8th > May > 1999 Archive

Intel hand seen behind SIA roadmap shift

The Semiconductor Industry Association (SIA), which is a trade body for chip firms, has revised its roadmap and shown every sign of playing follow my leader. The organisation, which exists to lobby governments and promote its membership, said yesterday that .15 micron process has been "eliminated, shortening the time needed to arrive at .13 micron in 2001." This echoes strangely with an Intel presentation we reported earlier this year, in which the chip giant said .18 micron was "really" .13 micron. (Stories: Intel confuses world+dog over .18 micron and slides, Why Intel's .18 micron is really .13 micron) George Scalise, president of the SIA, did little to clarify the position yesterday. He said: "This is the successful start of the international effort to produce a global roadmap, which will culminate in the 1999 Edition of the Roadmap." There is more information on these "roadmaps" here. ® * And the SIA continued to say the chip market is recovering, with sales in March up seven per cent to $11.6 billion, compared with March last year.
Mike Magee, 08 May 1999

Why Halla's head should be on the block

We first met Jerry Rogers, who co-founded Cyrix, about nine years ago at an Etre conference, in a bar. He's a tough boy. The Texan, who formerly worked for Texas Instruments, had some strong ideas about microprocessor design and was finding venture capital to put those ideas into practice. The guy has guts and also a sound engineering background which made us listen to what he had to say. You didn't need fabs to compete with Intel, he said, and you didn't need to do what AMD was doing then, essentially cloning Intel designs. Jerry Sanders III's statement that "only real men have fabs" was just talk, Rogers said. Over the next few years, we saw Intel boot up its profit and loss based legal division several times in a bid to knock Cyrix designs on the head. We were amused to see a fake cemetery with this headstone at its Richardson, Texas HQ, when we visited Cyrix a few years back. Cyrix had beaten the Intel legal machine. Rogers, after he'd shouted down a corridor: "Who let The Register into this building," proudly showed off the patents under the Cyrix belt. AMD had fared worse with Thomas Dunlap, Intel's legal counsel, and bought in the technology it craved and needed from NexGen. But whether real men needed fabs or not, Cyrix did need some place to make chips and it struck a deal with IBM Microelectronics (and SGS Thomsen) to make the pesky things. Unfortunately for Cyrix, IBM Micro never played fair. It wouldn't use its chip allocation in its own PCs, preferring instead to sell them through the merchant channel at prices which undercut Cyrix prices. Eventually, Rogers left Cyrix and, as we know, National Semiconductor took over the company. IBM Microelectronics eventually became an albatross around NatSemi's neck and the fabbing deal was terminated, leaving Brian Halla, CEO of National, to foot a whopping deal. NatSemi has made losses over the last seven quarters so it isn't too much of a surprise that its discrete x.86 business has gone. But it is a shame. Despite the fact that the Cyrix roadmap was flagging, it still had some top engineers working for it. Intel now has even less competition to worry about, with only AMD, Rise and IDT left in the game. But if Cyrix has gone, surely Mr Brian Halla should follow? He, after all, masterminded the $550 million takeover, and now only rubble remains. His system on a chip idea might work, but there's lots of competition there as well. It turns out that the gravestone that Jerry Rogers put in the reception area of the Richardson site was actually engraved with the wrong epitaph. Intel Inside lives. RIP, Cyrix. ®
Mike Magee, 08 May 1999

Intel's 550MHz PIII comes in a week's time

As predicted here earlier on this year, Intel will debut a 550MHz Pentium III at $744 on May 16, while other members of the PIII family experience price reductions. The 550MHz part will arrive on platform Pentium III, unless there are overheating problems that Intel has not yet fixed. (Story: 550MHz Pentium III has overheating problems) The 500MHz PIII will fall from $637 to $482, and the 450MHz Pentium III from $411 to $268. The Pentium II family shows all the signs of packing up its bags for the trans-Siberian train to Intel's gulag. The 350MHz Pentium II will, essentially, die on the journey. Its price will be $163 for the second price round in a row, while the PII/450 will cost $268 and the PII/400 $193. Because the Pentium III/450 costs $268 from May 16, it's gulag time for its Pentium II equivalent. There will be no changes in prices for Intel Celeron parts, given that the 466MHz Celeron is now the premier model and has some way to go before its price declines. All prices above are when you buy 1,000 parts. Intel, meanwhile, is denying that the Pentium III/550MHz has over-heating problems. ® See also Intel desktop chip prices How Intel's geese lay golden eggs
Mike Magee, 08 May 1999

Sun switches Java standards efforts from ISO

Alan Baratz, Java Software president, announced yesterday that Sun has decided to standardise Java through the Geneva-based ECMA rather than through the ISO directly. The JTC1 (Java Technical Committee) changed the rules in January, so that the committee would have responsibility for Java enhancements. Sun had the status of a publicly approved submitter (PAS). The Open Group and The OMG are backing the submission. The Java 2 Standard Edition submission (J2SE Version 1.2.2), which Baratz says was developed through the Java Community Process, will be made next month at the ECMA General Assembly in Kyoto and consists of the Java language spec, the JVM spec, and the Java API core class library. Once ECMA has approved it, it will go to ISO by a fast track. It may well be that Sun is wary of the pressures that Microsoft brought to bear on W3C in the "extend" phase of its embrace, extend, and extinguish strategy. ECMA, now called the European Association for Standardising Information and Communication Systems (formerly the European Computer Manufacturers Association), was created in 1960. It became a liaison member of TC97 of ISO and IEC for standardisation in the computing field, and subsequently part of ISO/IEC JTC1. Sun had an amusing flirt with ECMA in 1993. The aim at the time was to make it possible for Unix users to run 16-bit Windows apps in an open environment, without DOS or Windows, and also without emulation or modification to the Windows program. Sun did this because Microsoft would not license Windows source code to Sun on reasonable terms, although it had licensed it to Insignia, Bristol and Locus. In March 1994, Sun hosted a meeting of Microsoft, Novell, the European Commission and Scottish Enterprise at Gleneagles to discuss the desktop environment and Windows/open systems convergence. Imagine Microsoft's embarrassment when Sun made the meeting the launch of a Public Windows Interface (PWI), and followed this up by putting the proposal through ECMA. At the time, Microsoft was not a member of ECMA, and found itself powerless, although later of course, Microsoft became a member. And here's a rather jolly miscellaneous fact: Rule 1.1 of ECMA says: "The English language, as written in the United Kingdom, will be the official language of the Association." ®
Graham Lea, 08 May 1999