eWEEK content and product recommendations are editorially independent. We may make money when you click on links to our partners. Learn More.
Manufacturers have begun tapping new processes that are capable of producing smaller, more power-efficient processors, in order to help dual-core chips proliferate in 2006.
Dual-core chips, which include two processor cores in place of one, have been limited to a relatively small number of high-end desktops and servers to date.
But now manufacturers such as Intel Corp. and Advanced Micro Devices Inc. are looking to new 65-nanometer chip manufacturing processes—the means by which manufacturers knit together the transistors that make up the circuits inside their chips—to help them expand the market for the chips in the coming year.
Chip manufacturers generally move to new and successively smaller manufacturing processes every two years.
The shift, which costs billions and takes years of development, allows them to produce chips with greater numbers of transistors, but still make them smaller by packing those features more tightly together.
The cycle—as dictated by Moores Law, which observes that chip transistor counts will double every two years—has allowed the chip makers to drive up performance with each generation of manufacturing technology.
However, with the coming generations move from 90-nanometers to 65-nanometers, the chip makers will emphasize their dual-core designs.
Thus, the new crop of 65-nanometer dual-core chips will run faster, incorporate larger onboard memory caches, and still have space to add circuitry to support virtualization or other on chip features, while fitting within power budgets similar to those of todays dual-core chips.
“The really big challenge in any [manufacturing] technology transition is getting it right in smaller geometries,” said Nick Kepler, vice president of logic technology development at AMD.
However, once there, he said, “You could produce a [65-nanometer] chip thats the same size [as a 90-nanometer chip] and put two cores on it. You can just fit more in it.”
Big-name chip makers such as AMD, Intel and IBM all report that, at a minimum, they have begun the early stages of 65-nanometer production. That means businesses and consumers can expect new crop of 65-nanometer chips over the course of 2006.
For its part, Intel appears to be the first brand-name chip maker to hit the new mark. Intel said its shipping Presler, a 65-nanometer, dual-core desktop processor, for revenue, and aims to ship hundreds of thousands of the chips by the end of this year.
Presler, which will come out in systems in early January, just about two years after Intels first 90-nanometer Pentium 4 chip, will be joined by Yonah, a dual-core processor for notebooks thats also due in January, and a Xeon server chip, dubbed Bensley, that will also arrive in the first quarter of 2006.
The 65-nanometer mark “equals high volume production of dual core in all three segments—thats the bottom line,” said George Alfs, a spokesperson for Intel.
Presler, in keeping with 65-nanometer manufacturings advantages, is expected to offer more clock speed as well as extra cache. However, its expected to fit within current dual-core Pentium D chips envelopes for power consumption.
The first Preslers are expected to top out at 3.4GHz and offer twin 2MB caches. Intels Pentium D, on the other hand, hits 3.2GHz and offers two 1MB caches. Intel will offer the chips for both corporate desktops and consumer machines.
Next Page: Costly chip investments will pay off.
Costly Chip Investments Will
Pay Off”>
To date, chip makers have invested billions of dollars to ensure their ability to move to 65-nanometer production on time.
Intel has begun production of 65-nanometer chips at its Fab 12 factory in Chandler, Ariz., after an 18-month, $2 billion renovation. Fab 12 is Intels second 65-nanometer factory.
By the end of 2006, the chip maker said, it expects to be turning out 65-nanometer Pentiums and Xeons from four factories.
The transition to 65 nanometer will serve other components inside PCs as well.
The transition will eventually free up capacity for building things like chip sets, chips that assist a computers processors, in 90-nanometer plants that are now dedicated to building processors.
For Intel, moving chip sets to 90 nanometers also offers a chance to load them with more features, such as expanded graphics capabilities.
AMD, Intels chief rival in x86 chips, has also been executing a 65-nanometer plan.
AMD, which has worked with IBM on 65-nanometer manufacturing technology, has begun pilot production at its recently opened Fab 36 in Dresden, Germany. AMD expects to begin full 65-nanometer production in the latter half of 2006, Kepler said.
“Our progress on 65-nanometer technology is going very well,” Kepler said. “Weve gotten very good results on the technology at this stage.”
IBM of Armonk, N.Y., said it also plans to convert its chip plant in East Fishkill, N.Y., to 65-nanometer production from 90-nanometer production over time.
It has already begun prototyping its 65-nanometer process while it moves equipment into a special annex, also designed to produce chips at 65 nanometers and then later at 45 nanometers.
IBM has yet to say exactly when it will start its 65-nanometer chip production. However, a company executive told Ziff Davis Internet earlier this year that its aiming for 2006.
Despite the cost, the companies stand to benefit greatly from their technology transitions, analysts say.
“Intels 65-nanometer products will hit en masse in 2006. That alters the cost and margin equations for Intel,” said Dean McCarron, analyst with Mercury Research. “Its combining a move to [larger] 300-mm wafers and 65 nanometers in some factories, so its a double whammy. Yields in terms of functional devices and speed bins [otherwise known as chip clock speeds] will increase.”
Thus, by moving to 65-nanometers sooner, Intel could possibly make gains against rival AMD, which gained market share during the third quarter due to its strong product line, McCarron said.
However, AMD has similar designs.
“Were making the same product and get twice as many on a wafer. That means the cost of each of those products is lower. We can move costs down,” Kepler said. “The other thing that we can do—the other side of that—is we can put twice as much on a chip that was the same size at 90 nanometers. You can just fit more in it.”
Check out eWEEK.coms for the latest news in desktop and notebook computing.