Intel's 65 nm Process Breathes Fire into Double-Core Extreme Edition

The 65 nm Era Begins

Almost three months after our preview of Intel's 65 nm Presler dual-core processor, today marks the device's availability in retail channels. The processor that marks the entry point into a new manufacturing process is nothing less than an Extreme Edition, and for the first time in its history it is a seriously promising package.

Intel has been receiving scathing criticism for its current desktop processor family, because all Pentium 4 and Pentium D processors consume clearly more energy than their AMD counterparts, while at the same time delivering less performance on average. The stalling flight was initiated in early 2004 by the 90 nm production process, because has been plagued by considerable leakage currents. These cause high levels of heat dissipation, which has prevented Intel from ramping up the chip's clock speeds as intended; and high clock speeds are the prerequisite for the controversial NetBurst processor architecture to deliver.

We have performed two early technology previews of Intel's upcoming 65 nm NetBurst products. The first one was a look at the single- core Pentium 4 6x1 series, code-named Cedar Mill; the second was our take on the double-core Presler (Pentium D 900 series). Cedar Mill will replace the current Pentium 4, while Presler is the replacement for the Pentium D 800 and Pentium Extreme Edition 840 series. We deliberately call Presler a double core rather than a dual core device, because it comprises two Cedar Mill single cores into one processor package. The single core does not have significantly different performance compared to its 90 nm Prescott 2 MB counterpart, but power consumption of both single and double core solutions could be reduced significantly.

Hence, our expectations for the all-new 65 nm Pentium Extreme Edition 955 based on Presler are rather high: It is based on a state-of-the-art manufacturing process, comes with double the L2 cache (2x 2 MB rather than 1x 1 MB), an accelerated system clock speed (FSB1066 instead of FSB800) and a core clock speed increase of approximately 8.25%. Will this finally be enough to pose a threat to AMD's current performance superiority?