2900 XT lacks a UVD

Page 4 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

apoppin

Lifer
Mar 9, 2000
34,890
1
0
alienbabeltech.com
kinda hard with your link :p

Your support portal session has expired due to inactivity and the requested action has not been completed. If you entered data in the previous screen and would like to save it, please follow these steps.

1. Click this button.
2. Select and copy your unsaved text.
3. (optional) Paste your copied text into a temporary text file.
4. Refresh your browser and begin your process again.
5. Once you have arrived at the desired page, paste your data in the appropriate text boxes.

If you just want to restart your support portal session and discard any possible unsaved data from the previous screen, click below.
 

olmer

Senior member
Dec 28, 2006
324
0
0
^^ http://support.ati.com ^^ it is the most recent topic added

Cannot see how it helps them though :)

Free opportunity to play with HIS 2900 - i'll keep it within 7 days though to be able to get a full refund including delivery under UK DSR and not involve Trading Standards/merchant . Hardly possible to get satisfactory refund within 7 days (looking for at least £50 off £250 price) from a manufacturer. I advice against trying GeCube and Sapphire ? there is virtually no customer services there. Will see if HIS is any better. They will be trying to flog a lot of ?open box? 2900s with 15% off all over the net pretty soon I guess.
 

apoppin

Lifer
Mar 9, 2000
34,890
1
0
alienbabeltech.com
Originally posted by: olmer
^^ http://support.ati.com ^^ it is the most recent topic added

Cannot see how it helps them though :)

Free opportunity to play with HIS 2900 - i'll keep it within 7 days though to be able to get a full refund including delivery under UK DSR and not involve Trading Standards/merchant . Hardly possible to get satisfactory refund within 7 days (looking for at least £50 off £250 price) from a manufacturer. I advice against trying GeCube and Sapphire ? there is virtually no customer services there. Will see if HIS is any better. They will be trying to flog a lot of ?open box? 2900s with 15% off all over the net pretty soon I guess.

how hard is to do this?
737-27824: Radeon? HD 2900 series - Wrong Information Regarding UVD Support on Retail Box

The information in this article applies to the following configuration(s):

* GeCube Radeon? HD 2900XT
* HIS Radeon? HD 2900XT
* Sapphire Radeon? HD 2900XT
* Windows Vista 32-bit Edition
* Windows Vista 64-bit Edition
* Windows XP Professional
* Windows XP Home Edition
* Windows XP Media Center Edition
* Windows XP Professional x64 Edition



Some AIB Partners indicate UVD (Unified Video Decoding) support on their Radeon? HD 2900 retail box.

Note: UVD is the abbreviation for Unified Video Decoding and not Universal Video Decoding.


Explanation:
Radeon? HD 2900XT does NOT support UVD. Radeon? HD 2600 and 2400 series will be supporting UVD.

ATI has been advised of this issue and is currently communicating with AIB partners. Any updates will be published when they become available.

an *admission*

... of course notice that AMD says it is "ATi's fault"
ATI has been advised of this issue
... no doubt AMD will fire some more former ATi employees :p

 

nullpointerus

Golden Member
Apr 17, 2003
1,326
0
0
Hey, it took four edits to get a link into that post that worked on *my* computer. :thumbsdown:

It's not my fault AMD is using a KB system designed by a moron. IIRC, Creative's KB system has the same design flaw.

Here's a cool thought: create a KB system where each article gets its own HTML page. Then number them like KBxxxxxx. That way people who aren't support drones can say:

Oh, that's a known issue: go to http://support.microsoft.com/kb/930375/en-us for more information.

:shocked:

OMG! Someone actually built this KB system! Like, at least ten years ago!

/anti-stupid-programmers rant
 

apoppin

Lifer
Mar 9, 2000
34,890
1
0
alienbabeltech.com
Originally posted by: nullpointerus
Hey, it took four edits to get a link into that post that worked on *my* computer. :thumbsdown:

It's not my fault AMD is using a KB system designed by a moron. IIRC, Creative's KB system has the same design flaw.

Here's a cool thought: create a KB system where each article gets its own HTML page. Then number them like KBxxxxxx. That way people who aren't support drones can say:

Oh, that's a known issue: go to http://support.microsoft.com/kb/930375/en-us for more information.

:shocked:

OMG! Someone actually built this KB system! Like, at least ten years ago!

