Gary Griffiths, who is director of business development at IBM’s Power Personal Systems division, says that roughly 1,000 copies of the PReP specification have been distributed to potential system developers and that between 15 and 20 companies have been actively involved in its development. Most want to remain anonymous: PowerHouse Systems Inc is an exception. PowerHouse first got involved in defining the specifications about a year ago, says its vice-president of hardware engineering Glen Miranker. At that time the Reference Platform’s whole approach was different, he says. Whereas today, the sample machine is a mere appendix to PReP, in the initial conception it, it was at the heart of it, with most of PReP dedicated to describing it. The initial approach was basically totake the sample implementation and describe it, he says, describing this draft specification as completely redundant. Since then, PReP has evolved swiftly, thanks to a constant flow of electronic mail and phone discussions between the participants. Electronic communication has been supplemented by regular face-to-face meetings: at IBM’s Austin, Texas facility, at Motorola Inc, at PowerHouse and at the premises of others, more shy. At no time does the name Apple Computer Inc turn up – Apple’s attitude to PReP will have to wait for another, longer article.

Belongs to IBM

Currently the PReP spec belongs to IBM and to IBM alone, but the company is committed to handing control over to an impartial third party as soon as it is mature. The only reason that the hand-over hasn’t happened already, claims IBM’s Griffiths, is the urgency to get it developed quickly: I can’t possibly, at this time, take on negotiations with a third party, he says giving the impression of a man in a hurry. Indeed the speed with which the specs have stabilised would shame most formal standards bodies. The alpha version was officially released in November 1993 and March 15 1994 saw the ‘publication’ of the beta document – though just how public this will be remains to be seen. Griffiths says that the changes from alpha to beta are a question of fine-tuning: if someone had started to develop in November, there would be nothing [in the beta] that would cause them much trouble he believes. The final document is scheduled to be ready in the next quarter. And where will it go then? The PowerOpen Association? That’s certainly one option, according to Griffiths, though he stresses that no-one has had the time to consider PReP’s final resting place. And yet, the potential for conflicting interests with IBM itself is real enough. On the one hand it has to be the impartial custodian of the PReP standard; on the other, IBM’s Power Personal Systems division needs ruthlessly to keep a competitive edge to sell its forthcoming PowerPC-based machines. Griffiths reasonably points out that the original Personal Computer cloners weren’t deterred last time, and that was in the face of IBM hostility – this time around they are being actively courted by IBM – although IBM is these days not perceived to have quite the clout to set industry standards as it had back then – less than a decade ago.

By Chris Rose

Moreover, back when the original Personal Computer was launched, the industry, such as it was, would probably have jumped upon any personal computer standard bearing the IBM label. Today’s task is more difficult, since IBM has to convince them to leave an established standard that many seem quite happy with. In addition, the Power Personal Systems division currently co-ordinates the development and distribution of PReP in such a way that it could, if it wanted, glean quite a bit of information about what potential competitors are up to. Griffiths says that he has never looked at it that way and argues cogently for the need for the standard to be open. It doesn’t seem to bother PowerHouse unduly: Miranker, says it’s my belief that IBM has a profound understanding of the necessity for clone manufacturers. There is another, potentially more serious, conflict and that involves IBM Microel

ectronics, currently the sole manufacturer of the PowerPC 601 processor. The question of just how level the playing field will be for PReP-cloners was thrown into sharp relief by the saga of the shadowy PowerPC 615 processor. The rumour of a PowerPC chip, customised to emulate a fast Intel Corp processor had been doing the rounds since last year, but it was PC Week that put a name to the alleged chip and reported that it would be reserved for IBM’s exclusive use in its own machines. Frankly we don’t know if the 615 exists: IBM spokespeople simply call the story pure speculation; but Unigram/X reports that a source at IBM admitted, unofficially, that such a beast is still on the drawing board. Griffiths’ immediate reaction is to says that the 615 does not exist, but swiftly adds that it would be inappropriate for him to comment. But there are more important questions than whether it exists or not, and these are raised by the mere rumour. If IBM can produce a specialised version of the PowerPC chip with some special properties, and if it can keep it for the exclusive use of Power Personal, where does that leave the clone manufacturers? Would they really be able to compete against an IBM so equipped? And wouldn’t IBM corporate be tempted, as the market and the competition grew to take advantage of the special relationship between the chip and machine builders? It is not something that the original clone manufacturers had to face – the CPU was controlled by Intel, not a personal computer vendor. Compaq Computer Corp’s recent complaints about Intel’s involvement in the motherboard business show what can happen when worlds collide. It’s not an issue, says Griffiths, who describes the links between IBM Power Personal and IBM Microelectronics as a customer-supplier relationship, nothing more: they are clearly separate businesses, he adds. The concept of autonomous business units is relatively new to IBM: it took hold in the dying days of John Akers’ stint as chief executive.

Go to Motorola

Under Louis Gerstner’s stewardship, the policy has come under scrutiny and the chattering classes have suggested that there is a return towards centralisation. IBMers themselves say that autonomy is alive and well: Gerstner may have halted the proliferation of businesses with peculiar names, they say, but the underlying commitment remains (but not the openness or the separate books). So, the basic process of IBM Power Personal knocking on the door of IBM Microelectronics should be the same as it as for any other potential customer. Anyone that wants specialised PowerPC processors is at perfect liberty to approach the chip maker and negotiate. Alternatively they can go to Motorola for their needs. Presumably this separation accounts, at least partially for Griffiths’ reluctance to comment on the 615’s existence – he shouldn’t actually know the answer, unless it was a chip that Power Personal had commissioned. Given that, in Griffiths’ words: IBM Microelectronics is in the business to get as many customers as possible, it is not clear how it could commercially justify selling something like the 615 to Power Personal only – when it could sell it to others too. But now we begin to heap conjecture on top of hypothesis and it is time to stop. Only time will tell how well IBM handles the practicabilities of juggling PReP, its chip-making business and Power Personal. To engender absolute confidence, the company could do more to place a ‘Chinese wall’ between the two hardware arms. Currently it places a lot of emphasis on a formal separation between Power Personal and Microelectronics. A separation that isn’t that obvious to those of us that don’t work at IBM. Still, if all goes well, perhaps in five years’ time, we will be able to ask the question why are there so many PowerPC machines on people’s desks and point back to IBM’s decision to embrace the clone makers.