3rd > May > 1999 Archive

The Register breaking news

UK boffin plans Linux box for comeback

Sir Clive Sinclair, the 80s UK computer guru, is being strongly tipped to make as comeback with a machine running Linux. It's not Sir Clive's fault that the latest rumour outbreak was in yesterday's Sunday Express, although the material the Express majored on -- his love life -- is his fault, we suppose. Rather more worrying for The Register is the fact that well-known UK journalist Chris Bidmead claimed in a recent column that he'd be sending a copy of Linux to Sinclair. This means that if Sir Clive does go with a Linux box, Chris will become even more self-important than he is already. Linux does, however, have obvious attractions for Sir Clive. His previous creations have been low-cost, at least intended to be robust, and have gone for innovation rather than whatever the industry standard has been perceived as. The ZX family of little black boxes (early computing appliances, perhaps?) brought home computer prices down to levels (£125, anyone?) below anything we've seen for some years. Sinclair Research lost the ensuing price war, aided to some extent by Timex lousing-up Sinclair's entry to the US market (the Timex 2000). A later machine, the Z88, wasn't commercially successful, but was the weapon of choice for quite a few UK hacks for some years, and could maybe be seen as a kind of signpost to any future machine Sinclair will produce. Sinclair has a long-standing aversion to moving parts, wasted machine horsepower and pointless adherence to industry standards. So the Z88 was a thin black pad with rubberised keyboard, a smallish mono (non-backlit LCD), Z80 CPU, EPROM storage and no disks of any description. It had a suite of built-in applications, ran off four AA batteries and was a pretty successful dedicated word-box and mobile email terminal. If Sinclair was putting together something similar today he'd be inclined to go for a low power CPU that delivered maximum bangs per buck, so an ARM would be attractive. Linux these days seems a pretty obvious candidate for an OS, and would give a development team something solid to build on. Sinclair would however have to figure out how to deal with various developments that have taken place since his last outing. The machine will have to deal well with the Web and communications, and these might be the areas where he would try to introduce the characteristic Sinclair innovation (unfortunately, the Sinclair innovation frequently turns out to be only nearly a good idea). The display will be a problem if he goes for a portable, as Sinclair characteristically resists components that cost a lot of money. Storage will also help determine the kind of machine he produces - he may have got over his dislike of moving parts by now, but it's not guaranteed. For what it's worth, we reckon Sir Clive is likely to find a low-cost, robust machine positioned somewhere in the mobile and/or appliance areas attractive. He could well cut corners with the display, maybe making it at least partially home Internet box, and might even go for consumer electronics flash storage rather than rotating media. But don't hold your breath -- the Z88, although quite fun, was years in the making, and barely made it out of the UK at all. ®
John Lettice, 03 May 1999
The Register breaking news

