Top 10 Chip Stories to Watch in 2008

4. Virtualization Goes Mainstream

Virtualization may have been the hottest technology in 2007, what with VMware's boffo IPO, Citrix's acquisition of XenSource as well as Oracle and Microsoft's belated entry into the fray. But 2008 will be the year when all the buzz becomes reality, panelists said. Bob Anderson, executive vice president of business development and strategy at ZT Systems, saide he thinks that the push for power savings in data centers is the overarching story that affects all other trends in the market. And as IT professionals continue to fret about reducing the energy consumption of servers, virtualization will be regarded as the No. 1 cure.

Patrick Moorhead, vice president of advanced marketing at AMD described 2006 as the "proof-of-concept" year for server virtualization, 2007 as a year of testing and small-scale integration, while, "2008 is the year in which you're going to see actual volume."

That means continued jockeying for virtual pride-of-place on Windows and Unix servers between HP, IBM, Dell and Sun, as well as plenty of opportunities for smaller system builders. Managed IT service providers will also have a piece of the action, according to Moorhead, who said, "Virtualization is going to make a huge statement in the MSP channel." He also believes client-side virtualization is now in the position that the server-centered technology occupied in 2006.

Keith Millar, vice president of product management at Liquid Computing, added, "There's going to be a race to support Virtual Machines with processor features, such as memory mapping and virtual page management."

5. Graphicspalooza

With the explosion of high-definition digital media and gaming, graphics and video processing is integral to PC performance across all form factors and price points. User experience isn't just for enthusiasts anymore. That means that while the traditional path of cutting-edge graphics trickling down to the mainstream from high-performance systems is still operative, that cycle is now a whole lot tighter. Nvidia and AMD's ATI division are the most obvious drivers of this trend, but watch out for Intel, especially in the ultra-mobile space, said panelists.

Meanwhile, the particular strengths of GPU processing -- like grinding through critical computational algorithms in mathematically demanding work -- is being exploited as never before to accelerate system performance. Areas to watch include stream computing, where Nvidia and AMD will continue to go at it with competing products in 2008.

"The strongest part of a mainstream [whitebox] business is the $1,000 system. That's not the biggest part of the market, but we see it as the strongest and the biggest growth area. Every single one of those systems has a GPU in it, so that tells you something," Toste said.

"Graphics are more than just for games," Moorhead added. "On the mainstream client, all things Vista, like Google Earth and Powerpoint, have become 3D-aware and that's just the start. You will see software that will take video encoding and use the GPU to bring what would be a 10-hour rendering process into real-time. On the server-workstation side, you're looking at real-time rendering for not just movies, but pharmaceutical companies running intense math computations. Or weather modeling, which is just one giant mathematical computation, taking current data and doing what's called pattern-matching. This is the year the whole graphics thing is going to bust out of the seams."

6. Small Victories

Is this the year we finally get "the whole Internet in our pocket?" That's been the dream for some time, and 2007 was notable for the release of several dynamic ultra-mobile personal computers (UMPCs) and smartphones. The iPhone got the biggest buzz, but slightly larger x86-based UMPCs from companies like OQO are arguably more impressive, ounce-for-ounce. OQO's O2 UMPC runs on a CPU from VIA, an up-and-coming maker of ultra low-voltage chips.

Apple showed us that a smartphone can sell like hotcakes. But to get the full Internet experience on a tiny device, a lot of work remains to be done. According to Intel CEO Paul Otellini, the hardware -- from those ultra low-voltage CPUs to solid state flash drives the size of a penny -- is already there. But at September's Intel Developer Forum, Otellini said two big hurdles stand in the way of the full PC experience on a UMPC or handset. First, software developers must figure out clever ways to squeeze the large-format, graphically rich interface of the Web onto the smallest of screens in user-friendly fashion. The other big challenge, Otellini said, is connectivity. On that front, Intel is putting a whole bunch of its eggs in the WiMAX basket. However, a list of currently deployed WiMAX networks around the globe shows how far we are from widespread adoption of that standard, let alone whether it will ultimately be the one we adopt.

Intel's not hedging its bets on this space blowing up -- the chip giant has its first UMPC platform, the Silverthorne "system-on-a-chip", coming out in 2008. Our panel's not as eager to bet it all on small just yet, however. They have questions as to whether progress on UMPC technology will live up to the hype, if a stable ecosytem of OEMs will emerge, and to what extent sales of such devices will grow.

"2008 is a very interesting year for UMPCs, with lots of hype. The technology will be catching up with the vision of the product. VIA plays in this segment. They've picked up some wins and we should see some competition between them and Intel in this area," said Mercury's McCarron.

Added IDC's Rau: "You need an SOC [system-on-a-chip] to create a performing device. It's a well-into-2009 development. Demand for UMPCs is still very slight, representing about 1 percent of volume according to IDC numbers for UMPCs out of mini-PCs."

7. Tick and Tock

Before the extent of AMD's annus horribilis became widely understood, chip watchers were eager to see if either AMD or Intel would be able to race ahead of the other on both micro-architecture and die fabrication in 2008. With its "native" multi-core architecture, AMD is believed by some to hold the advantage in "tock", while Intel's already raced ahead on "tick" with its late-2007 launch of 45nm-process devices. And while the smaller chipmaker's woes have put a bit of a damper on that horse race, it will still be exciting to see whether AMD gets its next die shrink rolling before Intel succeeds in ramping its scheduled micro-architecture reboot, codenamed Nehalem -- or vice-versa.

Nehalem is widely expected to be a major redesign that will bring Intel's micro-architecture in line with AMD's, meaning independent power supplies to individual cores and a memory controller on the CPU die. The folks at AMD believe that's a tacit admission from the market leader that the smaller guys got it right in the first place. Meanwhile, Intel can't have helped but notice all the delays AMD experienced in moving to quad-core. Building chips the way AMD does is exceedingly difficult to do. You have to wonder if Intel's going to be able to hit its second-half deadline for shipping such a complicated upgrade. Interestingly, AMD's target for 45nm is the second half of 2008, as well. We'll see which company gets to its target first.

"Our architecture with Barcelona and Phenom is the one Intel wants to get to," AMD's Moorhead said. "And we'll ramp up 45nm in the second half of next year, so we'll be on the same footing there. It will in some ways break Intel's business model, because they'll have a memory controller on there also. How will they price that?"

Steve Dallman, general manager of the Worldwide Reseller Channel Organization at Intel, said, "So when we get Nehalem in the second half of 2008, that's going to be a heck of an opportunity for a lot of the system builders. It's an opportunity for high-end products, and later on, a tier below that. Nehalem will probably be one of the biggest architectural changes the channel will have seen in the past few years."

Brian Corn, vice president of marketing and development at Source Code, concurred. "With Nehalem, we'll finally be able to compare Intel with AMD. All this time, we've been comparing apples and oranges because of the fundamental differences in their micro-architectures. Now the technology, like the way the memory architecture is built, is going to be really, really similar."

Next: Channel Developments to Watch