The next-gen Game Boy: Isn't it of strategic importance to ATI?

[Megadrive1988]

Assuming Nintendo will give the 3D chip contract for the next-generation Game Boy to ATI, would you not imagine that it is of great strategic importance to ATI given the likely volume of units involved ? with the potental to sell tens of millions of new Game Boys from launch date through the end of the decade -- and probably sell more than both Xbox360 and Revolution put together, one would think this is a priority for Orton and his company. I was somewhat surprised that DS did not have an ATI chip in it, but I almost cannot imagine the next Game Boy without one.

 

[AlphaWolf]

Assuming Nintendo will give the 3D chip contract for the next-generation Game Boy to ATI, would you not imagine that it is of great strategic importance to ATI given the likely volume of units involved ? with the potental to sell tens of millions of new Game Boys from launch date through the end of the decade -- and probably sell more than both Xbox360 and Revolution put together, one would think this is a priority for Orton and his company. I was somewhat surprised that DS did not have an ATI chip in it, but I almost cannot imagine the next Game Boy without one.

You are talking about relatively low dollar items. I imagine they would like to have the business but I don't see it contributing that much to their bottom line.

 

[Lazy8s]

With portables being the largest market for graphics and ATi all but shut out of it currently, an inside track to the next Game Boy contract, while not as high volume as a major chip manufacturer's application processor, might be ATi's best opportunity. Their difficulty as a company who typically produces their own chips has been that the major chip suppliers to the portable market all license their graphics IP and are already settled in with a solution for at least the next generation. The likes of ATi and nVidia have a lot of inroads left to make in the industry in order to secure a significant spot in the market.

ATi does have a processor for handhelds on the way that achieves the requirements for Shader Model 3.

 

[AlphaWolf]

With portables being the largest market for graphics and ATi all but shut out of it currently, an inside track to the next Game Boy contract, while not as high volume as a major chip manufacturer's application processor, might be ATi's best opportunity.

Largest market? Excluding pcs and cellphones perhaps.

 

[Lazy8s]

Cellphones are portables.

 

[AlphaWolf]

Cellphones are portables.

Well if you are counting cellphones as portables, then ATI isn't shut out of that market. The problem is that the chips that go into cellphones have an ASP of $5. If they have a 20% margin you are only looking at $1 per unit.

 

[wireframe]

You are talking about relatively low dollar items. I imagine they would like to have the business but I don't see it contributing that much to their bottom line.

I don't know that the margins woud be any real concern. These handheld gaming cosoles are a big business and is probably going to grow. I am sure someone like ATI could get some nice income from licensing IP and reaping the rewards on quantities.

It seems logical to me that portable gaming consoles require higher performance graphics than mobiles phones and can therefore offer better margins and more commonality with the core business to make it interesting.

 

[Lazy8s]

Cellphone volumes reach into the hundreds of millions.

 

[AlphaWolf]

Cellphone volumes reach into the hundreds of millions.

So if ATI had the entire market to themselves they could increase their overall sales by 30%?

 

[SiBoy]

Their difficulty as a company who typically produces their own chips has been that the major chip suppliers to the portable market all license their graphics IP and are already settled in with a solution for at least the next generation.

Graphics in the current generation has been a non-event, with many companies taking out token licenses of whatever happens to be available (many of which will never be enabled). Once graphics in mobiles is taken more seriously, the choice of which vendor's graphics to use will be taken more seriously.

 

[Dave Baumann]

ATi does have a processor for handhelds on the way that achieves the requirements for Shader Model 3.

More or less it will be a cut down Xenos.

 

[Lazy8s]

Xenos's large tile approach could pressure the size limits of a portable chip with a lot of embedded memory.

 

[Teasy]

More or less it will be a cut down Xenos.

Cut down would have to be one hell of an understatement if true.

 

[Lazy8s]

Presumably, ATi's processor is custom designed for wireless requirements using a Xenos-like architecture.

 

[Megadrive1988]

Cut down would have to be one hell of an understatement if true.

yeah like maybe 8 unified shader ALUs instead of 48, and maybe 2 MB EDRAM

 

[Dave Baumann]

In terms of architecture, I wouldn't know whether it has eDRAM or not (although enough for 640x480 may be wise).

Some time ago AJ from Futuremark joined ATI, his job is to evangelise to handheld developers the use of shaders - this it kind of an odd job at the moment seeing as ATI have no shader enabled handheld parts yet. I met with Richard Huddy recently and asked after AJ and having pointed out that he must be having an easy time of it at the moment () because he's got nothing to evangelise; Richard pointed out that he can already evangelise the capabilities as they already exist. I already knew that they were working on Xenos for the mobile space, given what Bob had said when he was initially wheeled out at E3, but from Richards comments it sounds like the shader capabilities will be very similar, if not the same.

Its wouldn't surprise me if 65nm was the target for the first version of this.

 

[nelg]

In terms of architecture, I wouldn't know whether it has eDRAM or not (although enough for 640x480 may be wise).

Some time ago AJ from Futuremark joined ATI, his job is to evangelise to handheld developers the use of shaders - this it kind of an odd job at the moment seeing as ATI have no shader enabled handheld parts yet. I met with Richard Huddy recently and asked after AJ and having pointed out that he must be having an easy time of it at the moment () because he's got nothing to evangelise; Richard pointed out that he can already evangelise the capabilities as they already exist. I already knew that they were working on Xenos for the mobile space, given what Bob had said when he was initially wheeled out at E3, but from Richards comments it sounds like the shader capabilities will be very similar, if not the same.

Its wouldn't surprise me if 65nm was the target for the first version of this.

Would it be too much to infer that since Xenos is the MS name for the C1 that perhaps there is more to read into that comment from Richard?

 

[arjan de lumens]

Well if you are counting cellphones as portables, then ATI isn't shut out of that market. The problem is that the chips that go into cellphones have an ASP of $5. If they have a 20% margin you are only looking at $1 per unit.

$1 per unit?? Sounds unsustainable to me, at least. ARM claims that they are on average collecting ~8 cents for each CPU core; IME, it's very hard to sell a GPU at 12 times the price of the CPU.

The mobile phone market for 3D IPs may be gigantic in terms of number of units; in terms of actual revenue it's very small (~1-3% of the total 3d graphics hardware market, if you're optimistic). There are also a lot more companies offering 3D IPs in this market than on the desktop, which also puts a strong downward pressure on what prices you can expect.

 

[AlphaWolf]

$1 per unit?? Sounds unsustainable to me, at least. ARM claims that they are on average collecting ~8 cents for each CPU core; IME, it's very hard to sell a GPU at 12 times the price of the CPU.

The mobile phone market for 3D IPs may be gigantic in terms of number of units; in terms of actual revenue it's very small (~1-3% of the total 3d graphics hardware market, if you're optimistic). There are also a lot more companies offering 3D IPs in this market than on the desktop, which also puts a strong downward pressure on what prices you can expect.

I was just using that as a potential upper limit, I am only aware of their ASP from their conference call earlier this year.

http://www.beyond3d.com/forum/showthread.php?t=24342

 

[Lazy8s]

Assuming ATi is trying to compete directly with PowerVR for the next generation, their processor could be in consumer products not much later than first-half 2007. SGX is already prototyping in silicon, presumably as the wireless processor Stanwood at Intel. Some of the major fabricators could have their 65nm processes ready around that time.

Some idea of royalty and licensing rates for wireless graphics processor IP can be figured from ImgTec's recent financial reports which list some general results on sales volume and revenue.