CES 2009: The price of falling prices, and a path to Intel's potential demise

Atom is a short-term success, but does the netbook trend represent a long-term threat to Intel's beefy revenue and profit margins—as well as a potential salvation for graphic-processor makers?

By Brian Dipert, Senior Technical Editor -- EDN, 1/8/2009

Here's how I opened today's value x86 CPU benchmarking cover story:

It's unclear to what degree Intel anticipated the fiscal crisis now gripping the globe and therefore developed the economical Atom CPU as a means of preserving its market-share position. (The company publicly states that its primary intent with Atom was to broaden the application base for the x86 architecture.) It's also unclear what negative impact on long-term corporate revenues and profits may result from Atom's potential "cannibalization" of more lucrative Intel microprocessors.

Nonetheless, any indication of robust sales is a rare and encouraging sign in today's economic malaise. So-called netbooks currently account for the bulk of PC-market growth, and Atom is the CPU engine powering most netbook designs, so Intel deserves kudos for pragmatism in using Atom to obsolete its own CPUs instead of allowing competitors to do so.

Truth be told, given the multiyear timespan that elapses between when a new microprocessor architecture is proposed and when the first products based on that architecture appear, I honestly doubt that Intel anticipated the economic crisis now gripping the globe. So while I can't give the company credit for prescient forecasting, I'm still struck by the lucky timing of its Atom unveiling. Note, though, that as in my late May 2008 cover story, my praise is muted; although Intel is "making hay" (as we used to say back in Indiana) in the near term, Atom may do acute long-term harm to the company roadmap and resultant fiscal aspirations.

To explain, I'll tell you a bit about a conversation I had last week with a longtime industry observer who was attempting to get a handle on Nvidia's current status and future prognosis. She rattled off a laundry list of company strengths, shortcomings, aspirations, and hinge factors, which largely echoed points I'd made in recent writeups on the company. And she specifically offered up Tesla (i.e. GPGPU for supercomputing) and Tegra (a multimedia-enhanced ARM processor family for handhelds) as particularly encouraging opportunities for the company.

I agreed that both products had notable upside potential, but I pointed out that they both had notable risk factors as well: supercomputer profit margins are impressive but volumes are not (especially considering that only a percentage of the total market is GPGPU-applicable), and Nvidia has plenty of legacy ARM licensee competition. In response, however, I offered a bright spot that she had overlooked (and that I'd written up just two weeks prior); the pairing of the company's GeForce 9400M single-chip core logic cluster (aka Ion) with Intel's Atom CPU.

The crux of what I find so compelling about the netbook (and nettop) phenomenon is that it represents the first time I can recall since the initial unveiling of the notebook (versus desktop) PC when CPU horsepower has taken a substantive step backwards in the face of some other more compelling factor. In the desktop-to-laptop transition, consumers consciously chose to lessen their CPU demands in exchange for factors such as lighter system weight and longer battery life. And in the modern notebook-to-netbook (and desktop-to-nettop) era, system price reductions are the fundamental motivation.

Intel and partner Microsoft are doing their damndest to keep the notebook and netbook PC segments distinct, in the hopes of preserving the size (and profitability) of the former fed by its higher-end CPU families. Take the Atom N270 CPU found in the MSI Wind U100 netbook that I tested in my study, for example. Intel's own Web site advocates pairing that particular Atom processor flavor with the Mobile Intel 945GSE Express Chipset with 82801GBM I/O Controller Hub (ICH7M). That particular core-logic chipset has an archaic integrated graphics core, and it also only supports up to 2 Gbytes of system memory. OEMs also report that Intel insists that they use the N270 only with systems containing 10-in. (diagonal) maximum LCD sizes.

Back in late October, Dell introduced an Atom-based netbook that seemingly violated Intel's "recommendations," although subsequent research has convinced me that it's far less controversial than it appeared at first glance. The Mini Inspiron 12, as its name suggests, offers a larger 12-in. diagonal LCD, seemingly enabling it to make a serious intrusion into traditional notebook PC turf. But as it turns out, the system is based on the Atom Z520, which has even more restrictive chipset "suggestions." The Intel SCH US15W Chipset with I/O Controller supports only 1 Gbyte of maximum system memory, and its graphics core is even more primitive—not even capable of handling Windows Vista's "aero" user-interface option.