/anti-stupid-programmers rant
hey ... amigo ... i was NOT picking on you

i was kinda hoping for a C&P, though

.. .. .and did you just see?

MODERATION in the video forum?
-a nasty troll thread just got locked!
:shocked:

---was the mod 'invited' ? passing by ? about to take up "residence" in Video ?

i need some time to think
... and recover
 

fierydemise

Platinum Member
Apr 16, 2005
2,056
2
81
Its good to see AMD own up to it, not that they really had a choice, but still its nice. Anyway as I said earlier Driver Heaven found that the HD2900XT still has lower CPU utilization on HD video without a UVD (here and here) so I think this issue is as dead.
 

olmer

Senior member
Dec 28, 2006
324
0
0
Originally posted by: apoppin
how hard is to do this?

I do not know about US, but in UK you are protected by Distance Selling Regulations, where you can return any item you have purchased not in shop (without seeing/trying it) within at least 7 days (most of merchants make it 14 days) for a full and unconditional refund (restocking fee is unlawful). Apart from this there is a clear consumer law with regards to misrepresentation of the product, where merchant /manufacturer misleading to get sales is severely punished. It does not matter what site?s t&c says at all as long as ?contract? is forged (merchant accepted your money) and you can prove that product is misrepresented ? you are guaranteed at least a full refund (with any delivery fees) plus any interest you may have incurred. This, however, has to go through Trading Standards if merchant is stubborn (rarely happens as the merchant is obliged to pay TS?s fees) ? so just threatening works 100%. Works with UK/foreign companies regardless. Then there is also an option to claim a chargeback on your cc ? takes longer and you have to prove you point to the letter, but no one wants to mess with mc/visa ? their license is in stake.

With HIS ? contact in writing (email) with an ultimatum (if you return the card used they again by law cannot sell it as new ? so at least 10% losses) clearly stating you have nothing against the merchant and do not want them to bear these costs ? therefore it is between HIS and you. You are willing to accept a certain discount to keep the card. If they just disregard as GeCube does ? return the card when you see fit to the merchant for a full refund, giving them a copy of your email. If they/AMD find some sneaky way out of it and replies ? GFY, etc ? just return the card within 7 days.
 

apoppin

Lifer
Mar 9, 2000
34,890
1
0
alienbabeltech.com
i was actually referring to "how hard is it to C&P"
-but thank you for the UK info

i expect that ANYONE that bought a HD2900xt in the USA expecting UVD can also get a full refund from the e/retailer ... especially if they give this reason
... the colonies and former colonies are actually pretty civilized now and up-to-date on consumer protection laws, also
 

nullpointerus

Golden Member
Apr 17, 2003
1,326
0
0
Originally posted by: apoppin
Originally posted by: nullpointerus
Hey, it took four edits to get a link into that post that worked on *my* computer. :thumbsdown:

It's not my fault AMD is using a KB system designed by a moron. IIRC, Creative's KB system has the same design flaw.

Here's a cool thought: create a KB system where each article gets its own HTML page. Then number them like KBxxxxxx. That way people who aren't support drones can say:

Oh, that's a known issue: go to http://support.microsoft.com/kb/930375/en-us for more information.

:shocked:

OMG! Someone actually built this KB system! Like, at least ten years ago!

/anti-stupid-programmers rant
hey ... amigo ... i was NOT picking on you

i was kinda hoping for a C&P, though

.. .. .and did you just see?

MODERATION in the video forum?
-a nasty troll thread just got locked!
:shocked:

---was the mod 'invited' ? passing by ? about to take up "residence" in Video ?

i need some time to think
... and recover

I know you weren't picking on me, and I wasn't angry at you. The situation just really irritates me. The whole point of a knowledge base is to allow people to write something once and link to it when it's needed. Why create a KB system that doesn't work with external links like the rest of the WWW? Next time I'll post the KB article body.

Moderation in the video forum isn't entirely unheard of. In fact, I've seen it happen before, even in my short lifetime. No idea why it happened recently, except that the OP of that thread was so blatantly a pro-nvidia troll. Maybe he was demoing to replace what's-his-name, the screwdriver king?

