“We’re going to build computers 100 times more capable than they are today… moving to exaflop computing,” says Kirk Skaugen, Vice President of Intel’s architecture group.
An exaflop is equivalent to a thousand petaflops, and a petaflop equals 1,000 trillion floating point operations per second. Intel’s exascale prediction is in line with its upcoming 22nm Tri-Gate technology — 3-D based transistors that rise up vertically from a wafer’s surface — continued size reductions of chips and its MIC architecture. The envisioned supercomputers will enable such things as real-time delivery of CT scans to an operating room and better forecasting of hurricanes, Skaugen says.
At the annual International Supercomputing Conference in Hamburg, which names the fastest computer in the world based on the benchmark application LINPACK, Fujitsu’s K supercomputer, was named the fastest computer in the world for its 8-petaflops-per second performance.
However Cray’s new XK6 supercomputer, capable of 50 petaflops per second, wasn’t entered in the contest because of ongoing software development projects. The race goes on.
Source: The Register