Since can not build any fast processors, are all high-performance parallel computers. A parallel computer is a computer that will be distributed simultaneously in the operations on multiple CPUs to increase the operating speed. Be tuned for optimal utilization of a supercomputer, the programming so as closely as possible on the individual parallel processors.
Supercomputers are now generally conceived as a vector or Skalarrechner calculator. They are based on different processor architectures. Vector processors (or vector computers, array processors called) perform a calculation simultaneously on many data (in a vector or array) from. Skalarprozessoren other hand, can process only a Operandenpaar per command. Skalarrechner therefore often based on thousands of standard processors that are networked together) (computer cluster.
Originally, the outstanding performance by maximum utilization of available technology was achieved by designs were chosen that were too expensive for large volume production (eg, liquid cooling, exotic materials and components, compact design for short signal paths), the number of processors was rather low. For a long time to establish more so-called clusters that are interconnected in which a large number of (mostly cheap) individual computers to a large computer. Compared to a vector computer nodes in a cluster have their own peripherals and only its own local memory. Clusters use standard components, so they initially offer cost advantages over vector computers. But they require a far higher programming costs. It is examined whether the established programs lend themselves to being distributed across many processors. Used programming languages include Fortran and C.
Closely associated with the term, the Cray supercomputer. It is named after its founder, Seymour Cray, and made the first supercomputer in the 1970s. The first Cray-1 supercomputer installed officially abolished 1976 130 mega-flops. (For comparison, a normal PC can now run multiple GigaFLOPS.)
No comments:
Post a Comment