Innovation is Even More Critical During Times of Economic Turmoil – Interview with AMD Graphics Products Group

Even though graphics processing units are very complex these days and take years to develop, just as central processing units, the right GPU at the right time can completely reshape the market in just one or two quarters. This is why it is crucial to make the right business decisions for the next month and right technology decisions for the next two-three years. With these rules in mind, we decided ask ATI, graphics products group of Advanced Micro Devices, a dozen of question regarding current state of the business and future products.

by Anton Shilov
12/24/2008 | 11:42 AM

X-bit labs: Hello, can you introduce yourself to our readers please?


David Cummings: Hi, I’m David Cummings, Director of Product Marketing, Discrete Desktop Graphics, Graphics Products Group, AMD.

Dave Nalasco: And I’m Dave Nalasco, Senior Technical Marketing Manager, Graphics Products Group, AMD.

X-bit labs: Some of our readers probably do not know about your career and would be interested to learn more about your role at ATI and now AMD, can you talk about that?

David Cummings: I joined ATI 10 years ago. The graphics industry has certainly changed a lot since then! I have worked in several roles in ATI and then AMD, most notably as Director of Notebook GPU Marketing prior to moving to the Desktop group in 2007.

Dave Nalasco: I also started working at ATI 10 years ago, originally as an ASIC designer (on our Rage II+ and Rage Pro chips, for those who remember that far back), before joining the marketing team. Today I help define new features and products, and also explain AMD’s GPU technology to anyone who is interested.

Graphics Adapters Business and Market Today

Throughout the last ten years graphics processors gained extreme amount of performance and features, but the market of discrete graphics processing units (GPUs) did not actually increase tangibly since the emergence of integrated graphics processors (IGPs) did not allow it to.

Now that the economic crisis is in full swing and it is likely that many will start to save on GPUs and buy IGPs, it is very interesting to know, what does the world’s second largest supplier of discrete graphics chips thinks on the matter and how it will act. Moreover, both ATI (AMD’s graphics products group) and Nvidia have to keep an eye on Intel, the king of IGPs, which is prepping its code-named Larrabee standalone graphics processor.

X-bit labs: How is the global economic crisis impacts AMD/ATI graphics cards business?

David Cummings: I think it is safe to say that most if not all semiconductor companies are feeling the effects of a global slowdown. We [AMD] recently (December 4th) updated our fourth quarter outlook, estimating that our expected revenue for the current quarter would be down relative to our third quarter. I can’t offer you more than that as we are currently still in our Q4 and will not be reporting our Q4 revenues until January 22nd, 2009.

X-bit labs: How do you think the global economic slump will affect further development of the market?

David Cummings: Given recent history, slowdowns would appear to be part of the global economy. During a slowdown, it is even more important for companies to continue innovating in their respective markets, and developing new products based on new technologies and new benefits to the end user. We did that with our ATI Radeon HD 4000 series, completely redefining the performance per dollar equation for graphics cards, and it has paid off in increased sales and market share gains.

X-bit labs: Do you think that because end-users reduce their spending, companies like ATI or Nvidia will concentrate on development of entry-level or mainstream graphics processing units and not high-end GPUs?

David Cummings: We have our most compelling, complete graphics product stack in the company’s history. Our ATI Radeon HD 4000 series is bullet proof in that we have something for everyone, from our value priced ATI Radeon HD 4350 through to the ATI Radeon HD 4870 X2, a card that continues to be the fastest single graphics card on the market today. If we find sales are slowing down in a given price band and picking up in another, we are ready.

In terms of determining future products, the engineering efficiency approach we adopted with the ATI Radeon HD 3000 series and perfected with the ATI Radeon HD 4000 series (also referred to as our “sweet spot strategy”) gives us a great deal of flexibility in building that one graphics chip that can then serve as the building block for our entire product stack. This approach gives us tremendous latitude in building the right products for the right market segments.

X-bit labs: Will ATI/AMD reduce research and development (R&D) spending because of the economic crisis?

