Coppermine key for Intel's holiday

The fourth quarter for chip giant Intel is going to largely depend on a chip code-named Coppermine.

The "Coppermine" processor, a deluxe version of the Pentium III debuting October 25, will likely become one of the major issues for the semiconductor industry over the next three months.

What makes the chip special? At a speedy 733 MHz and faster, it should excite consumer demand. At the same time, it will cost less to manufacture than current Pentium IIIs, making it an accountant's dream.

If sales take off, the chip will help reinvigorate the performance segment of the market and help Intel shrink the performance gap between its products and AMD's high-end Athlon chip.

"It puts the Pentium III on more or less a clock-for-clock parity," said Nathan Brookwood, principal consultant at Insight 64. "When Intel pops with a 733 MHz, there is a likelihood that they will be the fastest guy on the block."

On the other hand, if consumers continue to flock to budget computers, Intel may continue to see the average selling prices for its chips decline and, with that, less-than-spectacular growth from Wall Street's perspective. In fact, the chip has already had an effect. Originally due in September, Intel put Coppermine off until October, a move which contributed to lower-than-expected earnings.

"We underestimated the effect of the push-out of the high-performance, low-cost processor," said Andy Bryant, Intel chief financial officer, during a conference call with analysts. "We think we have a better cost story and a better product story going into the fourth quarter."

Coppermine essentially is the code-name for the next generation of Pentium III processors. Current Pentium III chips are made on the 0.25-micron manufacturing process, a designation which means that several of the transistor-level components on the chip measure 0.25 microns in width. These processors also come with 512KB of secondary cache, sort of a data reservoir, that comes on chips located close to the processor.

The new chips will be made on the 0.18-micron process, resulting in smaller, less-expensive, and faster chips. The smaller dimensions will also mean the chips will run cooler, and allow Pentium IIIs to be put into notebooks for the first time.

Advancing the manufacturing process also permits Intel to integrate an "advanced transfer cache." The advance transfer cache contains 256KB of memory, but it's four times as fast as standard caches. In the end, the cumulative effect of these changes result in cheaper processors that deliver a higher level of performance that would be achieved by merely boosting the speed.

"The advance transfer cache will improve the performance by about 10 percent," said Brookwood. "They [Coppermine and Athlon] will be very close." Coppermine, however, will fall behind Athlon if Intel is unable to pair it with next-generation Rambus memory. Some estimate that Rambus memory won't hit PCs until November at the earliest.

On October 25, consumers should expect to see an onslaught of these systems. Several major PC makers will likely appear at an event in San Jose, California, to mark the release of the processors. Desktop versions running at 700 MHz and 733 MHz will be released, while versions for notebooks will come out at 450 MHz and 500 MHz.

The cost dynamic, however, will be equally important. The lower cost will allow Intel to bring up its gross margins, stated Bryant and other Intel execs. The chip's delay was cited as one of the two reasons profits dipped this past quarter.

"We didn't have as much success as we would have liked in the higher end," he said.

This also spells problems for AMD. Athlon currently is a more expensive chip to manufacture, according to recent estimates from MicroDesign Resources. Further, Intel will change the chip packaging in the future to make Coppermine even cheaper.

Still, Intel isn't out of the woods yet. Consumers still appear to be drawn more strongly to ultra-cheap computers, which could blunt interest in Coppermine computers. The serial delays Intel has experienced also bring up the possibility that something may occur in the future.

"We are in general moving our road map faster than in the past," said Paul Otellini, general manage of the Intel Architecture Business Group, in response to question about the various bugs found in high-end Intel efforts this year. "I would like to think they are random."