More Gates smoking emails support Caldera case

Microsoft did build software that checked for the presence of rival operating system DR-DOS into its own MS-DOS, according to evidence produced by Caldera. There also seems to be some evidence that the company at least intended to use some international versions of Windows to stop applications running with DR-DOS. Microsoft has admitted it devised a code-checking routine, but has denied including it in production software (Gates Smoking emails). DR-DOS users from way back have however insisted to The Register that this code did appear in production versions, and Caldera, which is suing Microsoft for anti-competitive activities, says that this and other weapons were used to undermine DR-DOS sales. Caldera's evidence suggests that Bill Gates was a driving force here. Asking for information on "what things an app would do that would make it run with MS-DOS and not run with DR-DOS," Gates writes in 1988: "I am not looking for something they cant [sic] get around. I am looking for something their current binary fails on." The answer he got from his staff was that there were things an application could do to break DR-DOS, but that these were things an application had no business doing. Which makes the Korean versions of Microsoft applications from this period interesting. Caldera does however have some traffic from Microsoft Korea to Redmond from the period: "Bill Gates ordered to all application business units to include checking routines of operating environments and if it is Microsoft DOS, nothing will happen. But if it is non MS-DOS (such as DR-DOS), application will display messages saying that "This application has been developed and tested for MICROSOFT MS-DOS. Since you use different environment, this application may not work correctly. . . ." It gets more interesting. "The question from [Microsoft Korea] is 'How to check the DOS is MS-DOS or clone'. [Microsoft Korea] wants to include such routine in Hangeul Windows so that Hangeul Windows can run only Hangeul MS-DOS. Could you tell me to whom I can ask to resolve this problem?" Gates denies that such a check was implemented, but according to Caldera: "At least one Korean Windows version from this era contains this warning: 'Hangeul Windows 3.0 should be executed on Hangeul MS-DOS. For correct execution, please run on Hangeul MS-DOS. Press any key to continue.'" Not a breakage, perhaps, but worrying for users who don't know the truth. There may however have been breakages. Caldera says: "Microsoft has been unable to produce copies of the Korean versions of their application software from this time period. Even so, evidence in the record suggests that -- consistent with Gates' directive -- Microsoft coded 'software locks' into several of its Hangeul (Korean) applications, including Word, Works, and Excel." Caldera is still seeking copies of this software - any of our Korean readers able to oblige? Microsoft is also claimed to have denounced DR-DOS as "a copy of MS-DOS, and anybody that used that product would be sued by Microsoft" in seminars in Korea. Something called the "non-tested DOS warning code," which alerted/scared users if a non-MS-DOS OS was detected, was devised, and according to Microsoft's then MS-DOS product manager Microsoft developers "have committed to implementing it in all new MS application and language releases from this point forward, including international." The warnings themselves hurt, says Caldera, while FUD campaigns and that old leveraging of the OEM relationship hurt some more. Says one Microsoft manager: "It only takes a couple of reports about non-compatibility to give the kiss of death to a PC: we've seen that on the hardware side as well in as the operating system area." A purported DR-DOS bug list was put together for the OEM sales force, who "would use that information as they saw fit in competitive situations." These bugs seem to have been somewhat exaggerated, as MS-DOS 5.0 product manager Mark Chestnut wrote that he disagreed with the tactics of "taking what you perceived to be a minor bug and trying to make a big deal, a big story out of it." Chestnut was replaced by Brad Chase in November 1990, and according to Caldera he "was not so similarly restrained." There's plenty more of this stuff, and we'll no doubt get back to it shortly - but here's a snippet from the period when the story - apparently - came to an end, as Microsoft engineered the switchover from Dos to Windows. Gates wrote: "I doubt they will be able to clone Windows. It is very difficult to do technically, we have made it a moving target and we have some visual copyright and patent protection.." (Our italics) So making Windows more complex and moving the goalposts, while extending Microsoft's IP protection, was a Gates objective? Possibly: "First, we have to make sure Windows isn't easy to clone for both technical and legal reasons. Who is smart that thinks about this -- patents and such. I can do it at some point and I think we will be able to achieve it. DOS being fairly cloned has had a dramatic impact on our pricing for DOS. I wonder if we would have it around 30-40 % higher if it wasn't cloned. I bet we would!" Naughty Bill. ®
John Lettice, 03 May 1999
The Register breaking news

Can AMD seize the corporate day?

