Brian's Brain

Sep 27 2006 10:47PM | Permalink | Email this | Comments (0) |
Blog This! using:  Blogger.com | LiveJournal |
Digg This | Slashdot This | add to Del.icio.us

Intel's aspirations in the PC space over the years can be summed up in the following two straightforward strategies:

• Continually migrate the platform, and therefore the customer base, to an increasingly richer and broader suite of functions, in order to sell leading-edge, therefore maximally-profitable and sole-sourced, microprocessors, and
• Capture as high a percentage of the overall silicon landscape in the PC, beyond the CPU, as is profitably possible

Back in my Intel days, the first flash memory chip I worked on was the 28F001BX, which became the first high-volume flash BIOS storage device. Although antitrust safeguards prevent a company like Intel from selling its chips to sibling divisions at lower prices than it charges its tier 1 customers, Intel's motherboard group in Hillsboro, OR was a natural partner for the Folsom, CA flash memory group's BIOS aspirations. And, for any customer attracted to the simplicity of one-stop shopping, a CPU-plus-BIOS (plus core logic chipset, plus Ethernet transceiver, plus....) bundle was appealing.

I'm getting a strong déjà vu tingle here at IDF, because the flash-plus-processor package pitch is playing out once again. Only this time the densities, and therefore the stakes, are much higher. Back in May, as part of my WinHEC coverage, I updated you on the latest news regarding Microsoft's dual ambitions for exploiting bulk flash memory in the PC; ReadyBoost, which uses flash memory as a higher-performance read/write alternative to the HDD in virtual memory paging situations, and ReadyDrive, which caches frequently-read HDD information in faster-access flash memory, whose contents also survives system hibernation and reboot.

Microsoft's documentation spells out the most common to-date flash flavours for ReadyBoost and ReadyDrive, respectively a USB flash drive and a flash array embedded within a HDD. But Intel's got its own spin on cache-in-flash, Robson Technology, which the company claims is able to support both ReadyBoost and ReadyDrive. The company's performance and power consumption improvement claims for Robson, voiced during today's mobility keynote by David Perlmutter (Senior Vice President and General Manager of the Mobility Group) were eyebrow-raising; ½ the resume-from-hibernation and application load times of a conventional system, along with a 400 mW reduction in system power consumption.

Robson prototypes showcased at the conference came in 256 MByte and 1 GByte density variants, and in the PCI Express x1 Mini Card form factor (the Register's writeup includes a photo), amenable both to OEM density flexibility and to end user upgrade scenarios. You can also, of course, solder Robson flash memory arrays directly on the motherboard. Intel was careful to point out that Robson inclusion was an optional part of the upcoming Santa Rosa platform bag-o-chips, not a requirement in order to obtain Centrino certification. However, you've gotta believe that Intel's going to be pushing hard for its customers to adopt Robson, now that it's got fab-filling NAND flash technology in-house. Said another way, Robson was probably the linchpin factor that convinced Intel's upper management and board of directors to do the Micron partnership deal in the first place.

HDD manufacturers see the hybrid hard drive as an attractive avenue to propping up their oft-scant profit margins. Memory suppliers, and their retail channel partners, view ReadyBoost as a compelling path to shipping more, bigger, and faster (translation: more expensive and profitable) USB flash drives. Intel aims to short-circuit both groups' aspirations with Robson. Coopetition still reigns supreme in the PC application space.