:)
 

guezz

Member
May 10, 2006
45
0
0
Originally posted by: fierydemise
Its good to see AMD own up to it, not that they really had a choice, but still its nice. Anyway as I said earlier Driver Heaven found that the HD2900XT still has lower CPU utilization on HD video without a UVD (here and here) so I think this issue is as dead.
Er, the driver used don't support decode acceleration - 8.38 will be the first. I smell BS.

Originally posted by: Dave Baumann (ATI employee)
HD 2900 does not have UVD; it'll be paried with HD video capable CPU's. The drivers used in that test don't yet have decode acceleration enabled either, first revision of that comes with 8.38.
http://forum.beyond3d.com/showpost.php?p=1009569&postcount=3
 

Creig

Diamond Member
Oct 9, 1999
5,170
13
81
Originally posted by: apoppin
Originally posted by: olmer
^^ http://support.ati.com ^^ it is the most recent topic added

Cannot see how it helps them though :)

Free opportunity to play with HIS 2900 - i'll keep it within 7 days though to be able to get a full refund including delivery under UK DSR and not involve Trading Standards/merchant . Hardly possible to get satisfactory refund within 7 days (looking for at least £50 off £250 price) from a manufacturer. I advice against trying GeCube and Sapphire ? there is virtually no customer services there. Will see if HIS is any better. They will be trying to flog a lot of ?open box? 2900s with 15% off all over the net pretty soon I guess.

how hard is to do this?
737-27824: Radeon? HD 2900 series - Wrong Information Regarding UVD Support on Retail Box

The information in this article applies to the following configuration(s):

* GeCube Radeon? HD 2900XT
* HIS Radeon? HD 2900XT
* Sapphire Radeon? HD 2900XT
* Windows Vista 32-bit Edition
* Windows Vista 64-bit Edition
* Windows XP Professional
* Windows XP Home Edition
* Windows XP Media Center Edition
* Windows XP Professional x64 Edition



Some AIB Partners indicate UVD (Unified Video Decoding) support on their Radeon? HD 2900 retail box.

Note: UVD is the abbreviation for Unified Video Decoding and not Universal Video Decoding.


Explanation:
Radeon? HD 2900XT does NOT support UVD. Radeon? HD 2600 and 2400 series will be supporting UVD.

ATI has been advised of this issue and is currently communicating with AIB partners. Any updates will be published when they become available.

an *admission*

... of course notice that AMD says it is "ATi's fault"
ATI has been advised of this issue
... no doubt AMD will fire some more former ATi employees :p

No, it's NOT an admission of any kind. Read it again.

It says "Some AIB Partners indicate UVD (Unified Video Decoding) support on their Radeon? HD 2900 retail box."

ie - It was an error on the part of the AIBs, not ATI/AMD. One of AMDs slides clearly shows that they themselves did not say the 2900 contains the UVD, only the 2400 and 2600.

http://www.beyond3d.com/images/news/R6FamilyUVD.jpg

If there is proof out there that AMD advertised the 2900 as containing UVD, then I'll happily retract my statement. But I just got done searching and couldn't find one. The only times I saw it was from retailers and AIBs.

As far as "ATI has been advised of this issue and is currently communicating with AIB partners", it just means that people have contacted ATI about the AIB advertising error and ATI is attempting to get the AIBs to correct it.


Originally posted by: smitty3268
There are basically 2 approaches - the way ATI did it in the last generation where they use their existing shader hardware to do the acceleration, and the NVIDIA way which is to add a seperate chip on the card which exclusively handles it.

The former way is cheaper in terms of silicon but requires a lot of shader power, especially for HD content. AMD is saying that the 2900XT has enough horsepower to do it that way, but the drivers haven't been finished yet. On the other hand, they had to add another chip (UVD) to the lower end parts because they wouldn't have been able to do it. This is what everyone seemed to assume the 2900 had as well, and is really the better solution because it would use less power due to the specially designed hardware being more efficient for the task and would also almost certainly use less CPU as well due to the drivers having to do less work.