Analysis Despite getting heckled by an octegenarian, ceding control of day to day issues to Atiq Raza, and appointing Bob "walk in closet" Palmer to the board, Jerry Sanders III, CEO of AMD remains upbeat about his company's future. So much so, that Sanders wheeled out his famous gorilla slides again. He is fond of describing Intel as a gorilla. In terms of scaleability, a word US computer firms use over and over again, if Intel is a gorilla, then AMD is not even a chimp. A Mountain Gorilla weighs around 600lbs. Intel had a turnover of $26 billion in 1998, while AMD had a turnover of $2 billion. In terms of primates, if Intel is a gorilla, AMD is a Mandrill -- one of the largest of all monkeys. AMD is not even a chimp, which weighs around 150lbs. While Sanders' portrayal of Intel as a gorilla is obvious bunk, the chip giant is concerned about AMD. Andy Grove's famous dictum: "Only the paranoid survive" is alive and well in Santa Clara. It means that Intel will take every step to ensure that it stays big and AMD stays weeny. AMD has a number of factors on its side, some of which are emotional ones. Intel is seen as being the Great Satan of Chips, the topdog, while AMD is sometimes cast as the underdog, bravely pitting itself against the might of Goliath. Intel, of course, only has itself to blame for this perception. Over the years it has traded, it has gained a reputation for fierce and aggressive competition. The head of Tandem UK once said to us: "You don't tread on the tail of a tiger". Intel still has lessons to learn about treading lightly. At the shareholders' meeting, Sanders said that Intel's semiconductor revenues exceeded the combined revenues of the next five largest US companies combined. "We are one of those five companies, and the only one to compete head-to-head with Intel in microprocessors," he said. This statement may be either wisdom or folly. Certainly, other companies such as Cyrix have preferred to take a back seat in this strange contest between the gentle Mountain Gorilla and the Mandrill. Cyrix says it can make a tidy sum of money without getting involved in head-to-head battles with Intel. IDT and Rise probably feel the same. Sanders said that earlier on in this decade, AMD made tidy profits from its 386 and 486 alternatives and was faced with a barrage of law suits from Intel as a result. In fact, we remember this period well. As Intel phased out some of its processors, AMD manufactured clone chips and mopped up the aftermarket. That changed in 1995 when AMD acquired NexGen and with it Atiq Raza and other high flying execs such as Intel's Vinodh Dham and Dana Krelle. This was probably the best thing AMD did in the last five years. (See, for example, what The Register said about the takeover.) NexGen technology was perceived as being technologically advanced, while AMD was floundering around attempting to develop processors that it had real problems designing. Raza, Dham and Krelle, frankly, had a far more professional way of approaching AMD's market than the people they replaced. PC vendors do not like relying on one source for products. When Intel introduced the Pentium Pro, that had all types of problems which caused headaches for its customers. Unfortunately, Intel was the only company then supplying chips suitable for the server market. AMD has made a difference to the x.86 market. It forced Intel to follow it, forced down prices and even, if our well informed source at the chip giant is to be believed, resurrected Katmai because of the threat AMD posed. Intel was late to realise the threat other companies posed in the sub-$1,000 market and rushed out Celeron processors that were underperformers. Now AMD has a large share of the US retail market, 60 per cent, according to Sanders. He said: "We are mounting a successful challenge to an enormously rich and influential company that we can neither out-invest nor out-produce. We are challenging with the most powerful competitive weapon of all: innovation. Or, to put it another way, with better ideas." He obviously realises the dangers in taking Intel head-on: "The determination to be the nucleating force for an alternative platform represents simultaneously one of our greatest opportunities and toughest challenges." He said that fierce price competition from Intel compounded AMD's yield problems, resulting in its largest quarterly loss ever. "As the dominant processor supplier, Intel sets the pace of the processor roadmap. It controls the schedule of introduction of higher-clock-speed devices, and in response to competition from the AMD-K6 processor family, it has accelerated the pace of the clock speed race." Sanders realises that the clock speed, raw MHz fight is a spurious one. He said: "Numerous benchmarks exist for comparing PC processors, and they are subject to much marketing manipulation. Clock speed has thus become the principal battleground of the marketing war for consumer acceptance. AMD is committed to matching Intel clock speeds even as we provide more system performance at a given price point." Because AMD decided to play the raw MHz game, Sanders admitted that yields were affected. "To support our customers, we pushed both our process technology and our AMD-K6-2 processor design to achieve maximum clock speed...Tuning the process and the design to achieve maximum clock speed while maintaining acceptable yields is an ongoing challenge that ideally involves a series of engineering iterations to achieve optimum results. "Pushed by Intel to match an ever-faster pace of new product introductions, and pulled by the demands of our largest customers for competitive higher-clock-speed devices, we encountered yield problems last December that carried over into the first eight weeks of 1999 with a devastating impact on our aggregate production in the first quarter," he said. Sanders said that AMD will spend $750 million on R&D this year and claimed that the yield problems of January and February are a thing of the past. That means, he added, that AMD is now in a stronger position than ever in process technology, production capacity and platform competitiveness than ever. He claimed that AMD can attain the right levels of revenues to sustain profitability and achieve adequate profit margins, but only if the company continued to do well in the consumer sector and achieve better penetration in the commercial sector. "Corporations buy a lot more computers than consumers do, and the current pricing structure in the commercial sector supports higher processor pricing. Today, our participation in this segment is insufficient to impact Intel's ability to extract monopoly prices there. This enables Intel to subsidize their ability to be extremely aggressive where we do compete," he admitted. Making K6-IIIs at speeds of 500MHz and above is a significant challenge, he said, and means it will have to move to .18 micron technology using aluminium interconnect. But the K7, claimed Sanders, will allow AMD to break into the enterprise market. "The AMD-K7 processor will be the first seventh-generation processor on the market," he said. "Based on initial product sampling, we are encouraged by the extremely positive feedback from customers. Nonetheless, the challenges we face in assuring the success of the AMD K7 processor are substantial and unprecedented in our history," he said. He said that AMD will move AMD K7 technology to copper interconnect technology at its Dresden Fab 30 factory in Q3, giving raw MHz speeds of 1GHz or more by the end of 2001. Of course, Bob Palmer was the CEO of Digital before Compaq acquired it, and so has more than a passing knowledge of the Alpha platform. And, as we revealed exclusively earlier this year, there are plans to make the Alpha and the AMD K7 platforms sing together. Atiq Raza will act as the day-to-day executive in charge of the company and he has a formidable reputation. (See story: Alpha chip to clock 1.4GHz in copper) All AMD has to do now is to persuade the corporate market to switch to the K7 platform. That, coincidentally, is what Compaq has to do with the Alpha platform. This is a matter of life-and-death for AMD, but Sanders has never been afraid to take risks. Nor, for that matter, has Intel, in this risky chip business. ® See also Compaq, AMD continue to squeeze Intel
Mike Magee, 03 May 1999
The Register breaking news