Back to Nvidia. As my mid-October 2005 cover story and its Web addendums pointed out, Intel and its graphics partners have long advocated that spending a little more of the overall system budget on a more powerful GPU makes more sense from an overall return-on-investment standpoint than spending a lot more of the overall system budget on a more powerful CPU. Unfortunately, the graphics upstarts have been fighting (and largely losing) against a very powerful current; since the days of the first PCs and Macs, consumers have been trained that faster (and in more recent days, higher core-count) CPUs directly translate into better and slower-to-obsolesce PCs and Macs.

"The netbook phenomenon has opened consumers' eyes to the reality that they've long been buying far more CPU processing muscle than most of them require. The era of robust sales of competitively isolated high-performance Intel microprocessors is, I suspect, therefore largely drawing to a close."

Now, however, the ascendance of the CPU-deficient netbook and nettop symbolize a substantial change in the rules of the game, thereby providing perhaps the first (and last?) tangible opportunity for Nvidia to redefine the fundamental parameters of the computer platform. I never took seriously the rumors that Intel was refusing to sell Atom CPUs to potential customers interested in Nvidia's Ion, and apparently my skepticism was justified; in-progress lawsuits and worldwide antitrust investigations hamper whatever pressure Intel might bring to bear under other circumstances (though have no doubt that the company will do its utmost to make an all-Intel offer too fiscally attractive for customers to refuse). And the recently standardized OpenCL API, scheduled to appear in Apple's forthcoming OS 10.6 ("Snow Leopard"), will span the gap between Nvidia's proprietary CUDA and AMD's equally self-focused Stream to enable dynamic processing allocation between CPU and GPU. OpenCL probably won't natively show up in Windows 7, analogous to Microsoft's lukewarm-at-best embrace of OpenGL, but the company plans conceptually similar capabilities in upcoming DirectX v11.

Speaking of lawsuits, the netbook/nettop trend equally represents an opportunity for CPU (and core-logic) suppliers AMD and Via. Consider, for example, AMD's Neo CPU (the heart of the company's Yukon bag o'chips platform), which launched Monday inside HP's dv2 notebook (note the terminology) PC. As a sidebar in today's cover story points out, Neo is an aged, single-core K8 microarchitecture derivative fabricated on a 65-nm lithography process. What would have been a laughably noncompetitive silicon offering just one year ago is now a serious contender in the post-Atom era....like Nvidia (but due to matching in-house CPU expertise, one better than Nvidia), AMD's companion core logic also delivers robust graphics and video-processing performance. Similarly, as my testing shows, Via's Nano CPU substantially narrows (and in some cases collapses) the versus-Intel compatible-core-count gap of its C7 predecessor. If Via's willing to go toe-to-toe with Intel's budgetaries, it could grab a tangible slice of the overall business pie, especially in worldwide markets where Intel's brand doesn't carry particular cachet.

That new HP dv2 notebook PC, by the way, offers a 12-in. LCD. Intel will do its best to keep the notebook and netbook PC segments distinct, but I'll wager that competitive pressure will eventually compel it to cave and allow OEMs to embed Atom in more mainstream form factors, and with the company's modern companion core-logic offerings. At that point, today's artificial netbook-versus-notebook differentiation will collapse, and Intel had better hope that its unit shipments blossom in compensation for the resultant lost per-unit revenue and profit margin. The collapse may in fact be underway: MSI's MacBook Air-reminiscent X-Slim 320 embeds a 1.6-GHz Atom..and touts a 13.4-in. LCD.

The netbook phenomenon has opened consumers' eyes to the reality that they've long been buying far more CPU processing muscle than most of them require (excluding serious gamers, videographers and the like, of course). Eyes, once opened, are rarely coaxed into closing again, no matter how persuasive the pitch. The era of robust sales of competitively isolated high-performance Intel microprocessors is, I suspect, therefore largely drawing to a close, and the resurrected graphics processor may be correspondingly emerging from its 3D-centric tomb. What do you think?