Intel’s next chips will be too clever for today’s programs. Jack Schofield reports on a scheme to help rewrite the way we work
The PC industry is preparing a great leap forward that will more than double the power of computers. The problem is that there is hardly any software to exploit the extra capability. To bridge this gap, Intel, the world’s biggest chip maker, has just announced a $250-million fund to support the switch from 32 bit to 64 bit processors when the company delivers Merced, its first 64 bit Intel architecture (IA) chip, next year.
The Intel 64 fund is being bank-rolled by Compaq, Intel, Dell, NEC, Silicon Graphics Incorporated (SGI) and Hewlett-Packard. A dozen non-computer companies – including Bank of America, Ford Motor, Reuters and SmithKline Beecham – are also contributing to the fund, which will be managed by venture capital investment company Morgan Stanley Dean Witter.
The last processor transition started in 1985 when Intel introduced the 32 bit 80386 (386) to replace the 16 bit 286 chip used in the IBM PC AT (advanced technology). However, it was not until Microsoft Windows 95 appeared a decade later that most users gained the benefits of 32 bit computing, and even that retains 16 bit operations for backwards compatibility. Intel hopes the $250-million investment in 64 bit software will make the transition faster this time.
Microsoft is well advanced with the development of a 64 bit version of its next generation Windows 2000 operating system, formerly known as Windows NT (new technology). However, the Intel 64 fund will target “a full range of volume operating systems” including Unix, which is the standard on the large server computers and high-powered graphics workstations where Intel’s 64 bit chips will first come into use.
In this market, Intel’s ideas are far from new. Workstations and servers started moving to 64 bit processing when Digital Equipment Corp- oration announced the Alpha, an advanced reduced instruction set computer (Risc) processor, in 1992. But the IA64 fund’s backers show the failure of the Risc processing revolution to kill Intel – one of the sillier articles of faith among supporters of Risc chips used by companies like Acorn, Apple and IBM.
Hewlett-Packard, the first com pany to commercialise Risc, has been a co-developer of IA64 chips with which it plans to replace its own HP-PA (precision architecture) Risc chips.
SGI has floated off its Risc chip division, Mips, and switched to Intel. Digital was taken over by Compaq, the world’s largest manufacturer of Intel-based PCs, and the Alpha is now made under contract by Intel.
Intel expects that Merced and the next IA64 chip, code-named McKinley, will enable it to expand its share of the server and workstation businesses. This does not mean it has lost interest in the PC business. There will be a “price point” version of the IA64, code-named Deerfield, for use in desktop computers.
Intel is also working hard to win back the low-cost PC market. Last year its market share fell from 85% to 76%, according to United States-based Mercury Research, as PC manufacturers adopted cheaper Pentium compatible chips from AMD and Cyrix. But over the past six months, Intel has repeatedly cranked up the speed and slashed the price of the Celeron, its low-end version of the Pentium, to attack this market.
As a result, Brian Halla, chair, president and chief executive of National Semiconductor (NatSemi), Cyrix’s parent company, announced last week that it would immediately cease “slugging it out” in the PC processor market. “Believing that Intel would not and could not compete in the low- end [PC] space was a mistake,” said Halla.
AMD is also struggling. In March the company said it would lay off 300 workers. Shareholders called for the resignation of chair and chief executive Jerry Sanders, who has run AMD for 30 years. He asked for more time to try to compete with Intel. “You should reserve judgment for at least another year,” he said.