FTC to appoint guardian for Alpha

A year ago From The Register No. 75, May 1998 The Federal Trading Commission (FTC) is making an extraordinary condition on Digital's sale of its semiconductor operations to Intel - a trustee will have to be appointed to supervise the licensing of Alpha to other manufacturers, and if the FTC doesn't agree with decisions that are made, it will step in and look after the chip's future itself. This condition is included in the consent decree permitting the Intel-Digital deal to go ahead, and is an unprecedented piece of government intervention in the free market, in order to preserve it. The decree seems tacitly to conclude that, left to their own devices, Intel and Digital (or Compaq. The new parent company) would take decisions about the future of Alpha that would not necessarily be in the interests of free competition in the 64-bit market. Digital's Alpha technology has become more obviously key to the 64-bit generation over the past few months, so it makes sense to the FTC to force it into free licensing, if not quite into the public domain. Negotiations over Digital Alpha licensing in the future will certainly cover Samsung, whose current position seems unclear. The company has been a second source of Alpha for some time, and in February struck a deal which allowed it to develop its own high volume version for 'specific markets,' and at the time it seemed clear that Samsung could make design changes, so long as it maintained compatibility. It has however been reported elsewhere that the FTC is under the impression that Samsung still only has manufacturing rights, which contradicts February's announcement. Samsung is however now rather important to the success of the Intel-Digital deal. The FTC wants Alpha to be licensed to two other manufacturers, and the company has the virtue of actually wanting it, whereas others like AMD and IBM are playing hard to get. So now all Samsung probably needs is some hard currency to run with it. ®
John Lettice, 03 May 1999
The Register breaking news

HP buys Transoft

Hewlett Packard said today it has bought Transoft for an undisclosed amount. The move is part of HP's efforts to grow the storage area network (SAN) market, and to compete with companies like Compaq. Transoft makes fibre channel based SAN networks for the NT and Unix manager, including SAN manager. ®
Mike Magee, 03 May 1999
The Register breaking news

Intel talks 600MHz StrongArm

Chip giant Intel said today that it will introduce StrongARM devices of speeds between 150MHz to 600MHz but with consumption of only 40 to 450 milliwatts. Intel's nextgen StrongArm chips will use superpipelined Risc architecture including seven stage and eight stage memory pipelining to hit the higher clock speeds. It will also introduce dynamic branch prediction and data bypassing to help data throughput. The company will use dynamic voltage management for better battery life and a set of low power modes. The chips will sample in the first half of next year, the company said. ®
Mike Magee, 03 May 1999
The Register breaking news

Compaq slashes prices on Linux Alpha servers

Compaq today announced it has cut the price of its Alpha low end servers by $2,000 for the Tru64 DS20 boxes and by 20 per cent for the Linux boxes. The dual server boxes are set to fall to a level where they directly challenge Intel servers. A DS20 using two Alpha 500MHz 21264 chips now costs under $12,000. Prices of other Alpha boxes are set to fall steadily later this year as Samsung's Alpha group rolls out its solutions. ®
Mike Magee, 03 May 1999