1. AMD Financial Analyst Day Products Group Rick Bergman, Senior Vice President & General Manager November 11, 2009
2. Yesterday GPU CPU 110 million transistors @150nm 2D and 3D gaming Nascent video processing 105 million transistors @130nm Compute tasks including video decode
3. Today TeraFLOPS-class GPU Multi-core CPU 2.15 billion transistors @40nm 3D OS Multi-panel HD gaming Full HD video and audio 758 million transistors @45nm Multi-tasking Most compute tasks
15. Unprecedented Server Performance Gains AMD Opteron™ 244 AMD Opteron 250 AMD Opteron x75 (dual-core) AMD Opteron 285 (dual-core) AMD Opteron x356 AMD Opteron x384 AMD Opteron 2435 “Interlagos”* “Magny-Cours”* 2010 is projected to be the beginning of unprecedented leaps in server performance-per-watt for AMD * “Magny-Cours” and “Interlagos” data is based on AMD projections
16. Six-Core AMD Opteron™ Processor Draws less power at load and idle than competitive offerings* Six-core virtualization performance leader Sub 7-watt cores without compromise Compatibility designed for faster upgrades Powers world’s first 2 petaFLOPS supercomputer * Power numbers based on 3rd party testing in http://it.anandtech.com/IT/showdoc.aspx?i=3653&p=9
18. 2010: Target the Volume Server Market 4 Socket, 0.4M $10K+ “traditional 4P” market Volume 4P server $6-10K Redefine the market Fully-featured 2P G34 delivers more value with no compromises on features 2 Socket, 5.0M New C32 platforms take true server features to environments placing a premium on power-efficiency 1 Socket,1.6M <$500 “desktop as a server” market Source: IDC Workload Study and Server Forecast
23. New levels of valueand power efficiency1P Platforms ~20% of Market* *AMD internal estimates of total server market as of Q309
24.
25.
26. “San Marino” Platform: New Levels of Power Efficiency and Value 4- and 6-core “Lisbon” processors 1- and 2-socket Platform optimized for high energy efficiency Available with sub 6-watt CPU cores Ahead of schedule
27.
28.
29. New levels of value and power efficiency*AMD internal estimates of total server market as of Q309
31. 2010 Priority: Win with Superior PC Platforms Visual technologies leadership Optimize for most popular usage areas Continued battery life improvements
33. AMD Mainstream Notebook Platform Battery Life Core Silicon Power Note: Based on AMD notebook reference design Assuming 6-cell Li-Ion battery, 55Whr for 2008-2009, 62Whr for 2010-2011 Battery Life (Resting)
34. Mainstream Notebook Platform: "Danube” Available H1’10 45nm quad-core performance DX®10.1 IGP: performance and efficiency Designed for 7 hours battery life DirectCompute and OpenCLTMapplication acceleration DX®11 mobile discrete graphics: only from AMD Ahead of schedule
35. 3rd Gen Ultrathin Notebook: "Nile" Available H1’10 45nm dual-core mobile processors Designed for 7+ hours battery life DX®10.1 IGP: performance and efficiency DX®11 mobile discrete graphics: only from AMD Enhanced UVD capabilities Ahead of schedule
41. 2010 2009 2011 Segment AMD Desktop Platform Roadmap 45nm 32nm “Dragon” Platform AMD Phenom™ II CPUsUp to 4 CPU CoresDDR2/DDR3 AMD 790FX/GX chipsetATI Radeon™ HD 4000 Series discrete graphics “Leo” Platform “Thuban” CPUsUp to 6 CPU CoresDDR3 RD890+SB850 chipsetDX® 11 discrete graphics “Scorpius” Platform “Zambezi” CPU 4/8 CPU “Bulldozer” CoresDDR3AM3 Socket Next-Gen ATI discrete graphics Enthusiast “Lynx” Platform “Llano” APUUp to 4 CPU CoresDX® 11 GPUDDR3 “Pisces”/”Kodiak” Platform AMD Athlon™ II CPUsUp to 4 CPU CoresDDR2/DDR3 AMD 785G chipsetDX®10.1 IGP “Dorado” Platform AMD Athlon II CPUsUp to 4 CPU CoresDDR3 RS880P+SB810 chipset with DX® 10.1 IGP Mainstream Roadmap subject to change without notice
46. The GPUs must be powerful, scalable, affordable, efficient$200 “Juniper” $100 Q4 2009 Q1 2010 “Redwood” TIMELINE “Cedar” Product schedule subject to change without notice
47. AMD Wins Visual Computing Technology Transitions Unified Shaders 3DC Dynamic Branching HDR & HDR+AA API Support Gaming and Media Technology GDDR3 GDDR4 GDDR5 Memory Technology 130nm 110nm 90nm 80nm 65nm 55nm 40nm Process Technology
50. ATI Radeon HD 5800/5700 Graphics Leadership ATI Eyefinity™Technology ATI Stream Technology DirectX®11 Technology
