marketing department needs a kick in the nuts and I wish Hector Ruiz fell off a bridge for ruining the company I invested my life savings in" thread, thank you very much. Not that there is such a thread on the forums, of course, but there are more than enough non-trollish discussions of AMD's prospects in various other threads already.

And as a final and more personal note, maybe it's just because I'm not from the USA, but I really don't get this lawsuit stuff - although it's still worth a good laugh, I guess!

 

[nicolasb]

The Inquirer has some AMD presentation slide images that (in its opinion) caused the confusion, plus a number of references to reputable on-line sales sites that advertised 2900XT cards as having UVD (because they believed the AIB makers, who in turn believed AMD).

http://www.theinquirer.net/default.aspx?article=39884

The headline is "AMD is to blame for 2900XT/UVD fiasco".

So, after seeing suspicious slides, what can we conclude? While it is certain that several partners and media got their assumptions wrong, there is no other than DAAMIT to blame for the confusion. This confusion is going on in the channel, among partners and press for over half a year now. All of the delays and confusion can be blamed to pathetic power-games between AMD and ATI people....

All of the statements made by partners were confirmed by AMD, thus Sapphire or HIS did not work on their own, by blindingly following the documents and quotes they had in hand - this messy situation was caused. It could have been easily avoided, but even yours truly would leave Tunis with impression that R600 has UVD, given the fact that Xilleon chip is integrated inside the chip. Sadly, this was not case with UVD.

 

[Silent_Buddha]

Interesting, the UVD: Decode Comparison would be the most damning. Although ATI can still save face if they can get the HD 2900 XT to do that Bitstream processing/Entropy Decode done on chip in the shaders.

From the slide it seems to be implying that the entire process is done on chip without having the CPU do any video decode.

It'll be interesting to see if this proves to be the case when finalized drivers for video decode are finally out. Or if we'll have a repeat of the Nvidia Fiasco where video decode just didn't work.

Regards,
SB

 

[AlexV]

I`m not sure they can do that. Entropy decoding is one of the most unfriendly unwieldy things, and I think an UVD including chip has dedicated hardware for that.

 

[Geeforcer]

The whole thing is not really that hard to resolve: All AMD and/or partners need to do is issue a statement: "There has been some confusion, this is what it actually does, if you are not satisfied please return for the full refund." End of story.

 

[Kanyamagufa]

I may be mistaken, but from what I've read I think the main source confusion for people is that they believe they have been conned out of functionality with the 2900 XT because of it's lack of UVD.

But this isn't the case, as the 2900 XT is able to perform all the of the same video decoding as the 2600 and 2400, just without the reduction in CPU utilization. There's no difference in the resulting image, all that is different between the cards is how that result is achieved.

 

[AlexV]

You`re expecting ppl to be fairly reasonable...that`s seldom the case.

 

[dizietsma]

Strange that nvidia did not "correct" the websites assumptions with this and those websites had to stumble upon it themselves / in conversation / reading these threads ....

In the old days nvidia would have been first to put people right if it showed the competition in a bad light. Maybe they did, or maybe they are just getting like the fat cat?

 

[neliz]

My 2900 has UVD, end of story...

 

[_xxx_]

The whole thing is not really that hard to resolve: All AMD and/or partners need to do is issue a statement: "There has been some confusion, this is what it actually does, if you are not satisfied please return for the full refund." End of story.

That's a pipe dream, though

Looking back at NV and their pure video fiasco, that was strikingly similar to this. What a coincidence.

 

[santiagodraco]

Are links to the "original" thread not allowed or is that to be allowed to die and hide away in forum oblivion?

I can understand creation of a thread more specific to the feature (obviously the other thread wasn't for this topic) but locking it also makes the discussion, and ATI comments via Dave, "history".

I don't want to link it here myself fearing some retribution from the forum gods

 

[neliz]

Are links to the "original" thread not allowed or is that to be allowed to die and hide away in forum oblivion?

