Intel keeps ratcheting up the pressure on AMD in the race toward smaller chip processes, opening up a $3 billion factory in Arizona this week slated to churn out 25,000 45 nanometer Penryn wafers a month. This comes barely a year after AMD moved from 90 nanometer to 65 nanometer with the Athlon 64 X2, putting AMD once again in catch-up mode if it hopes to stay competitive.
For enterprise users, however, the more salient question is what kinds of benefits come with Penryn-equipped servers, desktops and laptops, due to show up within six months. That the Penryn will result in more powerful hardware is no secret, but it also delivers on power consumption and overall cost, according to people who've studied the designs and have seen the preview devices in action.
But what kind of power are we talking here? While most pundits put the clock speed at 3.0 GHz -- not the fastest by any means, but the fastest for a quad, so far -- Intel's designers are boasting that a 3.2 GHz speed is more likely at launch, with enough headroom for 4.0 GHz before long.
Apple fans seem particularly jazzed at the Penryn's arrival, with the Mac Pro slated to be one of the first to use the Penryn. Insiders say the company is planning an eight-core configuration with a 1600 MHz bus and a 12 MB L2 cache. Even entry-level machines sporting dual-core Penryns will range from 1066 MHz to 1333 MHz bus speeds.
With all this talk of 45 nanometer and, soon, 32 nanometer, chips, it's helpful to remember just exactly what those numbers mean. This article on EDN points out that back in the days of 350 nanometer chips, they used to indicate the line width on the wafer. But since current printing technology is incapable of anything less that 193 nanometer, the number now refers to the process that produces bulging lines and jags that have to be masked and decorated, taking up space on the chip. The result is still a more powerful chip, but it is by no means as elegant a design as in the past.
Overall, then, it's hard to imagine that the Penryn will be anything but a winner. Tighter designs mean more processors per chip, simultaneously lowering costs and power consumption while boosting performance. Now, it's up to developers to find new ways to harness that power.