51. AMD Wins in Mobile Discrete ATI Mobility Radeon™: #1 choice for notebook discrete graphics 200+ design wins from leading OEMs worldwide Best gets even better with upcoming introduction of ATI Mobility Radeon HD 5000 series DirectX® 11 mobile discrete graphics: Q1 2010, only from AMD
53. ATI Stream: Platform Performance Breakthroughs Heterogeneous: Developers leverage AMD GPUs and CPUsfor enhanced application performance Standards Based: Industry-standard OpenCL™ and DirectCompute 11 for cross-platform development High Performance: Massively parallel, programmable GPU architecture delivers dramatic performance and power efficiency Gaming HPC Entertainment Productivity
55. Adobe and OpenCL David WadhwaniGM and Vice President, Platform Business Unit Adobe Systems, Inc. “Adobe and AMD are collaborating to optimize Flash Player 10.1 for AMD platforms. We are excited about the joint effort, which will help accelerate and deliver smooth video playback of rich Flash based Web content on AMD-powered devices.”
56. ATI Stream-Enabled Multimedia Applications MediaShow 5 MediaShow Espresso PowerDirector 8 PowerDirector 7 SimHD™ Plug-in for TotalMedia Theatre Roxio Creator™ 2010 Roxio Creator™ 2010 Pro
57.
58. Pro Graphics Momentum Accelerates 42 10 Record design win numbers for ATI FirePro™ Professional Graphics
59. 2010 AMD Product Priorities Deliver four new winning PC platforms in H1 Increase both notebook performanceand battery life by at least 25% Expand OpenCLTMand DirectCompute 11 developer tools and hardware optimizations Win DX®11 notebook transition Launch 12-Core AMD Opteron™ processor and 6000/4000 series platforms Sample “Bulldozer”, “Bobcat” and 32nm AMD Fusion™ based products to customers
Three key points:The future of computing is AMD Fusion Update on AMD’s Product Strategy and 2010 platforms How we leverage our key differentiator which is our GPU technology
In just the past 5 or 6 years, the PC industry has undergone some remarkable changes. Let’s look at 2003, for example.On the left you see an Athlon 64 processor with 105 million based on 130nm process technology which handled the bulk of the work on the PC. On the right you have a Radeon 9800 GPU, DX9 class with 110 million transistors based on 150nm process technology. At this time the GPU handled highly specialized tasks as the applications were not written to take advantage of the inherent GPU power.
Let’s look at today’s compute platforms:You have a Phenom II with 758 million transistors on 45nm process technology on the left On the right you see a 5870 DX11 GPU with 2.15 billion transistors on 40nm process technology. Today, with the emergence of visual computing, you see more work than ever before for the GPU. Especially with, arguably for consumers, the most important workload: video.The explosion of HD video and now HD gaming, means the GPU matters more than ever in the PC platform. More user-generated content puts more of the work onto the GPU such as video processing and rendering and 3D user interface.The era of visual computing is already becoming more about mobility and being able to do more of what I’ve just described on the go. However, users do not want more compute capabilities at the expense of battery life or smaller form factors.Favoring one component over the other or taking a niche approach to balanced visual computing platforms does not meet the needs of the mass market. Usage scenarios favor a combination of GPU/CPU balance and low power..
The need for this optimal energy-efficient balance of CPU and GPU represents the beginning of a new era of computing in 2011.The Fusion of CPU and GPU compute power is what the next chapter in visual computing requires – a powerful visual computing experience at home or on the go without compromise. Our AMD Fusion™ design is driven by mobility and is based on a low-power visual compute architecture that will enhance active and resting battery life while increasing both CPU and GPU performance. This is the culmination of the vision of ‘One AMD’ and only AMD can deliver the GPU and CPU combination that will be the future of computing
As you are well aware, there’s an explosion of online video happening and its something we all use on a daily basis nowThere’s also a wide range of applications that are now able to leverage the parallelism that GPU deliversYou can see that today’s applications are coming together to leverage the AMD Fusion capabilities.
