AMD Quits Handheld Market?

Its interesting to consider who will buy the handheld unit from AMD?
Qualcomm is supposedly using their graphics core in snapdragon.
 
We commented on it in the AMD Execution thread.

They claimed in the call they aren't expecting it to be a net cost to close it down, and that they expect that by the end of the year.

Said another way, they think they can sell it by the end of the year to someone else and for more money than whatever it costs them to separate that business from the rest of their operations.

It will be interesting to see if they sell all of Consumer (DTV + handheld) to one buyer, or split it into two and sell each piece separately.
 
Thing is what would AMD be selling off? I'd be very surprised if they where prepared to sell of any ATi derived IP, that leaves the ex bitboys stuff which I guess just comes down to the OVG core and related business...

John.
 
Well, presumably they would keep license to some of the IP as well, like the rights to continue to use Xilleon IP in Radeon products. But the engineers and where applicable the locations, presumably go to the buyer.

Where Xenos tech is in the handheld part, ghu only knows how engineers for that get split up, if they do at all or instead just stay with AMD but license to continue using the tech goes to the buyer.
 
Well, presumably they would keep license to some of the IP as well, like the rights to continue to use Xilleon IP in Radeon products. But the engineers and where applicable the locations, presumably go to the buyer.

Why would they keep the license, its unlikely that the terms if any sale would prevent them from continueing to use any underying tech in the future should they choose to.

Where Xenos tech is in the handheld part, ghu only knows how engineers for that get split up, if they do at all or instead just stay with AMD but license to continue using the tech goes to the buyer.
I find it hard to beleive that they would sell off any core ATi tech in this manner, although that said "handheld xenos" is more of marketing BS than fact imo...

Anyway, is all conjecture until we see a buyer and what they subsequently offer, and of course you have to wonder who wouldr actually buy a loss making business...

John.
 
Given that they specifically use the "Xilleon" name with regards to Radeon video capabilities, and they'd be selling "Xilleon" with the DTV stuff, yes I'd assume they'd need specific contract language of some sort --whether you call it a license or not is something else-- that would let them continue to use that tech.

Who knows, they might even back license for support and new versions down the road. That might make it even more attractive to a buyer --built in long-term contract revenue.
 
Who knows, they might even back license for support and new versions down the road. That might make it even more attractive to a buyer --built in long-term contract revenue.

I'm not sure how that would be divesting themselves of their handheld business...

John.
 
Why not? If you don't make it yourself anymore and still need it, then where are you going to get it? ATI used to own fabs. They don't anymore. . . so that precludes them from getting fab services elsewhere? It better not, or they're out of business.
 
I wouldn't be worried about UVD; it's a done deal, and allowing ATI to keep using it shouldn't be a problem at all. H.265 won't be a requirement for many years. On the other hand, ATI's handheld GPUs do reuse significant PC GPU IP and presumably their roadmap is also based on the capability to do so.

And no, it's not just marketing FUD; it's obviously modified quite a lot and trimmed down, but there's still a clear lineage. So that'll complicate the situation for the acquirer. What's more interesting though is what it'll mean for the IP market if ATI is acquired by, say, Qualcomm or another wireless company that wouldn't be willing to take part in the IP model. That means current licensees would switch to PowerVR's SGX and ARM's Mali in the long-term...
 
Why not? If you don't make it yourself anymore and still need it, then where are you going to get it? ATI used to own fabs. They don't anymore. . . so that precludes them from getting fab services elsewhere? It better not, or they're out of business.

My assumption is that they are divesting themselves of everything to do with handheld, that includes the IP part of the business, do you know differently?

John.
 
And no, it's not just marketing FUD; it's obviously modified quite a lot and trimmed down, but there's still a clear lineage. So that'll complicate the situation for the acquirer. What's more interesting though is what it'll mean for the IP market if ATI is acquired by, say, Qualcomm or another wireless company that wouldn't be willing to take part in the IP model. That means current licensees would switch to PowerVR's SGX and ARM's Mali in the long-term...

Sorry but irrespective of the lineage claims like this are marketing fud that have the sole intent of making the uniformed masses think they are getting something that will deliver xbox360 visuals/performance, when in fact it will do nothing of the sort.

It will be interresting to see who/if somone buys the business, alhought the prime contenders to buys are I think the only licenses that ATi had (Qualcomm and ST, where there anymore?). So there's probably only one new licensing opportunity going to come out of this, but it could obviously mean less competition going forward.

Cheers,
John.
 
My assumption is that they are divesting themselves of everything to do with handheld, that includes the IP part of the business, do you know differently?

John.

Well, I know that they won't sell very many video cards if they can't show HD video on them. How do you expect them to do that if they've "sold off the IP" and the engineers who created it both?
 
