Mobile platforms
AMD is committed to introducing two mobile hardware platforms in 2010: One enthusiast mobile, one mainstream mobile. These platforms are known as Danube and Nile, respectively. The big efforts in the upcoming year will be a focus on 32nm GPUs, a transition to DirectX 11 discrete adapters, and improvements in battery life.
Another leaked slide from AMD Japan. (Image credit: ASCII.jp)
AMD's 2010 mobile platform roadmap.
Danube
Purpose: Enthusiast mobile
Date: 1H10
Chipset: RS880M northbridge, SB820M southbridge
CPU: Champlain (Most likely: Mobile-oriented Phenom II X4)
GPU: Manhattan
Leading the way for AMD’s renewed efforts in full-size notebooks, the firm is finally playing catchup to its rival by pairing the Danube platform with the company’s first quad core mobile chip. Not much is known about Champlain, but slides leaked from a Japanese AMD event suggest that it’s a mobile-oriented Deneb (Phenom II X4) in the same way that the company’s current Turion II chips are mobile versions of Callisto (Phenom II X2).
On the chipset front, notebooks are also receiving an overhaul. Danube will be fitted with the new RS880M/SB820M north and southbridge combination for SATA, PCIe 2.0, USB 2.0, HDMI and DVI support. If you were looking for SATA 6Gbps or USB 3.0, you’ll most likely be out of luck until 2011.
Undoubtedly the most interesting aspect of AMD’s 2010 mobile platforms is the introduction of the Manhattan family of GPU cores. Whereas AMD’s desktop division only recently initiated a volume transition to 40nm with the Radeon HD 5000, AMD is already grooming Manhattan for Globalfoundries’ 32nm GPU process.
Globalfoundries’ VP of Manufacturing Systems and Technology Tom Sonderman has previously expressed the company’s interest to serve AMD’s 32nm needs in April saying, “We intend on competing for AMD’s graphics business in the 32nm/28nm technology node.” It appears that interest is playing out, despite opinions to the contrary.
Manhattan is certainly planned as AMD’s trial run at the 32nm node pegged for the next-gen Northern Islands architecture. If it plays out as intended, the entire GPU market would finally follow the cadence of the desktop: Create a new architecture (Evergreen) on an existing process (40nm), shrink it to a new node (32nm, Manhattan), then follow with a new architecture (32nm, Northern Islands) once the node has been mastered. It’s a very wise strategy, and should avoid the pain of trying to intro a new architecture on a new node, like NVIDIA is experiencing with Fermi.
Another interesting perk of switching to Globalfoundries for 32nm bulk silicon is the company’s half-node 28nm process. The result of efforts in the IBM Technology Alliance, architectures produced by Glofo can be shrunk to 28nm in 4Q10 without a redesign.
This is a very ambitious plan. Should the switch to 32nm bear out on schedule, Manhattan parts based on the new node will begin appearing at the very end of the year.
However, Danube should launch in the first half of 2010, which means Manhattan will certainly not be launching at 32nm. Instead, the company has confirmed that it is preparing a lineup of Manhattan parts based on the proven 40nm process.
The family will contain a trio of discrete DirectX 11-ready parts known as Broadway (enthusiast), Madison (performance) and Park (mainstream). It is unknown what branding, clock speeds, core, or memory configurations will be used, but 5000 and 6000-series brand names are equally plausible.
The Manhattan family of discrete mobile GPUs.
Lastly on the GPU front, the RS880M contains a DirectX 10.1 IGP based on the RV770 (Radeon HD 4000) architecture for those who neither need nor want the horsepower of a discrete GPU.
Nile
Purpose: Mainstream mobile
Date: 2H10
Chipset: RS880M northbridge, SB820M southbridge
CPU: Geneva (Most likely: Mobile-oriented Phenom II X2)
GPU: Manhattan
The Nile platform is a spitting image of the Danube in terms of architectural families, but it will differ in terms of the power offered by the parts sourced for processing and graphics duty. Where Danube offers a mobile-tuned Phenom II quad core, Nile should have a mobile-tuned Phenom II dual core. Where Danube will have the most robust discrete Manhattan adapters, Nile will walk with the value versions, or stick to the Radeon HD 4000 IGP.
Nile-based notebooks should be excellent candidates for those conscious about battery life or notebook size, and they will certainly pack a harder punch than any AMD notebook on the market today. For a look at what Geneva-style platforms look like this year, take a look at our introspective on AMD’s first and second generation Ultra Thin products.
Continuing the crusade for battery life
AMD has spent the majority of 2009 on a crusade to not only extend battery life, but to call for honesty in marketing when it comes to battery life ratings (a crusade we agree with). AMD hopes to put its money where its mouth is in 2010 by improving battery life by 25% over the current stable of Turion II and Ultra Thin products. The company will do this in several ways:
- Refreshing the Turion line with chips inspired by Phenom II’s C3 stepping will allow lower idle powers, faster power switching, finer clock control and lower nominal voltages.
- The RS880M chipset itself consumes less power than its predecessor.
- DirectX 10.1 IGPs and DirectX 11 discrete GPUs will consume less power at idle, which is where most laptop GPUs spend their time.
Improving idle power states for improved battery life.
Succinctly, AMD claims it wants to improve the battery life of its third-generation Nile Ultra Thins by one hour. That’s a lofty goal, and you can bet we’re going to be exploring that avenue when the time comes.
Final thoughts
AMD has spent the last few years hiding in the shadow of the Phenom I and Intel’s considerable capital investments. Mind share is every bit as important as market share, and the company’s early losses to the Core 2 series cost them confidence and customers, the echoes of which persist today.
What we see in response is a company that has undertaken the Herculean effort to right its course through an iterative strategy. While AMD is not kicking out die shrinks every year, or new architectures every two years, the company has more than made do with the technology it does have. Through the introduction of 45nm, DDR3 support, new processor steppings and a mighty push with the platform—a tack no other company on earth can take—AMD has continuously offered new goodies for buyers.
Their efforts in 2010 are no different. As the company continues to master its existing processes, the firm is reducing power consumption, adding chipset features, adding CPU cores, improving battery life and reducing footprints. Not bad for a company with a market cap that’s thirty times smaller than its competitor.
The AMD of 2010 is more focused, ambitious and well-prepared than it has been in a long time. Through a six month lead on DirectX 11, unbeatable pricing, seamless component interrelation and outstanding community outreach, AMD is looking healthy. And damn if that isn’t good for everyone’s wallet.
Pages: 1 2
Articles RSS