Intel (NASDAQ: INTC) this week unveiled the Knights Corner co-processor that it says is capable of delivering one teraflop of floating point performance. Part of a new architecture called MIC (many integrated cores), the Knights Corner chip plugs into a PCI Express slot as an accelerator. This is similar to how graphics processors work, except that Knights Corner's works better with existing x86 code.

Underscoring the huge advancements made by microprocessors, Rajeeb Hazra, general manager of Intel's Technical Computing Group observed how Knights Corner's teraflop performance is equivalent to that of the ASCI Red supercomputer with 9,298 Pentium II Xeon chips in 1997. No release date for Knights Corner was offered, though Hazra was quoted as saying: "We don't build chips...and then hold on to them for years."

As reported on PC Mag, GPUs made by companies such as Nvidia and AMD currently provide better double-precision floating point performance than any x86-based CPU. The articled noted: "Intel's Core i7 980 XE has reached 109 gigaflops for double precision calculations, but Nvidia's GTX 480 has hit 672 gigaflops and paired HemlockXT 5970 GPUs from AMD peak at 928 gigaflops." Clearly, Knights Corner will allow Intel to leapfrog its competitors in terms of sheer floating point prowess.

More importantly, the Intel MIC architecture will run existing applications without the time-consuming process of porting programming code to a new instruction set. In a statement, Intel noted that "this will allow scientists to use both CPU and co-processor performance simultaneously with existing x86 based applications, dramatically saving time, cost, and resources that would otherwise be needed to rewrite them to alternative proprietary languages."

For more:
- check out this article at PC Mag
- check out this article at Computerworld

Related Articles:
Ultrabook shipments projected to hit 43% of all laptops by 2015
Intel's IPT gains new fan in MasterCard
Intel working on tablet-specific processors for 2012 launch