Perfect Pair: PowerPC and Linux

by Henry Keultjes

Most Linux fans are not old enough to remember that Intel did not always dominate the PC processor market. In the late seventies, Intel won the chip war with Motorola and Zilog by offering certain features in its 8086 chip that favored MS-DOS over then existing competitive OSes. Subsequently IBM selected the 8088 for the first PC, knocking both Motorola and Zilog out of the emerging PC market. In spite of that loss, Motorola has managed to remain an important force in computer processors, mainly because of the PowerPC. As a consortium with Apple and IBM, the PowerPC is now in a position to save us from Linley Gwennap's One World, One Processor scenario. AMD is expected to continue its role, started as an Intel second source, a producer of Intel clones. As such, AMD is most relevant to these discussions because their chips obscure the real One World, Once Processor danger.

There certainly is no danger of the PowerPC meeting the fate of the last Zilog PC chip, the Z8000, and its potential successors. However, its role can be dramatically enhanced if the Linux community were educated not just about the dangers of a PowerPC demise, as Linley Gwennap has done, but also about the ability of the Linux community to prevent such a demise by being instrumental in helping the PowerPC become a factor in Linux desktop systems.

The Linux community is in a unique position to assure such an expanded role for the PowerPC. The Linux community can assure that the PowerPC will become a factor in Linux desktop systems, by jumping on the LinuxPC bandwagon. IBM is already leading the way toward a LinuxPC with its POP project, an Open Reference Design for a more or less standard ATX/PCI motherboard, but with a PowerPC chip rather than an x86 chip. Because of IBM's need to emulate high quality standards for its own RS/6000 boards that use the same PowerPC family of chips, this design project has encountered some delays, but it now appears to be on track. I have been told that one manufacturer has started limited production of these motherboards, and chances of obtaining a sample board in the near future are very good. However, those boards are only the first step. More important is giving additional producers a good reason to also manufacture PowerPC-based boards--not just motherboards for servers but also planar motherboards for complete LinuxPCs, desktop systems with much higher volumes than servers. These higher volumes will bring the price for LinuxPCs down to a level competitive with WinTel boxes.

The best reason for a company to manufacture these LinuxPCs is to give that company an assured demand.

I must confess I have an interest in seeing the LinuxPC become a reality. Linux, in order to continue its phenomenal growth, needs enterprise applications. The non-profit MD-Linux Foundation I am in the process of establishing will facilitate the widespread availability of thousands of proven enterprise applications that now run on a family of Multi-Dimensional databases (mdRDBMS). For that purpose, the Foundation will devise an open-source Linux distribution with an integral, not a ported, mdRDBMS engine. Such an integral database engine will not only allow these existing database applications to run on Linux, those database applications will also be able to interoperate seamlessly with native Linux applications. Pure Linux! That integral database engine and OS mix is designed to run only on PowerPC-based hardware.

Undoubtedly, you have read references to the Post-PC era in which computerized "appliances" will be much more numerous than the PC. Because most of these appliances run some form of embedded Linux, and because most of these appliances also run on RISC processors like the PowerPC, these appliances have become a natural stepping stone for a LinuxPC hardware platform. That such LinuxPCs will have very little, if anything, in common with the existing WinTel platform is evidenced by the briQ from Total Impact and similar products from SiliconFruit and Eternal Computing now nearing production. A typical LinuxPC would be about triple the height of the briQ, about 6" x 5" x 9", and would contain within that space as much, or more, functionality as today's giant WinTel boxes.

Where can you buy such a LinuxPC? That is the crux of this whole article. During the last several years we have seen lots of activity from organizations like LinuxPPC, OpenPPC and Yellow Dog Linux intended to promote the PowerPC as a Linux platform. However well-intentioned those efforts have been, they tend to put the cart before the horse. While the typical Linux fan is interested in broadening the Linux market via hardware that cannot be shared by any Microsoft OS, taking the PowerPC and turning that into a 6" x 5" x 9" LinuxPC is not within their capability, similar to building PC's 25 years ago. Linux fans, like WinTel fans, now have a normal expectation that some manufacturer will produce the hardware on which they will run their OS and applications. The reasons for such expectations are quite simple. Producing hardware has become much more capital intensive, much more complex and much more risky than producing software.

So why can you not yet walk into a store and buy a LinuxPC? Are technical issues holding up the LinuxPC? Designing a new product always involves some technical issues, but those technical issues are not really preventing PowerPC-based LinuxPCs from being broadly available at competitive prices. The current lack of availability of such a LinuxPC is a marketing issue. It is a "NO guts, NO glory" issue where both IBM and Motorola, on first blush, appear to have exhibited a marked lack of guts. However, upon closer examination, their lack of enthusiasm to compete in the WinTel market may have been a blessing in disguise. Because of the binary aspect of WinTel software, competing head-on against WinTel with a PowerPC did not make sense. Had IBM and Motorola done so anyway, there would have been a lot more ongoing innovation. As it was, the lack of competition produced the typical result: bland hardware and software, an environment ripe for Linux and the LinuxPC to thrive.

Open Source Linux changed the competitive landscape in both software and hardware. Whereas Linus Torvalds developed Linux to take advantage of the inexpensive WinTel platform, he unleashed market forces that go way beyond the scope of the legacy PC. Linux, already a force in servers, typically on rackmount hardware that lacks all semblance to a WinTel PC, and a much bigger force than Microsoft on those so-called appliances, is now poised to play a similar role in bringing dramatic changes to desktop hardware. Evidence of such dramatic changes in Linux desktop hardware can be seen in the briQ and similar products that are certain to fill the demand for "better than WinTel" hardware, a demand that Linux has created by being a better product than a competing Microsoft OS. However, bringing those dramatic changes to desktop hardware will require a guiding light, a company that can set the pace in conceptualizing the LinuxPC. Who will be the guiding light in Linux hardware development?