A year ago we told you our plans for 2009 and as you can see, we did what we said we were going to do … and then some. We successfully transitioned to 45nm across all of our CPUs We worked with HP and others to help create the ultrathin notebook category and have significantly improved our battery lifeWe handily won two big inflection points in graphics - the DX11 and 40nm transitions. We continue to be the only company to offer DX11 solutions. More great progress with OpenCL and other developer tools We delivered the first six-core processor for 2P servers and higher to market … 5 months ahead of scheduleAnd looking ahead, we’re on schedule to bringing our first APU and 32nm solutions to market in 2011While 2009 was a great year for execution, we’re not resting on it. We must go faster
Let’s shift to our product strategy: We will continue to deliver winning platforms Leverage the power of our leading GPU technology And simply put, we will win with AMD Fusion CPU, GPU, Chipset & software optimizations
The launch of “Istanbul” in 2009 was the beginning of what will ultimately be the biggest leaps ever in performance for AMD Opteron processors.
Let’s talk about where we are today with our server platforms. According to IDC, power is the number one buying criteria today and Six-Core AMD Opteron processors, which we launched five months ahead of schedule, are clearly delivering in this area. We draw less power than competitive offerings and we brought to market 40-watt six-core processors … that’s less than 7-watts per core, all with the same features as our highest-end processors. End-customers and OEMs have seen tremendous value in the fact we’ve enabled them to upgrade to six-cores within the existing socket infrastructure. Next week, the latest Top500 Supercomputing list will be released and Six-Core Opteron is expected to make an impact. In fact, the Department of Energy’s “Jaguar” supercomputer at Oak Ridge National labs was upgraded Six-Core Opteron processors and has seen peak performance of more than 2 petaflops. Also, the National Science Foundation “Kraken” system at Oak Ridge has also been upgraded with six-core processors as well. Take a look at this video from the folks at Oak Ridge on the Kraken supercomputer and their upgrade experience.
You heard that last part about the Cray and AMD working together on heterogeneous computing for HPC. Interestingly enough, a very large cluster based on 5120 ATI RV770 GPUs is being deployed in China. We believe the Tianhe-1 cluster will be the first Petaflop GPU cluster, as well as being AMD’s first large-scale ATI Stream deployment in technical applications. Pay close attention next week to see where this lands on the Top 500 list.
Let’s talk about 2010.Earlier this year, we said we would bring to market two new platforms in 2010 that are designed to meet specific end-customer needs in the volume server market. Traditionally you’ve seen AMD have a strong presence in the traditional 4P market and higher. While we will still continue to deliver the traditional 4P solutions, our focus will be on gaining market share and targeting the volume or “sweet spot” of the server market.
Our server sweet spot strategy is centered around customer needs that are gravitating to two key areas:More performance and expandability as well as more power efficiency and better value. Today’s “3 platform world” of traditional 1P, 2P and 4P servers does not square up easily with customers’ 2 key needs. So we are changing our products to match the customers’ needs.Today’s 2P products are forced to straddle the worlds of higher performance as well as lower price because they cover such a large portion of the market. While today’s 4P products include the “4P tax”, a relatively higher price despite utilizing the same silicon. This prevents the 4P market from growing; and in reality it is shrinking. We believe there is an opportunity to redefine this space, introduce a new scalable “value 4P” product that allows customers to deploy a “2P+” product, far more scalable than today’s 2P products, yet at a more aggressive price than traditional 4P servers. With a single platform that can deliver clear scalability from 2P up through 4P while maintaining an uninterrupted continuum of value from low to high, the Maranello platform can meet customers’ needs for performance and scalability Today’s 1P market generally features compromised servers with more desktop-like features instead of a true server platform. Today’s value 2P servers have a hard time competing because they need to support all of the features at the top end of the 2P range, making them more expensive than they need to be for many of the “value” workloads. And the need to support higher end features creates a platform that consumes more power than it needs to. By simplifying the platforms, we can deliver a focused 1P/2P solution that delivers great value and lower power consumption
By utilizing common cores, common chipsets, common BIOS code, common drivers and common memory, AMD can deliver a set of platforms that makes it easier for OEMs to develop platforms and easier for customers to manage and maintain long term.Outstanding performance (in the way of performance/watt and price/performance), unprecedented value as well as tremendous consistency all come together to make AMD’s new platforms second to none.