David Cummings: As mentioned earlier, innovation is even more critical during times of economic turmoil. We intend to maintain the pace of research and innovation that has enabled us to maintain our technology leadership in the graphics market.

X-bit labs: Quite a lot has happened in the industry in the recent years, do you feel the changes in terms of demand on the market of graphics cards in general as well as requirements of end-users?

David Cummings: We definitely see the changes and in fact those changes were a big part of ATI and AMD coming together. First, we wanted compelling platform (CPU, GPU, chipset) solutions to offer our OEM and channel partners. Second, we wanted strong integrated graphics products. And third, we wanted to continue delivering exceptional discrete graphics products. We have succeeded in all three areas.

X-bit labs: What are the primary goals that you think AMD’s graphics products group have to achieve in the short term?

David Cummings: We must continue to offer compelling graphics products that deliver on our promise of the ultimate visual experience. That means an immersive game experience as well as a great multimedia experience.

X-bit labs: The market of discrete desktop graphics cards has been very stable in the recent years and with several exceptions TAM (total available market) was in the range of 20 – 22 million units per quarter (according to Jon Peddie Research). Do you think it will stay on than level, or will increase or decrease?

David Cummings: There continues to be a huge opportunity for discrete graphics. First, it is a common misconception that PC gaming is on the decline. Recent data that includes digital downloads and online gaming indicates that the PC remains an important gaming platform. At the same time, the trend toward more realism in games continues and each incremental improvement in game realism demands more graphics horsepower. Second, the price of big screen monitors and televisions continues to drop, so more people than ever before are buying large format screens for gaming, watching multimedia content, and working.  Bigger screens tend to draw attention to limitations of lower end graphics so more and more people are paying attention to graphics.

According to Jon Peddie Research, the demand towards performance-mainstream and enthusiast-class graphics cards will be increasing, whereas basic graphics boards are losing popularity because of higher-performance IGPs and growing demands among end-users.

X-bit labs: Unlike the TAM, average selling price (ASP) of a graphics card has been fluctuating dramatically in the recent years, which probably affected both ATI and Nvidia materially. Do you think that ASPs will continue to rise and fall or will stabilize at a certain level at some point in future?

David Cummings: ASPs are very much a function of supply and demand so it is tough to answer that question with any degree of accuracy. I know that performance per dollar will continue to go up and that will ultimately be good for the end-user as they get more graphics performance for their money with each new generation of graphics products.

X-bit labs: What are the markets that ATI would target on the first place in terms of revenue increase: discrete desktop, discrete mobile, professional cards?

David Cummings: We arguably have the best products in all three areas, however in terms of untapped market share, our biggest opportunity is in professional graphics. Current market share numbers do not at all reflect the competitiveness of our professional graphics product stack. We intend to change that.

X-bit labs: Can you talk about perception of standalone graphics cards in different regions around the globe? Which countries consume more high-end graphics cards and which are all about entry-level products?

David Cummings: There are definitely regional differences. As an example, the Japanese market tends to prefer console gaming over PC gaming so our sales mix is very different there when compared to a country where PC gaming is more prevalent, like South Korea.

X-bit labs: Do you think that there is a market for discrete graphics cards in developing markets, such as Africa or India?

David Cummings: Absolutely, there is a market for discrete graphics wherever personal computers are sold. Discrete graphics adds to the overall experience and in many cases unleashes aspects of the computer that cannot be fully experienced even with today’s more robust integrated graphics products. While a better video game experience is probably the most obvious benefit, the entire user experience can be improved, from an improved user interface through to photo editing and viewing multimedia content.

X-bit labs: ATI got number of new add-in-board (AIB) partners this year – Gainward, Force 3D, XFX just to name a few. Was it a result of very competitive ATI Radeon HD 4000-series GPUs, or is it because ATI/AMD changed its way of working with AIB companies?

David Cummings: We remain committed to working with a small group of AIB partners as that lends itself to deeper engagement with each partner. That being said, the tremendous success of the ATI Radeon HD 4000 series made it necessary to add more partners to meet demand. We chose those partners carefully.