I can understand creation of a thread more specific to the feature (obviously the other thread wasn't for this topic) but locking it also makes the discussion, and ATI comments via Dave, "history".

I don't want to link it here myself fearing some retribution from the forum gods

Lol, the other thread isn't locked so this post about "forum-nazi's" is pretty pathetic.

 

[Arun]

Actually, the "Previous 2900 XT Lacks UVD" thread, which represents all the posts in the RV610/RV630 thread about UVD, was locked - but that was only to make sure there wouldn't be two concurrent threads with people replying in both. Feel free to quote stuff from there, as long as you link to the post so that people can always see the full context if there is any... Not sure why we'd want to prevent that either, that'd be a bit weird to say the least!

 

[CarstenS]

I've already postet my view on this matter as a news item on our website:
http://www.pcgameshardware.de/?article_id=602231
(google translation into english: http://translate.google.com/transla...rticle_id=602231&langpair=de|en&hl=de&ie=UTF8)

Basically, if there's any confusion about this matter it's becaue "press" did not pay attention during the launch - for example in Tunis. Because in the presentation as well as in the "public" (journalists only of course) Q&A afterwards this matter was mentioned and AMD clearly stated, R600 does not have the dedicated silicon known as UVD in HD2600/2400.

 

[neliz]

Basically, if there's any confusion about this matter it's becaue "press" did not pay attention during the launch - for example in Tunis. Because in the presentation as well as in the "public" (journalists only of course) Q&A afterwards this matter was mentioned and AMD clearly stated, R600 does not have the dedicated silicon known as UVD in HD2600/2400.

So true, but don't forget AMD helps in the confusion by using the same TLA in another context when it describes it's products.

 

[AlexV]

There are some strings in the registry named, suggestively, HWUVD_DisableH264 and HWUVD_DisableVC1, with my HD2900XT and both the drivers off of ATis site and the latest leaked ones. What the strings do or do not do is anybody`s guess though.

 

[Geeforcer]

I've already postet my view on this matter as a news item on our website:
http://www.pcgameshardware.de/?article_id=602231
(google translation into english: http://translate.google.com/transla...rticle_id=602231&langpair=de|en&hl=de&ie=UTF8)

Basically, if there's any confusion about this matter it's becaue "press" did not pay attention during the launch - for example in Tunis. Because in the presentation as well as in the "public" (journalists only of course) Q&A afterwards this matter was mentioned and AMD clearly stated, R600 does not have the dedicated silicon known as UVD in HD2600/2400.

What about partners with "UVD" on their boxes? Were they not paying attention either?

 

[CarstenS]

I don't know about that - and when partners were briefed about possible features that could make a sticker on the boxes.

 

[santiagodraco]

Actually, the "Previous 2900 XT Lacks UVD" thread, which represents all the posts in the RV610/RV630 thread about UVD, was locked - but that was only to make sure there wouldn't be two concurrent threads with people replying in both. Feel free to quote stuff from there, as long as you link to the post so that people can always see the full context if there is any... Not sure why we'd want to prevent that either, that'd be a bit weird to say the least!

Thanks for the reply Arun. That's what I hoped but past experience with other forums (not here!) tend to make me cautious. Good to know you guys don't play those games

 

[santiagodraco]

I've already postet my view on this matter as a news item on our website:
http://www.pcgameshardware.de/?article_id=602231
(google translation into english: http://translate.google.com/transla...rticle_id=602231&langpair=de|en&hl=de&ie=UTF8)

Basically, if there's any confusion about this matter it's becaue "press" did not pay attention during the launch - for example in Tunis. Because in the presentation as well as in the "public" (journalists only of course) Q&A afterwards this matter was mentioned and AMD clearly stated, R600 does not have the dedicated silicon known as UVD in HD2600/2400.

If ATI wanted to prevent confusion they wouldn't have presented the slides in conjunction with 2900 reviews. However anything is possible.

But more damning are all of the product packages that also contain UVD as a feature. Who's to blame for that and shouldn't the consumer have recourse?