Sorry but irrespective of the lineage claims like this are marketing fud that have the sole intent of making the uniformed masses think they are getting something that will deliver xbox360 visuals/performance, when in fact it will do nothing of the sort.
Puh-lease. That's like complaining about the fact G86/RV610 because they're so slow yet trick the customer into thinking they'll get "DX10-level visuals/performance". If OEMs/carriers were dumb enough to directly advertise the XBox360 lineage as a major marketing point (or even just highlight it in feature list), then you might have a point. But it's very very unlikely that ever happens, so I really don't see the problem.

It will be interresting to see who/if somone buys the business, alhought the prime contenders to buys are I think the only licenses that ATi had (Qualcomm and ST, where there anymore?). So there's probably only one new licensing opportunity going to come out of this, but it could obviously mean less competition going forward.
Yeah, Qualcomm and ST are by far the most likely. The IP houses aren't rich enough to buy this, and neither would they have any good reason to do so. Their other major licensee is Freescale BTW (and Nokia, but I don't know how that aligns with their current chip strategy if at all). I guess another possibility is Samsung, because I'm sure they'd like to expand their handheld chip division and they might also be interested in the DTV stuff.
 
Puh-lease. That's like complaining about the fact G86/RV610 because they're so slow yet trick the customer into thinking they'll get "DX10-level visuals/performance". If OEMs/carriers were dumb enough to directly advertise the XBox360 lineage as a major marketing point (or even just highlight it in feature list), then you might have a point. But it's very very unlikely that ever happens, so I really don't see the problem.
There's a big difference in making claim for something that has a perf difference of say an single order of mgnitude and one that is perhaps three orders of magnitude slower. Its fud, there's nothing else to it.
Yeah, Qualcomm and ST are by far the most likely. The IP houses aren't rich enough to buy this, and neither would they have any good reason to do so. Their other major licensee is Freescale BTW (and Nokia, but I don't know how that aligns with their current chip strategy if at all). I guess another possibility is Samsung, because I'm sure they'd like to expand their handheld chip division and they might also be interested in the DTV stuff.

Had forgetten about Freescale, Nokia aren't a licensee, I beleive there was only a agreement to cooperate in graphics (whatever that means). You're correct that the other IP companies wouldn't be interested in buying the whole package, the engineering team may be an attractive resource to aquire, although I strongly suspect that anyone really good will be staying with ATi/AMD. I think the DTV stuff is probably of much greater interrest to a 3rd party than the 3D stuff.

Cheers,
John.
 
Well, I know that they won't sell very many video cards if they can't show HD video on them. How do you expect them to do that if they've "sold off the IP" and the engineers who created it both?

Selling the IP business won't stop them from continueing to develope and use their DTV technology internally, what it does is remove the cost of suport and sales team assocaited with the IP market.

John.
 
You're assuming the DTV engineers aren't going with the sale then. Why would you assume that? What purpose do they have to keep them for a lesser total addressable market thus even increasing the opex/revenue ratio in a worse way than from before the sale.

Surely decreasing opex is a major consideration in this decision.
 
You're assuming the DTV engineers aren't going with the sale then. Why would you assume that? What purpose do they have to keep them for a lesser total addressable market thus even increasing the opex/revenue ratio in a worse way than from before the sale.

Surely decreasing opex is a major consideration in this decision.

You're ignoring the fact that addressing the IP market, particularly for a company that isn't primarily an IP provider, actually costs additional money i.e. developing and using something internally does not give a you an IP business i.e you have to spend money to get the IP into a form that others can integrate, you have to spend money to support people who are integratng your IP and you have to spend money to sell that IP. In fact I beleive ATi had a team of 100-200 in the far east for IP sales and support purposes alone, so there are cleary significant savings to be made by dumping the IP business.

As for if the engineering teams are sold off etc of course I know no more than you, but I would strongly suspet that if AMD intend to continue to offer these feature that/ have on going development then anyone signicant will be staying with AMD (or at least given the option to). The alternative is that they go to the open market for any IP of this type that they need.

John.
 
In ATI's defense whenever I had heard about their OGL_ES2.0 chip, it was called "mini-Xenos". It's obviously some sort of internal naming scheme, yet it implies that it's a small form factor reduced version of Xenos.

On the other hand SGX is also in that sense a small form factor reduced version of the original Series5 GPU and albeit I'm not aware of the latter's exact specifications, no one could convince me either that I could expect from a SGX equivalent performance (unless of course the latter gets beefed up a lot more than the highest 20.3mm^2 @65nm core, which wouldn't fit into any SoC anymore).

The question is has AMD ever presented anything from that OGL_ES2.0 chip in hardware or did they just introduce the according SDK (honest question)? If not, then I'm not at all surprised with John's FUD accusations.

Anyway AMD quitting entirely (or partially...whatever heh) the handheld market isn't great news either; the more competition the merrier.
 
Back
Top