X-bit labs: Do you think that chipsets with integrated graphics processors (and eventually central processing units with built-in graphics cores) will impact the TAM of discrete graphics cards going forward?

David Cummings: Integrated graphics has already had a significant impact on discrete graphics and now represents the lion’s share of graphics sales by unit volume. In fact, AMD has a vested interest in integrated graphics as we not only make chipsets with integrated graphics, we are actively and aggressively developing graphics cores that reside on the same piece of silicon as the CPU.

AMD's highest-performance core-logic with ATI Radeon HD 3300 integrated graphics processor

If anything, I believe that on the other side of current economic uncertainty, the market for discrete graphics will remain healthy for the foreseeable future. As mentioned earlier, PC gaming is alive and well. Visual computing continues to grow in importance, as demonstrated by everything from user interfaces to new applications like Microsoft Photosynth and Seadragon. Additionally, new applications of GPUs such as transcoding and stream computing are likely to have a positive effect on discrete sales.

X-bit labs: What do you think ATI/AMD should do in order to gain market share, revenue and profitability amid the economic crisis?

David Cummings: Continue to innovate.

X-bit labs: Do you think that the launch of Larrabee GPUs by Intel will fundamentally change the market of discrete GPUs?

David Cummings: To date, Intel has been offering controlled glimpses of their Larrabee vision. Based on information they have released, we know Larrabee will be a multi-core x86 architecture and that it will target the personal computer graphics market. This may or may not happen in the 2009/2010 timeframe.

First, I’d like to bring up the saying, “When all you have is a hammer, everything looks like a nail.” Intel is taking their existing x86 technology and attempting to apply it to what they see as a market they have yet to tap – personal computer 3D graphics. The challenge they are going to have is in convincing the existing PC ecosystem of software developers, OEMs, system integrators and many other stakeholders to add the additional cost of Larrabee to their existing BOM, or substitute Larrabee in place of existing graphics solutions that have evolved over the last two decades, in response to the needs of this very same ecosystem. The reality is that our discrete graphics products are incredibly powerful, multi-core processors in their own right that have developed in lockstep with the needs of the hardware and software vendors. I like our chances.

X-bit labs: Unlike graphics cards, TAM of video game consoles has been showing dramatic growth levels. Do you think that video game consoles also impact the market of graphics cards if not represent a threat to gaming PCs in general?

David Cummings: I think the key here is that while PC gaming is showing slower growth relative to video game consoles, it is growing and that is good for the graphics market.

As you probably know, AMD designed the graphics powering both the Xbox 360 and the Nintendo Wii. At the same time, we have a team that works very closely with the software development community. From this unique vantage point, we have watched the evolution of today’s video game market. What we have witnessed is the addition of thousands if not millions of new video game players, thanks to the Nintendo Wii and the many mobile devices that have provided another vector for delivery of video games. The casual gamer has never been so well served. At the same time, many of the gamers we would classify as mainstream and enthusiast now own both a gaming PC and an Xbox 360. Often, this group will prefer to play one genre of games on the Xbox 360 and another genre on the PC.

X-bit labs: Do you think that since the most popular video game console – Nintendo Wii – has very basic graphics capabilities, this will lower demands towards high-quality graphics in PC video games and consequently will impact sales of advanced graphics cards.  

David Cummings: I think the Nintendo Wii is a brilliant product that took a novel approach to the game play experience, an approach that has been incredibly successful. If anything the Wii shows that immersive game play is about more than just graphics. I think the Wii will ultimately be good for the video game industry, both console and PC, as it has introduced non-gamers to this entertainment medium.

X-bit labs: Do you think that emergence of high-definition video standard will help the PC graphics adapter market to grow?

David Cummings: Definitely. It already has. There is a growing audience of PC gamers with their [Blu-ray enabled] systems hooked up to 52” HD television sets. As prices continue to drop on LCD televisions, those numbers will explode.

Graphics Processors Strategy

X-bit labs: Back in the days both ATI and Nvidia introduced new high-end lineups of products once in six month and refresh the mainstream lineups once in a year. This time we see that both companies began to update high-end lineups relatively rarely. Can you explain why did this happen? Perhaps, because of the complexity of modern GPUs?

