Intel Has Leading Edge Multicore Processors In Plan For 2008

Intel is looking forward to bringing leading edge microprocessors and technologies to the market, as it unveiled its plans for the second half of this year. The company expects a six-core processor coming up later in 2008. In addition to “Dunnington,” the codename for the six-core processor, Intel will also bring its new Itanium processor named “Tukwila.”

“Dunnington” for expandable multi-processor servers that is said to become available in the second half of 2008 is based on the 45nm high-k process technology, offers compatibility with systems designed from quad-core Xeon processors and it supports FlexMigration technology, that allows a single compatible virtualization pool to support migration across servers. In addition, it will feature 1.9 billion transistors and 16MB of L3 cache.

Intel’s “Tukwila,” considered to be the next generation Itanium processor, has four cores, 30 MB cache, QuickPath Interconnect, dual Integrated Memory Controller and mainframe-class RAS features, the company’s Senior Vice President and General Manager Pat Gelsinger unveiled.

In addition to all that, Intel’s “Nehalem” will have 2 to 8 cores and will be capable of delivering four times the memory bandwidth compared to the Intel Xeon systems. Nehalem will be equipped with 731 million transistors, up to 8 MB L3 cache, QuickPath Interconnect which will offer up to 25.6 GB per second and integrated memory controller.

Intel AVX (Advanced Vector Extensions) were also part of the discussion, as Gelsinger said they will increase performance in floating point, media and processor intensive software when used by software programmer, and at the same time, it will increase energy efficiency and will be compatible with existing Intel processors.

“These technical improvements will result in performance improvements as well as flexibility for a wide range of eventual products based on the Nehalem architecture,” Gelsinger discussed in his presentation, but also noted: “Attempts to create new programmable architectures are painful heavy-lifting over time, and for the most part they fail.”