Thursday, May 25, 2000

A few months ago, IBM made a momentous shift in its PowerPC strategy with the announcement that it would freely share its "PowerPC-Open-Platform" (POP) reference design with any and all interested parties. This shift in policy is one in a long line of measures in IBM's concerted attempts to upset a status-quo that has seen it relegated to the position of "Just-Another-Computer-OEM", in a industry they largely helped found, by companies they unwittingly helped build to their current positions at the top of the foodchain (Microsoft and Intel). IBM's two more interesting paths to this end- adopting and supporting Linux, and the PowerPC microprocessor - represent "Big Blue's" big sticks (or maybe stakes) aimed at upsetting the positions of Microsoft and Intel, respectively. With POP, those two strategies have converged.

The PowerPC microprocessor - developed in conjunction with Apple Computers Inc., and Motorola Semiconductors, and based heavily on IBM's homegrown "Power" RISC architecture - has for years been marketed by the Apple-IBM-Motorola (AIM) alliance as being a superior competitor to the x86 family of processors which has come to dominate the microcomputing platform.

The alliance that brought together these three disparate companies though, has failed to put together a strategy to break the architecture free of its relative niche in the computing industry. This has been due in large part to intermittent squabbling between its members, who's sometimes conflicting needs and ambitions have euthanized attempts to open the market.

The last cohesive attempt at marketing a standard PowerPC platform- CHRP (Common Hardware Reference Platform) - died a messy death as all three members failed or backed out of their obligations. Even as I write this, Apple stands as the sole major vendor of PowerPC personal computers, IBM's own PowerPC-based RS/6000 line of servers occupies a niche at the high-end of the market, and the remainder of PowerPC sales are made up of sales to the market for embedded chips and the makers of PowerPC upgrade cards.

By all accounts, IBM is eager to crack the multi-billion dollar market for personal computer processors, to grab a larger piece of the market that has seen demand grow by exponentially over the last 10 years. Most of that demand has gone to Intel, AMD, Cyrix (now part of VIA), and WinChip, all of whom manufacture variations on the Intel x86 architecture. The insatiable demand for x86 processors stems primarily for demand for the world's most popular operating system: Microsoft Windows, whose many incarnations have come to occupy nearly 80% of the market. Unfortunately for companies such as IBM, who market chips other than the x86, Windows 9x only runs on the x86 family of processors; a circumstance (or arrangement) that has put many a competing product in its grave. Then, however, Linux appeared.

Linux (or GNU/Linux) is the technological and spiritual anathema of Microsoft. It is an open source, freely distributed, oft-ported Unix-like operating system that has caused a wildfire in ComputingGeek Land. From IBM's perspective, Linux's three most attractive characteristics are its rapidly growing popularity, its status as a completely free OS (both in dollar terms, and in terms of intellectual property), and the fact that it can be made to run on a PowerPC computer.

Several versions of Linux already run on Apple's PowerPC-based Macintosh computers: YellowDog Linux, MkLinux, and LinuxPPC. The opportunity for IBM is thus quite clear. With the market for Linux equipped computers growing, there is no reason why a good number of those computers couldn't be using PowerPC chips (preferably those made by IBM itself). The one very significant barrier towards such an outcome, is - or was - the lack of motherboards on which to place the PowerPC chip on which to run Linux. The only large-scale producer of PowerPC microcomputer motherboards is Apple - a company once burned and twice shy about the handing out it's hardware secrets to clone makers.

IBM's answer to this? Do the design, and let someone else do the building. The design they came up with is POP: a 66Mhz, 64-bit motherboard with 3 PCI slots, and support for ATA 33 and SDRAM, meant to support a single PowerPC 740 or 750 chip. POP stands as the intellectual heir to CHRP - an open PowerPC architecture designed to support a plethora of competing OSes. The trick, of course, is getting the systems built: Without POP systems being built, IBM has no new customers to sell its chips to.

Fortunately for IBM, the lure of a free motherboard design has attracted a few interested parties. Among those parties who have announced their adoption of the POP spec are Prophet Systems (A division of Eternal Computing), and Total Impact (an established producer of multi-processing upgrades for the PPC Macintosh).

Both companies have made clear their intentions to develop the POP spec to include new technologies outside the original CHRP spec.

Prophet Systems intends to sell its version of the POP motherboards as stand-alone products, as well as complete systems. They also plan on marketing their own POP chipset to OEMs and boardmakers. According to Prophet, their POP chipset will support PCI, 4x AGP, ATA/66, USB, and Ethernet on-board. Firewire, and PCI audio are also planned. Prophet has also stated they will forgo ISA - the expansion standard also included in the original CHRP spec - in favor of PCI, citing the undue costs of supporting a legacy standard. No information on shipping dates or expected pricing was available from the Prophet website.

Total Impact, on the other hand, is applying its experience in multi-processing solutions to the POP standard, with its announced intention to merge its TotalMP technology with the IBM designed POP motherboard. The system is targeted as a Linux system, with the base system including 5 PowerPC processors. Also, unlike the original POP spec, Total Impact intends to build systems around the Motorola designed PowerPC 7400 - or G4 - processor. These multi-processing machines will be internally expandable to 13 processors, with "100's of additional processors" being possible, via a passive backplane. Total Impact has stated that systems should be available "...In the first quarter of next year (year 2000)." Pricing is yet to be determined.

It definitely means the market for PowerPC computers will face a serious shake-up. Currently, Apple stands as the sole producer of PowerPC personal computers. If and when independently designed POP computers begin entering the market, Apple could possibly find itself in a situation hauntingly similar to the Cloning Fiasco. On the other hand, high-end PowerPC Linux boxes do not stand as an immediate threat to Apple's traditional customers: computing neophytes, graphic designers, and desktop publishers. They may, though, make inroads into a realm of servers and business where Apple has traditionally been weak. Under that scenario, Apple could benefit in the long run from the acceptance given to PowerPC products in general, and more so if their own Open Source OS - Darwin - is ported to the new platform. Even though Apple has expressed no intention to dedicate its own resources to porting Darwin to other platforms, it is eminently possible that Darwin Open Source enthusiasts will. Apple could very well see their installed base grow without lifting a finger.

Be Inc., another maker of a PowerPC OS, has recently announced that it has no immediate intention of porting the Be OS to new PPC motherboards. Though Prophet's site suggests that they are in talks with Be, no dedication has apparently been made to support POP systems, and the matter remains open speculation.

IBM, of course, stands to win big. Any system that ships with a PowerPC processor is potential revenue out of the pocket of Intel and into that of IBM, and defrays future development costs for the platform. Any system that ships with Linux, is a system that ships without Windows. Even if Big Blue doesn't make out like a bandit in this little adventure, they stand in a good position to poison the well of Wintel duopoly. Anything that knocks either of the two down a peg, puts IBM in a position to reposition itself nearer the top. Big Blue, in fact, has nothing to lose and everything to gain- and all it took was placing an old motherboard spec on the Web for anyone to grab hold of. Someone once said that revenge can be sweeter than nectar. Someone at IBM must have been listening.

You can discuss this article in the Forum