David Cummings: We’re still roughly on a once a year cadence. In fact, the time between the introduction of the 3870 X2 and 4870 X2 was around eight months.

X-bit labs: After AMD and ATI Technologies merged in late 2006, all graphics adapters vendors (ATI/AMD, Intel and Nvidia) started to tangibly increase performance of IGPs. Is the reason for boosting IGP performance in the fact that standalone graphics adapters [both desktop and mobile] are losing the importance, whereas integrated graphics cores is gaining the significance, or the cause is completely different?

David Cummings: Integrated graphics have succeeded because it has a place in the market. IGP typically only overlaps with value segment discrete graphics so by improving the performance of our IGP solutions, we are only marginally cannibalizing our own discrete products. At the same time, we want to offer the best possible product stack and IGP is part of that strategy. By offering better performance than competitive IGP offerings, we are differentiating our products from those of our competitors.

Despite of relatively low power consumption of its latest GPUs, AMD still does not have Mobility Radeon HD 4000-series products. The lag between the launch of desktop ATI Radeon HD 4800 in June ’08 and potential release of Mobility Radeon HD 4800 at Consumer Electronics Show or, perhaps, CeBIT 2009 is going to be dramatic. This raises questions about the strategy of ATI in the field of discrete GPUs for notebooks.

X-bit labs: Does the fact that Mobility Radeon HD 4000-series is still not introduced mean that it is less important for ATI to create discrete mobile GPUs or chipsets compared to desktop ones? Or maybe, it means that mobile IGPs are simply dramatically more important compared to GPUs for mobile computers?

David Cummings: It is a function of the market cadence. Our sweet spot strategy outlined earlier actually makes it easier than ever before for our GPUs to move from desktop to mobile, however the timing of our HD 4000 series launch precluded notebook designs shipping in 2008.

Indeed, it is crucially important for both ATI and Nvidia to introduce mobile graphics processors along with new mobile platforms from AMD or Intel, which means that graphics processors themselves should be ready well before the formal introduction so that to allow notebook designers to develop and tailor new mobile computers for new CPUs, chipsets and GPUs.

X-bit labs: After ATI left the market of IGPs for Intel processors, isn’t it in AMD’s interests to promote the discrete GPUs for both desktop and mobile PCs at any cost just to sustain the market share?

David Cummings: As mentioned earlier, given the tremendous growth of IGP, it is clear that it has its place in the market. Our IGP solutions allow AMD to provide our partners with exactly the platform or system specifications they want, often in all-AMD systems. We believe this customer-centric approach delivers significant value to our customers.

X-bit labs: Both ATI and Nvidia in the past tried to introduce cheaper/faster versions of high-end GPUs made using newer process technologies (Radeon HD 3870/RV670 and Nvidia GeForce 9800/G92, the lower-cost versions of Radeon HD 2900/R600 and GeForce 8800/G80 are among the best examples). Will this trend continue? Or does it make more sense to create new high-end products while continue selling the previous-gen high-end chips at lower prices?

David Cummings: With both the ATI Radeon HD 3000 series and HD 4000 series, we launched the families by starting with our performance and enthusiast products, and shortly thereafter rolling out value, mainstream and ultra high-end products. That pattern is likely to continue.

Graphics Processors Design

X-bit labs: Does the approach to GPU design change with different generations of GPUs?

David Cummings: GPUs continuously evolve on all levels, from process nodes and architecture to materials. Our HD 4000 series differs significantly from our HD 3000 series, even though they both utilize the 55 nm process.

X-bit labs: What are the primary constraints for GPU performance today, power consumption, memory bandwidth, anything else?

David Cummings: Power, and its byproduct, heat, are probably the two biggest challenges. They were the two biggest reasons behind our move away from simply creating bigger and bigger chips and instead focusing on smaller, highly efficient performance segment GPUs that could easily be adapted to enthusiast, mainstream and value designs. That choice has led to the success we’ve enjoyed with the ATI Radeon HD 3000 series and the HD 4000 series.