The “Maranello” platform, which continues to track ahead of original schedules, will feature:8- and 12-core x86 “Magny-Cours” processorsIncredible efficiency and threading for demanding data-intensive workloads100% more memory channels than current Opteron processors, and 33% more memory channels than the competition. We have also enhanced the integrated memory controller. All of this results in huge memory throughput gains. You will also see a host of new power efficiency and virtualization featuresPlatform consistency between series’ and platform generations
Let’s take at closer look at the DDR3 memory-based “San Marino” and “Lisbon” processors. Slated for the second quarter of 2010 and also ahead of original schedules. “San Marino” which is optimized for power efficiency will feature the “Lisbon” will be available in 4- and 6-core processors You will see low-power and ultra-low power platform choices with sub 6-watt x86 CPU cores Unlike our competitor, we will not make compromises on feature sets in order to lower the processor wattages. San Marino will be aggressively priced and consistent with Maranello to ensure future generation compatibility
Moving into 2011, both platforms will feature the Bulldozer architecture-based processors. Maranello will offer “Interlagos” at 12- and 16-cores while San Marino will feature “Valencia” at 6- and 8-cores. Both Interlagos and Valencia will drop into the existing G34 and C32 sockets, respectively. You’ll hear more from Chekib in a bit about the Bulldozer architecture.
For AMD in 2010, our priority is to win with superior PC platforms. It’s starts with leveraging our visual technology leadership. Focusing on how customers use our products And striving for that “all-day” computing experience
Now we have a quick video to show you which helps demonstrate the visual technology leadership AMD offers. Here you can see an innovative software partner, in this case ArcSoft, exploiting the superior IGP capabilities in an AMD platform to deliver standard upscale effects to standard definition video playback. You’ll note how much more defined the water spray is from the skier on the right than the one on the left. The greater depth of field in the waves, greater contrast and sharpness, etc.
I mentioned as part of our 2010 priorities we are striving for “all-day” computing. Here you can see we have made tremendous progress in our resting battery life improvements in 2009 and expect to continue to do so into 2010 and beyond.
Moving into our 2010 client platforms, we start with Danube, which is scheduled to be available in the first half of next year and will deliver new levels of multi-tasking to the mobile market with 45nm quad-core processors now part of our mainstream notebook platforms. Also, you get DX10.1 IGP with DX11 mobile discrete option, ONLY from AMD. We also bring the resting battery life up to 7 hours here.
We are ahead of schedule in bringing to market the 3rd generation of the platform we pioneered with our partners – the ultrathin notebook platform. In first half of 2010 we will have 45nm dual-core processors in the “Nile” platform which is designed for 7+ hours of resting battery life and brings new levels of visual computing to this rapidly growing market.
Here you can see our enthusiast and desktop platforms, both available in the first half of 2010. Our “Leo” platform will feature the six-core “Thuban” processor for enthusiasts requiring mega-tasking capabilities and DX11 graphics with ATI Eyefinity for eye-definition gaming experiences. “Dorado” will give consumers unbeatable value with the option of dual, triple and quad-core processors and the world’s best integrated graphics.
Here you can see a good overview of our “sweet-spot” strategy for graphics. Starting with “Cypress” which is our Radeon HD 5870 cards we launched back in September. Then we roll out its derivatives “Hemlock” which will be available next week and will deliver 5 TeraFLOPS of performance and the recently launched mainstream performance card “Juniper”. Moving further down into the mainstream, you have derivatives of “Juniper” which will launch in Q1 of next year, “Redwood” and “Cedar”. All of these cards available within a 3-4 month timeframe, all DX11 capable and all validating our sweetspot strategy.
AMD has a long history of innovation and being the first company to embrace new technologies.We were first to market with DirectX 9, DirectX 10.1, 3DC (now part of DirectX 10 and is the standard for compressing normal maps), useable Dynamic Branching (key feature for Shader Model 3.0). We were first to show and ship with unified shaders, first with HDR and HDR+AA (every HDR game now supports this)First with GDDR 3, GDDR 4 and GDDR 5 with a more than one year lead over our competition We led six straight transitions to new process nodes, from 130nm low K to the current 40nm process. Six years, seven generations of process technology. AMD has been first every single time.
