The IBM Corp-Apple Computer Inc alliance was negotiated in such short order that when either side talks about it, one has to be cautious about taking everything that is said as Gospel – and at the new IBM, a lot is being said, which is very welcome. The starting point for the relationship is the current single-chip implementation of the nine-chip set that made up the original processor (not nine different chips – some are duplicated). IBM will be using the chip in the baby RS/6000s now due early next year, and says it will offer them OEM as well – indeed third parties will be able to buy chips, boards or workstations OEM.

PowerPC

The PowerPC will be the next version of this chip, optimised for single-precision work – but the commitment is to make the RS/6000 line binary compatible from the PowerPC to the very top. Chips should be available in early 1993, with products able to be launched in late 1993. And IBM and not Motorola Inc will have have to be the initial source of the chips: once the design is complete, it will take Motorola time to get the process technology in place and working. It is at present being insisted that IBM, Apple and third companies will get chips at the same time – and don’t look for IBM and Apple to share a machine design: the idea is to start with the same chip and the same PowerOpen operating environment – and then for each to design a machine that will knock the other’s out of the water. The PowerOpen environment – based on the forthcoming AIX 4 – should be ready in one to two years – way before object-oriented Pink. It will provide the Macintosh user interface and 68000 Mac applications should run. It will include X Window, embrace Apple’s A/UX as well as AIX and support the Macintosh desktop and tools. The plan is for laptops and portables, entry and mid-range workstations, and servers, all using the PowerPC processor.From the IBM side, the intention is to license everything to third parties, and the aim is to rally supporters of the PowerPC and PowerOpen in a club with the 88open supporters’ group backing Motorola’s 88000 RISC as a model. And just as Ford Motor Co wants a microcontroller version of the 88000, IBM conceives of a microcontroller being derived from the Rios RISC or the PowerPC. The real excitement attending the IBM-Apple alliance all surrounds Taligent, the joint venture company charged with turning Apple’s Pink development programme into an object-oriented operating environment. And objects are not entirely new to Apple – it has already been using Objective Pascal to write some of its system software. Pink is based on a microkernel developed from scratch by Apple rather than picked up off the shelf, and the product will be this microkernel with a modular object-oriented operating system on top.

Workplace Shell

The microkernel will accomodate applications in heterogeneous environments. The IBM contribution, including object-oriented developments from the Metaphor Computer Systems Inc acquisition will be added to Pink – but also from IBM itself: according to AIX Systems Manager Bill Sandve, the Workplace Shell for OS/2 2.0 includes object technologies that are key to Taligent. Commenting on Apple chief John Sculley’s proclamation in Frankfurt the other day that IBM-Apple-Motorola was looking for a European company to join the party (CI No 1,796), Sandve pointed out that each of the present partners had a major contribution to make to the effort, and that any further addition would have to chip in with effort of comparable value. The areas where the partners feel that the collaboration could be strengthened are additional fabrication capacity – a third source for the PowerPC and support chips; the tailoring of AIX for different vertical markets; multi-processing expertise; 64-bit operation and high availability. Just about the only substantial European companies left are Siemens-Nixdorf Informationssysteme AG, Compagnie des Machines Bull SA and Ing C Olivetti & Co SpA; of the three, only Siemens-Nixdorf has access to semiconductor process technology – and is already bes’ f

riends with IBM on 16M-bit memory chips and up, and on private telephone systems. Siemens is keen to be a more substantial personal computer player and so far has not firmly aligned itself with any of the various clubs, although it is committed to Unix System V.4 on the MIPS Computer Systems Inc RISC. However now that IBM has torn up all the rule books, no partner can be considered out of the question, however unlikely, so that to bring in ICL Plc and its controlling shareholder Fujitsu Ltd would corral both the European and Japanese markets in one throw, and Fujitsu has the process plant to make the parts in Europe – the UK – as well as Japan. And while ICL is committed to Sparc for the mainstream Unix market for business, at the personal computer end, it has not yet moved beyond the undifferentiated MS-DOS – or Concurrent DOS – and Unix on Intel Corp iAPX-86.

SGS-Thomson

IBM is not giving any further clues on who the partner might be, but it is not at all clear that one has been chosen yet: indeed it may turn out after detailed negotiations that none of the partners available fit the bill. A further possibility is that we are talking about two European partners: SGS-Thomson Microelectronics NV would adore to take on board a high-volume microprocessor and could make very significant contributions to the effort with Transputer technology from Inmos International Ltd. And the Franco-Italian company could offer a way in for either Olivetti or Bull, with the former a little the more likely. The other company that is being established under the IBM-Apple effort is of course Kaleida, charged with pushing for and setting standards for multimedia technology, with the firm intention that anything that comes out of the effort should be freely licensed to all comers. It is not at present generating the same degree of excitement as Taligent, mainly because while most people are certain that multimedia will be a major extension of computing technology in the medium term, it is presently too expensive and looks like a solution seeking a problem to solve. But before any of this can happen, the Macintosh has to be embraced into the capacious IBM bosom: tying the Mac into the IBM Enterprise Network means extending systems and network management, file and print services, database access and more comprehensive SNA support to the Mac, as well as improving AS/400 and OS/2 communications with the Mac. Most of those things are available from third parties: the one positive move so far is to provide a Token-Ring Network board for the Macintosh.