Notebook buyers will have to wait until the second quarter for Dothan, the 90-nanometre version of Intel’s Pentium M chip.
In order to make sure Dothan could be manufactured in high volumes, Intel needed to modify some circuits on the chip, Intel president and chief operating officer, Paul Otellini, said.
The chip had been expected to ship this quarter, but will now ship in the second quarter, he said.
The modifications to some of Dothan’s circuits were not related to any thermal problems resulting from the jump to the 90-nanometre process technology, an Intel spokesperson said.
The chip industry is preparing to move its chip-making equipment from the 0.13-micron process technology generation to the 90-nanometre process generation. The number refers to the width of the circuit lines on a processor. As chips reach these tiny sizes, it becomes more and more likely that the electrons moving through the chips will be able to escape the ultra-thin circuit walls and leak out as heat.
Analysts have suspected that thermal issues resulted in the delay of Prescott, the 90-nanometre version of the Pentium 4, to the first quarter from an expected fourth-quarter launch. Heat dissipation is always a concern of chip designers, but the issue is magnified for notebook processors that need to deliver high performance while consuming little power in a small area.
But if it’s not related to heat dissipation, the delay could be chalked up to just about anything that chip vendors normally experience when validating a new part, editor-in-chief of US publication, the Microprocessor Report, Peter Glaskowsky, said.
“Things like this come up all the time, although usually they’re noticed sooner,” he said.
Intel might have had problems delivering the yields it needed to produce chips at higher clock speeds, Glaskowsky said.
Processors were all cut from the same wafer, and designed to run at a certain clock speed. Not every processor is capable of running at the target clock speed, due to minor imperfections, but those that fail to meet the target will usually work reliably at a lower clock speed.
If Intel couldn’t get enough chips that worked reliably at its target clock speed, it might have needed to redesign some of the circuitry to reach those speeds, Glaskowsky said.