So while the 2900 may lack a UVD, it's only because it has no need of one. The shaders on the 2900 can perform this function on their own. It's only the lower end cards that require a UVD because they simply don't have enough shaders to do it on their own. At this time, I believe we're simply awaiting the driver that will enable this.
 

apoppin

Lifer
Mar 9, 2000
34,890
1
0
alienbabeltech.com
Creig ... it's an admission
--here let me post your admission again for ALL to see
So while the 2900 may lack a UVD

thank-you ... it does lack a UVD as advertised
--"AiB's" ... 'whatever'


... your "need" is immaterial ... the lawyers will debate it and AMD - or its partners - will pay
 

fierydemise

Platinum Member
Apr 16, 2005
2,056
2
81
Originally posted by: guezz
Originally posted by: fierydemise
Its good to see AMD own up to it, not that they really had a choice, but still its nice. Anyway as I said earlier Driver Heaven found that the HD2900XT still has lower CPU utilization on HD video without a UVD (here and here) so I think this issue is as dead.
Er, the driver used don't support decode acceleration - 8.38 will be the first. I smell BS.

Originally posted by: Dave Baumann (ATI employee)
HD 2900 does not have UVD; it'll be paried with HD video capable CPU's. The drivers used in that test don't yet have decode acceleration enabled either, first revision of that comes with 8.38.
http://forum.beyond3d.com/showpost.php?p=1009569&postcount=3
All I know is that those graphs show less CPU usage, Driver Heaver could have screwed up its possible. My point is that UVD or not the HD2900XT still has (or will have) GPU acceleration for HD video so the lack of a UVD doesn't make the card a failure for HD video like some people seem to be claiming.
 

Creig

Diamond Member
Oct 9, 1999
5,170
13
81
Originally posted by: apoppin
Creig ... it's an admission
--here let me post your admission again for ALL to see
So while the 2900 may lack a UVD

thank-you ... it does lack a UVD as advertised
--"AiB's" ... 'whatever'

Admission? Huh? Apparently AMD/ATI never advertised it as having a UVD in the first place, their AIBs did. Why would what I said be an "admission" of anything?

Do you know what an AIB is? It stands for Add-In-Board. As is, card manufacturers. Apparently, some of the AIB marketing departments got their facts screwed up and thought that ALL of the new cards had UVD and advertised it as such. ATI has been informed of this and is attempting to get them to stop advertising the HD2900XT as having one.


Originally posted by: apoppin
... your "need" is immaterial ... the lawyers will debate it and AMD - or its partners - will pay

Need? Huh? What need? You make it so difficult for anyone to comprehend what you're trying to say with all your broken sentences and odd punctuation. If you would just make normal posts with normal sentences like everyone else it would be much easier to grasp what you're trying to get across.

This is not AMD's nor ATI's fault, but the fault of the individual card companies. It's unfortunate and hopefully ATI/AMD can get their board partners to straighten out their advertising departments.
 

apoppin

Lifer
Mar 9, 2000
34,890
1
0
alienbabeltech.com
This is not AMD's nor ATI's fault, but the fault of the individual card companies. It's unfortunate and hopefully ATI/AMD can get their board partners to straighten out their advertising departments.
You post as though you actually know for sure that it is the AiB's fault. i wonder where AMD's partners get their information ... from nvidia?

What access to info do you have that i don't that shows it was all the partners fault and not AMD's? After all, the *disclaimer* is also on AMD's site ... if AMD is not at fault to any degree they wouldn't need it at all.

... and thanks for "explaining" what AiB meant ... it IS pretty obscure and i am also pretty sure i used it without having a clue what it meant ... only you have the key of knowledge
 

apoppin

Lifer
Mar 9, 2000
34,890
1
0
alienbabeltech.com
Originally posted by: Wreckage
Originally posted by: Creig

Admission? Huh? Apparently AMD/ATI never advertised it as having a UVD in the first place

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

this is the guy that started the avalanche

http://techreport.com/onearticle.x/12552
I've just learned something that compels me to publish a major correction to our review of the Radeon HD 2900 XT GPU. I got the clear, distinct impression from AMD's presentations, statements, and conversations with me at its Radeon HD press event that its new UVD video decode acceleration logic was present in its entire lineup of Radeon HD graphics chips, and I relayed that information to you in our review of the product, promising to follow up with tests of this feature at a later date.

True to my word, I set out yesterday to test HD video decode acceleration on a Radeon HD 2900 XT using an HD DVD drive and a version of PowerDVD supplied by AMD for such purposes. To my surprise, CPU utilization during playback on our Core 2 Extreme X6800 test system ran between 40 and 50%, well above what one would expect from a solution with full HD video decode acceleration.

