The chip wars are far from over, as Intel looks to exploit processor 'headroom' to crank up Pentium III performance.
Intel, the world's leading chip manufacturer, is evaluating its options for introducing Pentium III chips running faster than 1GHz in the second half of this year.
The new chips, if introduced, would raise performance of mainstream PCs while keeping Intel competitive with rival Advanced Micro Devices (AMD).
The motivation to raise performance comes, in part, because of the available "headroom" on the chip. But analysts said it will also come out of necessity, due to competition with AMD and demand from its PC-maker customers.
Headroom is, essentially, the amount of additional clock speed a version of a processor can be enhanced with after its introduction, due to manufacturing refinements. Intel's current-generation Pentium III chip, according to convention, should have an additional 200MHz left in it.
The means Intel's 1GHz (1,000MHz) Pentium III Coppermine chip could, in theory, go at least as high as 1,200MHz. The chip, which replaced Intel's previous "Katmai" generation of Pentium IIIs, was introduced at 600MHz. (Coppermine Pentium III chips also are available at 533MHz and 550MHz.)
Likely speeds for the future gigahertz-plus Pentium III, based on Intel's 133MHz bus, are 1,066MHz, 1,133MHz and 1,200MHz, with 1,266MHz being the next possible speed. But will Intel exploit its headroom? Intel officials have been cagey about answering the question of what comes next with the Pentium III. "If it is possible, we will do it," Pat Gelsinger, an Intel vice president and general manager of the company's Desktop Products Group, told ZDNet News recently.
"I don't think that they've decided yet where to go. A lot of it depends on what they see out of the factory in terms of yields," said Dean McCarron, principal analyst at Scottsdale, Ariz.-based Mercury Research. "Market demand is another factor."
But the main factor may be the competitive landscape.
AMD in May is expected to launch the next-generation of its Athlon processor, the main rival to Intel's Pentium line. The new chip, based on the forthcoming "Thunderbird" core, promises to be a better performer than the current generation, but it is not expected to be much more expensive. They will come in a wide range of clock speeds as well -- both higher and lower than the current top clock speed of the 1GHz Athlon chip.
"When the competitive environment changed (with the introduction of AMD's Athlon), and we entered this tit-for-tat mode where Intel and AMD were trying to outdo each other, the entire market became much more reactive," McCarron said. So, to stay competitive, Intel will have to match Thunderbird with its Pentium III chip while it works toward the launch of its next-generation processor, code-named Willamette.
Willamette's claim to fame, Intel said, will be performance. The company demonstrated the chip running at 1.5GHz last February. It will offer integrated Level 2 cache, a 400MHz system bus and dual Rambus direct RAM (RDRAM) channels and a new multimedia instruction set, among other features.
Although an enhanced Pentium III could conceivably match megahertz with Willamette, Intel could avoid competing with itself by pricing Pentiums lower and claiming greater levels of performance for Willamette, McCarron said. PC makers would be able to offer, at least for a time, lower-cost Pentium III-based systems alongside more-expensive but higher-performance Willamette-based PCs.
PC makers would like to see both available simultaneously in order to give corporate customers time to make the transition.
That could be an important issue for business users, because premiums for high-end components -- including the new Willamette chip and Rambus memory -- could make Willamette PCs costly. "If the price differential between SDRAM and Rambus doesn't resolve itself, the price spread could be even wider," McCarron said.
Meanwhile, 1GHz and faster Pentium III-based PCs, which began selling between $3,000 and $6,000, will reach much lower average selling prices in the second half of 2000.
Now that the gigahertz barrier has been broken, what's next? Will anyone care when we pass 2GHz? Michael Caton thinks not. Go and read the news comment at AnchorDesk UK.