We are seeing great market momentum for our discrete solutions. Just take a look at our gains in notebook discrete unit share. We now lead with 57 percent market share based on the latest numbers from Mercury Research. Jumping quickly from around 40 percent in Q1 to nearly 60 percent today.
We are seeing great market momentum for our discrete solutions. Just take a look at our gains in notebook discrete unit share. We now lead with 57 percent market share based on the latest numbers from Mercury Research. Jumping quickly from around 40 percent in Q1 to nearly 60 percent today.
Here you can see a quick overview of what our new HD 5000 series cards bring.Eyefinity for new levels of gaming experienceThe Direct X11 APIAnd ATI Stream technology to bring more tasks on to the GPU
I showed you just a minute ago how our notebook discrete graphics are number one in the market today. The best will get even better in Q1 when we bring our Mobility Radeon HD 5000 series discrete graphics to market with DirectX 11.
Now, let’s listen to what some of the industry’s leading game developers have to say about why they are so excited about DirectX 11 and why it’s such an important milestone for gaming.
And speaking of bringing GPU power into everyday usage scenarios, here is Steve Guggenheimer from Microsoft talking about AMD and DirectX 11.
We are working closely with our partner Adobe to achieve Stream optimizations for their extremely popular Flash Player and enhancing user experience around video playback.
Here you see other applications that harnessing the GPU power CYBERLINK for video editingPowerDirector 7: This award-winning video editing software was the first Cyberlink application to use ATI Stream to accelerate HD encodingPowerDirector 8: With this latest version, Cyberlink enables more acceleration with ATI Stream for High-definition video editing including encoding and 20 visual effects and transitionsMediaShow Espresso: This is a firstARCSOFT for video playbackSimHD: This is a plug-in to video playback softwareArcSoftTotalMedia™ Theatre. With ATI Stream SimHD is able to take advantage of the GPU power, bringing close-to-ROXIO for content creationCreator 2010 Standard & Pro: This is a comprehensive software suite of capturing, editing and enjoying personal video, including high-definition content. It leverages ATI Stream technology to accelerate project encoding.
Moving on to professional graphics. You can see in 2008, we didn’t have many design wins to speak of. Moving into 2009, we have been able to secure many more design wins as the workstation market is quickly recognizing the value and efficiency that we can deliver.
Earlier I showed you our achievements in 2009, here you can see our list of product priorities for 2010.
The last component of our product strategy I highlighted earlier is for us to be successful in beyond 2010, we must win with AMD Fusion!
So how do we get to Fusion? It requires speed and direction? It requires AMD Velocity. AMD Velocity represents a shift in our design methodology and product introduction cadence. In bringing about the Fusion era of computing, visual computing technologies will increasingly drive the AMD pace of innovation. The AMD Fusion™ Design Methodology will build on AMD’s already established annual GPU design cycle to achieve a faster innovation pace than AMD previously achieved with a CPU-only development focus. AMD Velocity is designed to deliver performance breakthroughs via teraFLOPS-class GPU compute power in tandem with performance and low-power x86 core options. It is expected to result in clear, compelling platform differentiation for AMD, and the delivery of the best APU on the market every year.
AMD Fusion is designed to deliver breakthroughs in three key areas:PowerAMD’s most power efficient core silicon architecture ever projected to deliver all-day battery life for mobile computing.Optimized for power efficiency in resting mode (idle, display refresh) where notebooks spend much of their time, leads to significantly greater resting battery lifeExtensive power gating throughout the Fusion APUs to turn off power to the parts of the APU that are not being used PerformanceFlexible computing resources to take both scalar and vector computing workloads (balanced platform)Performance Boost feature to maximize system performance within a given power envelope – dynamically allocates performance between CPU and GPU processing cores to optimize performance on demand Cost StructureDesigned using 32-nm manufacturing process technology to optimize performance and costSome Fusion APUs (Brazos) designed to bring unprecedented levels of price-performance to the market
Here is the promise of AMD Fusion that we will deliver – Supercomputing power in a notebook that lasts all day.