Naturally, I contacted AMD to inquire about the problem. I received a reply from AMD's David Baumann discussing the issue that ended with this revelation:

Be aware, though, that HD 2900 XT itself does not feature UVD, this is specific to 2600 and 2400, so the levels of CPU utilization you see should be somewhat similar to the previous generation.

The UVD logic handles the CPU-intensive bitstream processing and entropy decode acceleration portions of the HD video playback pipeline. These are the most notable video decode acceleration capabilities that would separate the Radeon HD 2900 series from its direct competition, the GeForce 8800 series, if the HD 2900 XT actually had them. Turns out it does not. As the email states, the video playback capabilities of the Radeon HD 2900 XT are essentially similar to those of the previous-gen Radeon X1950.

So the essence of our correction is that the Radeon HD 2900 XT doesn't offer robust acceleration of HD video playback and will not likely reduce CPU utilization or power consumption substantially during high-definition video playback versus a GeForce 8800. We still intend to follow up with testing, but the lack of UVD logic on the GPU resets our expectations dramatically.

With that out of the way, I believe I ought to take a moment to explain how we came to believe the Radeon HD 2900 XT had full HD video playback acceleration, an impression formed by many layers of talk from AMD, starting with the Radeon HD name. Let me share a slide with you from a presentation titled "ATI Radeon HD 2000 Series and the Ultimate HD Experience," given by AMD's David Cummings, Director of Mobile GPU Marketing. The slide looks like so:

see slide - exhibit A

You can, of course, read for yourself that it says "Avivo HD technology makes full spec HD DVD / Blu-Ray (HD Disc) playback accessible at all price points," but I just like repeating it. That gives one a certain idea, does it not? Now, let's have a look at another slide showing what Avivo HD brings to video decode acceleration:


see slide - exhibit B

The bit labeled "Avivo HD" shows GPU acceleration of bitstream processing and entropy decode, and makes clear it's distinct from the Radeon X1000's Avivo video processing, which lacks acceleration of those stages.

Now, look at any specs list for the Radeon HD 2900 XT?say, this one from AMD's website, and you will find listed among its specs "ATI Avivo? HD Video and Display Platform" and a bullet point under that saying "HD decode acceleration for H.264/AVC, VC-1, DivX and MPEG-2 video formats." At the end of the day, one gets the impression that this GPU has Avivo HD, with all that entails.

Of course, AMD has left itself some wiggle room in its technical statements. The specs list above isn't technically untrue?just imprecise. The dodge built into the Cummings presentation, with its talk of making HD video playback "accessible at all price points" seems to be that high-end CPUs can handle HD video playback without as much assistance from the GPU. But that's a paper thin excuse, in my view.

To make sure this wasn't simply a matter of me missing the boat?it has been known to happen, and I've got a few gray hairs promising more of the same in the future?I checked with a couple of other journalists who attended a separate Radeon HD press event the week after the one I attended. Both Marco Chiappetta from HotHardware and Ryan Shrout of PC Perspective came away from their meetings with AMD convinced the Radeon HD 2900 XT had full HD playback acceleration via UVD logic, as well. I was not alone in gathering this impression from AMD. To their credit, some reviewers did sort through the fog and identify the Radeon HD 2900 XT's lack of UVD, but they were swimming against the tide of statements from AMD itself.

Nor could any of us have uncovered this fact prior to the publication of our reviews via testing, because AMD hasn't yet delivered a driver that includes the support for the Radeon HD's "full" multimedia capabilities.
judge for yourself if it was AMD or their AiBs that mislead

i believe the AiBs themselves were mislead by AMD ... hence the *disclaimer* on AMD's site
 

Matt2

Diamond Member
Jul 28, 2001
4,762
0
0
I think it was the board partners that mislead the public by saying their cards actually had the UVD.

But OTOH, it was AMD that mislead the board partners using the slides that can be seen on the article by the Inq.
 

Keysplayr

Elite Member
Jan 16, 2003
21,209
50
91
From those two links, it would appear that AMD led AIB's to believe, with documentation and the like, that the entire 2xxx line was equipped with the UVD feature.
So AIB's were only advertising what they were led to believe they "could".

