Intel packs teraflop capacity on a chip

On its road to boost tera-scale computing capacity for future PCs and servers, Intel Corp. researchers have developed the 'first' programmable processor that offers teraflop performance on a single 80-core chip while consuming only 62W-less than many single-chip processors today. Intel has no plans to bring the chip to market but is optimistic that this research chip will pave for specific insights in new silicon design methodologies, high-bandwidth interconnects and energy management approaches.

"Our researchers have achieved a wonderful and key milestone in terms of being able to drive multicore and parallel computing performance forward," said Justin Rattner, Intel senior fellow and chief technology officer. "It points the way to the near future when teraflops-capable designs will be commonplace and reshape what we can all expect from our computers and the Internet at home and in the office."

The 80-core chip achieves the same teraflop performance realized a decade ago on the ASCI Red Supercomputer built by Intel for the Sandia National Laboratory. In the new multi-core chip, Intel squeezed the teraflop performance of the supercomputer which then took up over 2,000ft?, was powered by about 10,000 Pentium Pro processors and consumed over 500kW.

The chip offers a tile design in which smaller cores are replicated as 'tiles,' allowing for easier design of single-chip multi-core processors. The research chip also features a mesh-like 'network-on-a-chip' architecture to allow ultrahigh bandwidth communications between cores and to process Terabits of data per second inside the chip.

In achieving further the goals of its Tera-scale Computing Research Program, Intel will focus on the addition of 3D stacked memory to the chip as well as developing more advanced prototypes with many general-purpose Intel architecture-based cores.