With all the deserved criticism for the roles they have played, both Microsoft and Intel also deserve credit for the frantic pace of hardware development. Had it not been for Wintel spurring an incessant drive for smaller and faster hardware, many of the hardware milestones, like 75GB hard drives, would not have been reached for a long time. Yet we all tend to forget that it was IBM, not Microsoft or Intel, that was the real guiding light not only behind the PC revolution itself, but also behind many of the revolutionary products and processes that have enabled us to enjoy unparalleled increases in computer power on our desk tops. Because of that earlier role, it is also my assessment that IBM will, once again, be the guiding light that will enable Linux to grow beyond the confines of its erstwhile WinTel platform. It is my assessment that IBM, already firmly committed to Linux across all its platforms, will bring its tremendous resources to bear on making both Linux software and Linux specific-hardware the most popular desktop platforms.

Will IBM and Motorola manufacture LinuxPCs? That IBM will brand LinuxPC hardware is almost a certainty. Both IBM and Motorola produce huge quantities of highly reliable and, consequently, pricey PowerPC-based professional hardware used in applications like telephony, the military and medical instrumentation, to name just a few. However, neither IBM nor Motorola play any visible role in marketing $600 WinTel boxes now. Chances that they will try to enter a market for LinuxPCs at similar price levels are highly unlikely. However, both companies have the enabling technology, the ability to produce the crucial PowerPCs and steer the development of the rest of the components, like planar motherboards and integrated chips, that will be the hallmark of the LinuxPC evolution.

If not IBM and Motorola, who then will produce LinuxPCs? The most likely producers of great quantities of LinuxPCs will be China and Taiwan. China especially seems to have a vested interest in freeing itself from its dependence on WinTel. Red Flag Linux has given China its independence from "Win", and the LinuxPC is capable of delivering China out of the "Tel" clutches. Companies like Acer, a Taiwanese company with ties to both China and IBM, are sure to play a role in bringing us affordable LinuxPCs, just as Acer has played a role in bringing us affordable WinTel boxes. However, companies like Silicon Fruit and Total Impact are certainly capable of playing a valuable role in the PostPC revolution. If one compares the possible labor content of today's typical WinTel box with the 6x5x9 LinuxPC, the labor content for the LinuxPC, as a percentage of manufacturing cost, will certainly be far less. Therefore, the customization that occurred in the WinTel market via manual assembly of boxes and boards and components will occur in the LinuxPC market via totally automated processes, not unlike those that Motorola and Apple used to revolutionize the manufacture or cell phones and the Mac, respectively. With such highly automated processes capable of producing incremental quantities of one of a kind LinuxPCs, domestic production might well account for a substantial portion of the total LinuxPC volume.

What about Apple's role in the LinuxPC? First, Apple defines the capabilities of the LinuxPC against the Wintel. We don't have to build LinuxPCs to know how they will perform against Wintel because Apple already has a LinuxPC, the cute little cube that Steve Jobs introduced about a year ago that can also run Linux. All other Apple products run Linux as well, so that gives us the needed benchmark to see how a certain LinxuPC will perform. The major difference between Apple products and LinuxPCs will be price and that the LinuxPC will not be able to run OS X unless Apple provides the enabling firmware. As pointed out, Apple is also the third leg of the current PowerPC consortium and has the capability for highly automated assembly, as evidenced by the fact that their Mac assembly labor factor is so low that they don't even track it. As such, Apple could play an invaluable role in furthering LinuxPC production. If for no other reason, Apple should assist wherever possible in furthering LinuxPC production because it is dependent on PowerPC production. If Apple can use its knowledge and influence to help convert annual production of about 25 million WinTel boxes into LinuxPCs, then, together with its own use of roughly 5 million PowerPCs, that 30 million piece annual volume would give the PowerPC a fair chance at gaining further market share against the roughly 95 million box market that WinTel boxes would then have.

What role can the Linux enthusiast play in bringing the LinuxPC to market? Success typically starts out as a dream, and you can help create a successful LinuxPC by responding to my dream. Think positive, not about what is but what can be. Think back about the defining moment for Linux. It certainly was not when Linus Torvalds started his project. It was when people responded to his dream of what Linux was to be. That enthusiasm was there long before Linux became a practical product. So is dreaming also required to assure that the PowerPC-powered LinuxPC becomes a practical product. The defining moment of the PowerPC will have arrived, when you and millions more like you, respond with enthusiasm to this article and to the future availability of the LinuxPC. The PowerPC will arrive when you create demand. When, instead of buying a typical WinTel box for your Linux now, you wait for a LinuxPC to become available.

Dream with me on how your LinuxPC will be so neat with that flat monitor you are lusting after. Dream with me of the ways the ergonomics of a typical cubicle can be improved using a wall-mounted flat monitor and a LinuxPC stuck on the wall with Velcro, and soon our dreams will ring in the ears of companies like Acer, Apple, IBM, Motorola, the companies that can make our dreams come true.

Henry Keultjes lives with his wife Alicia and daughters Claire and Hannah in Mansfield, Ohio. An enthusiastic user of multidimensional (MD) databases and an architect and interface designer of MD applications for more than twenty years, Henry is now more than knee-deep into a research project to give Linux those same unique MD database capabilities by natively integrating an MD database engine with Linux. His e-mail address is

Load Disqus comments