Creig, you need to rethink your stance on this. I mean, it doesn't really make a difference to me whether the 2900XT has UVD or not. Nor does it matter to me whose fault it was that it was advertised on the boxes. But apparently, it matters much more to you where the blame falls. I think we can all see for ourselves where that blame belongs.
So why do you care so much? What is the big deal? You're treading uphill here with some effort that is kind of futile. All you really have to do, is keep an open mind and look at the data from a "neutral" position and you will see what the rest of us see. And you won't look that foolish pissing into the wind. Not trying to aggravate you, just wishing you would keep things real for a change.
 

nullpointerus

Golden Member
Apr 17, 2003
1,326
0
0
Maybe R600 was designed with UVD, but it ended up being disabled in the final silicon because the GPU was just too damn big to get everything working properly in decent yields, so AMD is now doing the "corporation dance" to downplay the missing feature?

700m transistors...and we thought G80 was huge!

pure speculation btw
 

mruffin75

Senior member
May 19, 2007
343
0
0
This is the answer I got from AMD/ATI (although it took some pushing past the whole "You can do a comparison on our website" answer..)

Solution: 737-27824: RadeonT HD 2900 series - Wrong Information Regarding UVD Support on Retail Box
The information in this article applies to the following configuration(s):

* GeCube RadeonT HD 2900XT
* HIS RadeonT HD 2900XT
* Sapphire RadeonT HD 2900XT
* Windows Vista 32-bit Edition
* Windows Vista 64-bit Edition
* Windows XP Professional
* Windows XP Home Edition
* Windows XP Media Center Edition
* Windows XP Professional x64 Edition

Some AIB Partners indicate UVD (Unified Video Decoding) support on their RadeonT HD 2900 retail box.

Note: UVD is the abbreviation for Unified Video Decoding and not Universal Video Decoding.


Explanation:
RadeonT HD 2900XT does NOT support UVD. RadeonT HD 2600 and 2400 series will be supporting UVD.

ATI has been advised of this issue and is currently communicating with AIB partners. Any updates will be published when they become available.

 

MrPickins

Diamond Member
May 24, 2003
9,088
722
126
If the GPU is powerful enough to do hardware decoding, and AMD implements it in a new driver, what is the problem?

The only reason I can see for UVD is for GPUs too weak to decoding in shaders. Which is why they will be included in low/mid range cards.
 

Cookie Monster

Diamond Member
May 7, 2005
5,161
32
86
Originally posted by: nullpointerus
Maybe R600 was designed with UVD, but it ended up being disabled in the final silicon because the GPU was just too damn big to get everything working properly in decent yields, so AMD is now doing the "corporation dance" to downplay the missing feature?

700m transistors...and we thought G80 was huge!

pure speculation btw

Well, G80 should be around 700million transistors since NVIO chip does count as a part of the G80 chip.

What i think is that from the rumours ive heard, R600 was co developed with R500. Therefore it didn't have enough time to include features such as UVD due to the fact that it was already taped out as according to schedule. I heard the project took 2~3 years i think it was. This goes same as G80 since it probably missed out on adding the video engine enhancements seen on G84.

As you can see, you cant just add in stuff like here and there during the finalizations of an architecture. (Not for design reports/assignments however :D)
 

Aikouka

Lifer
Nov 27, 2001
30,383
912
126
Originally posted by: tuteja1986
In fact, the Radeon HD series is the first line of GPUs to integrate the HDCP encryption keys directly into the ASIC, no external CryptoROM chip is needed.

8800GTS and 8800GTX has this as well. I believe the 8600GTS also has it on-die, but the rest are via chip as they're not required.

EDIT:

Also, nVidia may not have mentioned that they had PureVideo 2 decoding, but the vendors did and vendors pull information from the manufacturer's page and where do they get their information from.... drum roll please... nVidia!

In fact, you can even see it on their website for the 8800:

http://www.nvidia.com/page/8800_features.html
NVIDIA® PureVideo? HD Technology2:
The combination of high-definition video decode acceleration and post-processing that delivers unprecedented picture clarity, smooth video, accurate color, and precise image scaling for movies and video.

I don't see anything special when using my 8800 compared to any other card that I've owned... and that's after using a 6800GT where the PureVideo didn't even work on :Q.

http://www.nvidia.com/page/8800_features.html
Discrete, Programmable Video Processor:
NVIDIA PureVideo is a discrete programmable processing core in NVIDIA GPUs that provides superb picture quality and ultra-smooth movies with low CPU utilization and power.

Comparing the 8600 and the 8800... we see this really isn't true.