Intel has announced that it has developed an accelerator chip capable of running at a speed of one teraflop, which is equal to one trillion calculations per second. Dubbed Knights Corner, the capability of the new chip was demonstrated using a test machine at the SC11 supercomputing conference in Seattle. The manufacturer explained that the Knights Corner chip acts as a co-processor, assuming responsibility for some of the most complicated tasks from the computer's central processing unit. To achieve this, Intel installed more than 50 cores onto a single piece of silicon.
According to Intel, the new accelerator is also the first - and thus far only - server processor to support full integration of the PCI Express 3.0 specification. Through the new technology data can be transferred at up to 32 gigabytes per second to compatible devices. This, Intel explained, is twice the speed of the previous generation.
"Collecting, analyzing and sharing large amounts of information is critical to today's science activities and requires new levels of processor performance and technologies designed precisely for this purpose," Rajeeb Hazra, Intel's general manager of technical computing, explained.
"Having this performance now in a single chip... is a milestone that will once again be etched into HPC [high performance computing] history," Mr Hazra added.
Intel's co-processor is reliant on the same instruction set architecture as x86 processors, another of the firm's products. Its rivals, by contrast, are taking a different approach, offering chips known as graphic processing units (GPUs). These are specifically designed to carry out the calculations necessary to draw, color and shade objects on the screen - and at very high speed.
But in contrast to GPUs - which require additional software coding - Intel's accelerator is said to be able to run existing applications at high speed without the need for software development.
Martin Reynolds, a vice-president at the research firm Gartner, explained: "Traditional supercomputers were built by putting thousands of processors in a room but in the last few years there has been a shift toward graphic processors."
He observed that GPUs "allow you to get results more quickly but will take longer to program so there is an interesting trade-off".
Intel, Mr Reynolds added, has a technology advantage over rival firms "because its manufacturing processes can make transistors half the size (of competing product) and more efficient". However, he observed that rival manufacturers will eventual catch up with Intel and challenge its market dominance.