Torrenza/CPU-GPU Hybrid

FiringSquad: How will this announcement affect your Torrenza initiative, specifically as it relates to hardware manufacturers in the gaming space? Do you feel a manufacturer like AGEIA or NVIDIA may be less willing to participate as a result?

Jon Carvill: Well that’s something that you’re really going to have to ask them. I really can’t answer that question for them. But from our perspective nothing’s changed on Torrenza, we still plan on opening up our architecture and allowing third parties to come in and innovate and to deliver solutions that provide unique propositions for the customer. So you really have to talk to AGEIA or NVIDIA to get their view on things but our approach hasn’t changed and they’ll be welcome to participate with Torrenza.

FiringSquad: One slide in your strategic vision presentation mentions four different platforms a general purpose platform, a data-centric platform, graphics-centric, and media-centric platforms. The general purpose and media-centric platforms appear to depict the combination of the CPU and GPU on the same die, while the graphics-centric platform depicts four GPUs. Can you go over this slide in a little more detail with us and perhaps provide a little more insight on where you plan on going with this?

Chris Evenden: Well the point there really is that we now have the chipset, we have CPUs, we have GPUs, and we can mix and match the appropriate level to each of those things to different customer requirements. So you might have a small CPU and a big video engine and very little graphics for the media-centric platform. Or you might have a big CPU with massive graphics processing power, and next thing you know you’ve got a high-end gaming machine. There are many combinations, we illustrated four different combinations in that schematic but of course that’s all very high level. I don’t think going into detail will give you any more specific information but the basic point is by having all this technology under the same roof it allows us to innovate on the PC platform in a way that just hasn’t been seen in the PC industry for years. I mean, the basic PC architecture has been fairly static, it’s been stuck in a rut for I’d say a decade. So what’s happened in the last few years? We got a faster bus/memory, PCI Express, that’s basically it.

This will actually allow us to do something genuinely different with the PC architecture, we’ll still keep it open, but we’re going to do interesting and new things that will allow us to reach customers in ways that we haven’t before.

Editors Note: At this point our digital recorder unfortunately ran out of memory and the remainder of the question was answered unrecorded. Chris, Brandon, and Jon ultimately discussed the slide in more detail and Chris basically reiterated that the slide is very general and only provides a basic outline for where AMD and ATI want to go moving forward, and that the combination could involve a mixture of both a discrete CPU, GPU, and chipset for some markets (read: high-end gaming) whereas others may offer a CPU-GPU hybrid which would combine some GPU functionality onto the CPU; this particular application could apply to the general purpose and media-centric platforms from the strategic vision presentation.

As mentioned in the preso this would occur sometime around the 2008 timeframe when AMD makes their transition to 45-nm. The smaller manufacturing process plays a key role in making the CPU-GPU feasible. We also specifically asked them if this merger meant the death of the traditional PC with a dedicated CPU, GPU, and chipset and they were quite adamant that wouldn’t be the case, in the conference call this morning AMD/ATI executives emphasized platforms that AMD hasn’t traditionally done well in (specifically at the corporate PC level and notebooks) but that in no way means that the combined entity will de-emphasize the gaming platform. In fact if anything they feel with the combined resources of both companies they’ll be able to deliver an even stronger product for gamers. We also asked if this merger would lead to the death of the discrete graphics card and ATI assured us that this wouldn’t be the case. The remainder of the interview was recorded with an analog tape recorder.

FiringSquad: How will this merger affect ATI’s discrete graphics plans?

Chris Evenden: ATI is committed to discrete graphics and will continue to blaze a path of customer-centric innovation and market leadership. This merger will only add to our ability to deliver world-class discrete graphics processors.

Again, ATI’s executives and board engaged in a deep strategic review of the company over the past twelve months. We have several overarching goals - >20% p.a. growth, technology leadership, #1 or #2 in all product markets and so on – and we took a long hard look at how we would continue to meet them. Up to two years out we felt that the current path would be successful. But to continue further in the future than that, we realized we needed a big move, a bold move, to be successful. We considered several options, such as making large acquisitions of our own, raising money on the markets to increase investment in R&D and others, and actually decided that an acquisition by AMD was the best fit. So when AMD called, we were already half way to the altar.

In the near term, there is strong growth available to the new company through increased CPU+chipset sales to the notebook and commercial markets. In the medium term, we’ll be able to innovate on the PC platform to produce solutions compelling to certain applications, whether it’s long battery life, media centers, high-end gaming or something completely different. Also medium to long term, we’ll be able to increase our investment in our Consumer businesses, accelerating their growth. And there are many technologies within AMD that will increase our competitiveness significantly. An obvious one is custom memories – at the moment, we design our chips at the cell level, a much larger granularity. AMD engages in a lot of custom design, and the quick win here is reducing silicon area by implementing custom memory designs. Another obvious one is AMD’s expertise in low-power application processors, something likely to be very useful to our cell phone business going forward. There are many others we know about and probably many other technologies we can leverage that we don’t even know about yet.