ATI Radeon HD 4800 also known as RV770 graphics processor

X-bit labs: ATI was heavily criticized for not implementing hardware FSAA resolve into the R600/RV670 generation processors. But as we see now, Nvidia also decided not to improve render back ends (RBEs) in the GeForce GTX 200 (G200) generation and Intel wants to get rid of hardware RBEs in Larrabee. Perhaps, ATI will do the same eventually and the move on the R600/RV670 was right, but was made too early?

Dave Nalasco: One lesson we learned from our previous generation of GPUs was the level of importance our customers place on maximizing performance at high image quality settings. Specialized hardware is more efficient than general purpose processors for handling tasks like anti-aliasing, and this was a big area of focus for the HD 4000 series design. Besides providing very fast and powerful fixed function MSAA, we also managed to include a high degree of flexibility that allows us to drive image quality higher, with features like Adaptive Anti-Aliasing and Custom Filter Anti-Aliasing (CFAA). DirectX 10.1 support allows us to expose this programmability to developers through an industry standard API.

X-bit labs: What level of computing performance is needed to safely remove fixed-function hardware from GPUs? When would you expect this level of performance to be achieved? Is two, three or four generations the right guess?

Dave Nalasco: In general, I don’t expect this will happen in the foreseeable future. Although we will certainly see GPUs handling a much broader range of applications beyond graphics and video processing over the next few years, they still have to perform these important specialized functions. Well-designed fixed function hardware will continue to provide more speed and efficiency for these tasks than general purpose processors.

X-bit labs: How much time is it spent to create a “shrunk” version of a graphics processor, e.g., transit from 65nm (R600) to 55nm (RV670) with minimal design shifts?

Dave Nalasco: There is no simple answer to that question. It depends on many factors, including the maturity of the new process technology, the amount of tweaking required to achieve performance and power targets, and the level of priority and design resources we assign to the project.

Future Graphics Products

X-bit labs: It is obvious that it is crucial to be the first with introduction of next-generation products. ATI was first-to-market with DirectX 9 generation and won the round, whereas Nvidia was first to introduce DirectX 10 GPU and won this time? Are you confident that you will be the first to launch DirectX 11 GPU?

David Cummings: Currently, we are the only vendor supporting DirectX 10.1, another first for us. As all features of DirectX 10.1 will be supported in DirectX 11, many developers are starting to develop for DirectX 10.1 as a preparation for the eventual move to DirectX 11. We feel very confident that we will make that transition to DirectX 11 right along with the developers.

X-bit labs: ATI has explained the reasons why it decided to switch to multi-GPU strategy for high-end market a number of times. But modern multi-GPU configurations are homogeneous multi-chip setups. Perhaps, it makes sense to look at so-called heterogeneous multi-chip configurations (e.g., one chip does load-balancing, other chip computes, etc)?

David Cummings: Certainly we are considering our options with respect to heterogeneous computing, however we have not made any decisions at this point.

X-bit labs: ATI has been heavily criticized for its 3:1 and then 4:1 texture to arithmetic performance ratio for DX9 and DX10 generations of products. But can we expect 5:1 or even 7:1 ratio for DirectX 11 generation of products, considering the fact that DX11 will have more computational power-based features?

Dave Nalasco: There is always some guesswork involved in determining the optimal configurations for GPUs, especially when programming models change and evolve. That said, we do an extensive amount of up-front simulation and characterization, and that is something we are improving with each generation. I think our process yielded some great results in our most recent designs, and I expect that trend will continue in our upcoming products.

X-bit labs: What levels of performance (in TFLOPs) should we expect from GPUs in the next 2-3 years? We do remember that Phil Hester (ex-CTO of AMD) promised petaFLOP performance in the next decade :).

David Cummings: More :).

X-bit labs: What do you think about external graphics cards? ATI and Fujitsu Siemens have launched one called Amilo Booster, but it has numerous limitations and is only supported by one laptop by one vendor…

David Cummings: I think the Fujitsu implementation of our XGP technology was a very good beginning for a technology that has tremendous potential.

