The new chips, codenamed Penryn, which come in desktop and server versions, promise higher clock speeds and yet the same power envelope as their larger 65-nm predecessors.

The goal, of course, is to keep pace with Moore’s Law, the observation by Intel co-founder Gordon Moore that the complexity of chips doubles about every 18 months.

The upshot to businesses should be faster and more robust applications. While the majority of the new chips – 15 dual- and quad-core Xeons – were aimed at servers (see separate story), the first desktop Penryn out the gate promises better performance for video, teleconferencing and workstation applications.

The desktop version, a dual-core called Core 2 Extreme QX9650, is aimed at gamers and prosumers. Ir boast a larger L2 cache and support for Intel SSE4, which is the company’s 47-set of media instructions the make video encoding faster for photo editing and high-definition media. Adobe and Microsoft are among the large software makers that have previously supported SSE4.

Intel claims Penryn will deliver more than a 20% increase in performance, with improved energy efficiency. The company said 40 OEMs were ready to produce new machines powered by Penryn, including workstations from Hewlett-Packard and Lenovo.

An additional 20 or so Penryn processors will ship in the first quarter of 2008, Intel said.

Penryn is about 25% smaller than previous microprocessor versions and, therefore, more cost-effective, Intel said. But the company has spent significant R&D, and manufacturing dollars on ensuring Penryn saw the light of day. It spent $3bn to build a new manufacturing fab in Arizona, which began 45-nm production late last month. Other 45-nm fabs in Israel and New Mexico are slated to be online soon.

In addition to better server and desktop processors, the 45-nm manufacturing node will also enable Intel next year to release smaller mobile chips, it said. Among them will likely be for so-called ultra mobile machines, which are basically fully loaded handheld PCs, as well as systems-on-chips.

Intel was able to shrink its manufacturing technology to 45-nm largely thanks to a new microchip material that promises to stem energy leakage, a problem that has plagued the semiconductor industry as chip geometries shrank.

Previously, silicon dioxide was used as an insulator within transistors, specifically within the device’s gate, which acts like the on and off switch for the device, as well as the gate dielectric, an insulating layer that helps contain the current. With each smaller microprocessor generation, the silicon dioxide had to become thinner, but it was beginning to be unable to contain electronic current. As a result, the current seeped out and caused excess heat and energy loss.

Penryn is the first microprocessor to use a new type of hafnium-based material to replace silicon dioxide. The new material has a property called high-k for gates and gate dialectrics, which would reduce leakage by more than 10 times compared to silicon dioxide, according to the company.

The high-k gate dielectric is not compatible with existing silicon gate electrode, so Intel has developed a new mix of metals for gate materials. The company said the specifics of this metal mix would remain secret for competitor reasons.

AMD is not expected to ship 45-nm microprocessors until next year. The company is in partnership with IBM, Sony and Toshiba on the IBM-led development of separate high-k metal gate technology. IBM had previously claimed that its high-k component was created without requiring major manufacturing tooling or process changes — unlike Intel’s. IBM said it had inserted high-k manufacturing into its manufacturing line at its East Fishkill, New York fab and will begin using it in 45-nm chips next year.

High-k will have little effect on existing roadmaps, but it will enable the chipmakers to continue Moore’s Law well into the next decade.

Intel has already demonstrated its proof-of-concept next-generation 32-nm chips.