November 11, 2005

Cell Processor Gets SW Boost… and Now to Market

I’ve been following the activities of Power.org for awhile now.  Power.org is the group IBM launched to drive collaborative development around the Power architecture.  You probably know that a Power product, the P5 processor is the base of IBM’s current iSeries, pSeries, and high-end storage architectures.  But another product of the Power architecture is the Cell Processor (officially the Cell Broadband Engine or CBE), co-developed by IBM with Sony Group and Toshiba. That processor is best known for its upcoming role in the next-gen PlayStation from Sony, but it has other uses as well including:

  • future digital TVs from Toshiba,

  • future Cell-powered engineering workstations from IBM and Sony designed to produce digital content for movies, TV shows, and video game,

  • future medical and chip-testing equipment from Mercury Computer Systems in the near term, and in military radar and sonar systems later.

So there are plans for Cell in consumer devices, as well as embedded devices, but naturally for the architecture to really thrive, it needs to be adopted by a broad base of technology developers.  The problem of market adoption is rarely issues with the hardware however, but is usually with the software.  If critical mass of software development doesn’t happen, a perfectly good hardware can languish or slip into a niche.  Additionally, if the development tools are lacking or the OS is too specialized – it just becomes too expensive for ISVs and developers to adapt.  Hence this week’s announcement by IBM and Sony Group is significant.  

In summary, IBM has released new software components and documentation, including software infrastructure extensions to the Linux operating system, compilers, and utilities.  IBM also has an alphaWorks site, where they routinely offer access to new and emerging technologies to developers.  Within that site, they have now created a new section for the Cell processor.

Making this technology available to developers is a critical first step.  The new challenge becomes disseminating the knowledge and building developer excitement around the platform.  Broadly speaking, there are three groups who IBM and its partners need to interest in this technology.  The first group is the über-geeks, hard-core programmers and technologists who will take this technology and run with it.  They are actually the easier group for IBM to reach, as this is IBM’s comfort zone.  The second group, is the leading edge of hipster-geeks, the multimedia, gaming, video, crowd that is frequently as much artist as technologist.  They are important because they will lead the coolness must-have end of the market. Many of these guys are also found at the university level too.  Finally there are the mainstream developers for embedded products and other ISVs that are more likely to consider Power as it grows in reputation and market familiarity.  

The question is how does a grassroots collaborative community like Power.org take advantage of marketing?  It can use word-of-mouth, but it’s not like there’s a marketing budget for Power.org as such.  IBM will drive some of the marketing, but there’s a risk in tying Power.org too closely to IBM.  Some of this involves seeding the university market and making Power interesting to future programmers.  It also involves major marketing to the Linux community, as it already markets to all three groups.  Just building and distributing good or great technology is not enough.  Building presence and awareness are the next big challenges to Power.org.


No comments: