AMD and nVidia Not Competing Directly

Lazy8s

Veteran
With nVidia introducing a family of application processors, equipped with a CPU and all, putting them in competition with semiconductor manufacturers the likes of TI and Renesas, and AMD continuing to license grapihcs IP to established families of application processors, like their recent deal with STM for Nomadik, putting them in competition with the likes of Imgtec and ARM, they're on two different tracks of business for the portable graphics sector.
 
I am not completely up to speed on ATI's design wins or new product announcements in the handheld space. I know they've had some successes with Qualcomm and a number of others.

Overall, I take the deal with STM as an indication that their HW business in this space is not going well (or they see the writing on the wall) and its sort of a natural to say, "well, how else can we get some revenue for this stuff we own".
 
I'd like to understand how you fathomed that.

Be aware that a dual track discrete and/or IP license business model has operated throughout the CE group - console chips are licensed, handheld chips are both discrete and IP licensed (QUALCOMM's deal, for instance, called for both - initial 6xxx series QUALCOMM designs shipped with discrete Imageon's, while the 7xxx integrated the IP).
 
Overall, I take the deal with STM as an indication that their HW business in this space is not going well (or they see the writing on the wall) and its sort of a natural to say, "well, how else can we get some revenue for this stuff we own".

But if they decided not to, they would be saying "we're just making too much money, so we don't want more revenue on stuff we own".
 
Limiting the discussion to handhelds (not CE):

The question is really, what is the trend of their Imageon hardware design wins vs. their licensing design wins?....and how does that correlate to HW Imageon product announcements in the same period (eg. how many have there been?)?

PS: My premise is really a general observation that previously successful HW chip manufacturer's usually go to a licensing IP model as their chip business falls apart.
 
The question is really, what is the trend of their Imageon hardware design wins vs. their licensing design wins?....and how does that correlate to HW Imageon product announcements in the same period (eg. how many have there been?)?
As I pointed out, the QUALCOMM business uses both models, specifically with Imageon. As for product announcements, its not difficult to find releases for discrete Imageon products - 3 models were announced a day before this release. I would advise reading the footnotes of that PR as well.
 
Dave Baumann said:
3 models were announced

I appreciate the link to the 2298/94/2192. That is interesting. As I originally said, I'm not entirely up to speed on ATI's recent activity. So that helps answer part of the question I posed.

Dave Baumann said:
As I pointed out, the QUALCOMM business uses both models
Factually, of course, I understand that the Qualcomm business uses both HW and IP from ATI. That doesn't answer my question of the trend of ATI's HW design wins vs. their IP design wins.

The original premise of this thread (which was not started by me) was:
With nVidia introducing a family of application processors, ....and AMD continuing to license grapihcs IP to established families of application processors, .....they're on two different tracks of business for the portable graphics sector.

I would say, given the discussion and data given that the answer to that premise is yes.
 
That doesn't answer my question of the trend of ATI's HW design wins vs. their IP design wins.
I'd say that's fairly irrelevent, however, considering ATI has always used both business models for their GPU designs, and that they are being extremely successful either way. Nearly $100M of revenue per quarter (4x NVIDIA's!) is nothing to sneeze at.

If their revenue stopped growing, then I'd begin worrying about any shift in ratio for IP/Discrete. Right now, it's still (good) business as usual for them, as far as I can see. The only good reason I could see that would make them incvrease their ratio of IP is if more and more SoCs already embedded in-house or non-ATI video/image/sound processing. In that case, AMD's solutions which also include those functions would be overkill. I don't think that dynamic is significant enough just yet though, and it's hard to predict long-term.
 
Arun D said:
I'd say that's fairly irrelevent (the trend), however,....
Yes, I'd agree with you in the context of whether Nvidia and ATI are competing in the handheld space (the original premise of this thread).

I raised the question of ATI handheld Imageon HW vs IP design-win trend as a question of ATI's motivation for doing the licensing deal with STM.....a different topic completely.

Arun D said:
The only good reason I could see that would make them incvrease their ratio of IP is if more and more SoCs already embedded in-house or non-ATI video/image/sound processing.
Yes, I think this observation begins to get at what I was wondering about. It depends on what your horizon is for determining who the winners are amongst the major players (Intel, AMD(ATI), Nvidia, TI, STM, Qualcomm, Nok, Marvel etc.)?

Arun D said:
ATI has always used both business models for their GPU designs, and that they are being extremely successful either way. Nearly $100M of revenue per quarter (4x NVIDIA's!) is nothing to sneeze at.
Well I've always questioned the real success of ATI's licensing models. I don't recall the specifics any longer, but I think the XBOX360 licensing deal worked out to a couple of bucks per unit which works out to about $5 revenue on a chip sale at 40% margins. I don't know much about their other deals, but I think overall you'd be hard pressed to convince me it was successful given their financial performance prior to being bought.....and by the way when one looks at success, one must look at lost opportunity. So the XBOX360 was a successful development project for them, but at the cost of what other of their businesses (I am of course thinking of the desktop add-in board market vs Nvidia.)?
 
Yes, I think this observation begins to get at what I was wondering about. It depends on what your horizon is for determining who the winners are amongst the major players (Intel, AMD(ATI), Nvidia, TI, STM, Qualcomm, Nok, Marvel etc.)?
You should bear in mind that ATI/AMD not only has business deals in place with QUALCOMM and ST now, but also has a strategic partnership with Nokia. ATI/AMD already have very good relationships with with many of the primary handheld vendors, so the onus probably shouldn't be on ATI/AMD's business model but the others that perhaps are now finding they need to go a different route in order to be competetive; however, in doing so they can find themselves in a much more competetive realm.

On the flipside of that, just because NVIDIA has signalled an intent on SoC, a lack of any type of annoucement from AMD does not preclude them from doing the same as well. ATI/AMD's handheld group hides it light under a bushel (as this thread bears evidence by the lack of understanding as to exactly how they are doing in this market) and they have no pressing need to pre-announce things before they are ready, unlike NVIDIA who has to prove their investment on this side of the business, both internally and the external aquisitions, is going to pay off.

I don't recall the specifics any longer, but I think the XBOX360 licensing deal worked out to a couple of bucks per unit which works out to about $5 revenue on a chip sale at 40% margins. I don't know much about their other deals, but I think overall you'd be hard pressed to convince me it was successful given their financial performance prior to being bought.....and by the way when one looks at success, one must look at lost opportunity. So the XBOX360 was a successful development project for them, but at the cost of what other of their businesses (I am of course thinking of the desktop add-in board market vs Nvidia.)?

I think you might find that Dave Orton had already said that the licensing was for $5 per unit sold, of course MS would have paid for the development of the chip as well. You don't need to stray too far from this site to get an understanding that the 360 chip development is being leveraged for both AMD's desktop graphics processors and the very design licensed to ST that sparked this thread - thats fairly widespread technology re-use.
 
I think if you go to AMD's job pages and search "SoC" you'll find more than enough activity there to justify the belief their interest is keen, ongoing, and with an eye on the future out into 2008-2010 (32nm, ftw?) even.
 
Back
Top