X-bit labs: Do you think that sometime in the future there will be external add-in graphics boards for desktop?

David Cummings: The notebook is the logical opportunity for external graphics as it has the dual need of being highly power efficient when on the move and yet having the same demands as a desktop placed on it when in a home or office environment. I could possibly see a desktop utilizing external graphics as a way of incrementally increasing graphics performance.

X-bit labs: ATI has been seriously investing into new memory technologies, which helped the company a lot back in the days. Will you continue to invest further into the evolution of GDDR? When can we expect the GDDR6 standard?

David Cummings: We are still actively involved in the development of GDDR5 memory technology. The payoff of that investment is clear in that we were the first (and still the only) provider of GPUs that take advantage of this standard. GDDR5 is still early in its lifecycle, and we believe there is plenty of headroom available to continue improving its performance and power consumption before a new standard is required.

X-bit labs: What are the reasons why you decided to drop ring-bus memory controller? Can we expect its return in the future?

David Cummings: The 512-bit ring bus interface used in the HD 2900 required a lot of room. Returning to a smaller memory interface allowed us to keep the overall size of the HD 4800 series down, an important requirement with more and more HTPC and SFF-based systems on the market.

X-bit labs: ATI advertised the “Sideport” communication bus on the ATI Radeon HD 4870 X2, but the technology is still disabled. Can you re-emphasize potential benefits of Sideport when enabled? Can you reveal the reasons why it is still turned off and when, if ever, it will be activated?

Dave Nalasco: The ATI CrossFireX sideport provides an additional communication path between GPUs, with the potential to increase the available bandwidth. Improvements in our CrossFireX drivers have since reduced the amount of bandwidth required to achieve expected levels of performance scaling, which obviated the need to enable the sideport on ATI Radeon HD 4870 X2 boards. We still have the option to enable it at a later date if we encounter cases where the additional bandwidth can provide a performance benefit.

X-bit labs: Apart from Sideport and unified memory [for two GPUs], are there ways to improve efficiency of homogeneous multi-GPU technologies?

Dave Nalasco: We are continuing to work on our CrossFireX technology, and we believe there are still opportunities to improve efficiency with the current hardware architecture.

Mainstream GPGPU? It Is Already Mainstream... Nearly!

X-bit labs: Do you think that GPGPU technology will ever go mainstream? Does it make sense to invest into that?

David Cummings: The GPU has a trick up its sleeve – it is massively parallel and as such, is very good at crunching through large datasets. It is much faster than the CPU in this respect. I would have to say that I think the GPGPU has a very good possibility of gaining traction both in commercial and consumer applications.

Dave Nalasco: This is what our ATI Stream initiative is all about. We believe that GPUs offer massive amounts of untapped performance potential for everyday tasks. We have made it a priority to support industry standards, build an infrastructure of development tools, and bring useful examples of this technology to the widest possible audience.

X-bit labs: Will you enhance your GPUs with general-purpose computing in mind going forward?

David Cummings: We started doing so with the HD 3000 and 4000 series and will continue to do so moving forward.

X-bit labs: When AMD acquired ATI Technologies back in 2006, many said that GPUs were floating point co-processors of our days. Given the recent events on the CPU and GPU markets, it does not seem so now. What is your opinion today?

Dave Nalasco: Not sure who said that, but I think GPUs have always been much more interesting than floating point coprocessors. Unlike the old coprocessors, GPUs have become a fundamental part of every PC, and basic functions like your operating system desktop are designed on the assumption that one is present. What is interesting today is that we are discovering how well-suited GPUs have become for accelerating applications that floating point coprocessors were used for back in the 1980s, like databases, spreadsheets, and numerical analysis. Our ATI Stream technology provides big opportunities to speed up tasks like these, by enabling a better balance of workloads between the CPU and GPU cores in a system.

X-bit labs: Thank you David, thank you Dave for your interesting and detailed answers and we hope to talk to you again.