Intel’s Cluster Tools Ease Programming And Optimization November 09, 2006
Targeting software developers, Cluster Toolkit 3.0 and Cluster OpenMP promise to ease the creation, debugging, and optimization of parallel applications for server clusters while facilitating the adoption of multi-core processor server clusters. The Cluster Toolkit combines the latest versions of MPI Library, Math Kernel Library Cluster Edition, and Trace Analyzer and Collector. The Cluster OpenMP for the company's compilers is a new application that extends OpenMP to distributed memory clusters, making OpenMP a programming method that works well for dual-/quad-core processors and clusters. The MPI Library now includes automated fabric selection, an array of performance optimizations, and full thread support level. Math Kernel Library Cluster Edition 9.0 now includes optimizations for multi-core Xeon 5100 and 5300 series processors, more VML functions (floor, ceil, round, trunc, and hypot), a FMGRES iterative sparse solver, and a FFTW interface in Fortran and C. Assuming title as the best analysis tool available for MPI applications, Trace Analyzer and Collector 7.0 adds comparison of multiple trace files, timeline display of performance counters, unique aggregation and filtering functions, a faster GUI, and an MPI-checking library. The Cluster OpenMP application for Intel C++ and Fortran Compilers debuts as the first commercially available OpenMP for clusters. The software features ease of use, the ability to run OpenMP code on a commodity cluster and to exploit existing SMP OpenMP codes on cheaper clusters, and performance comparable to SMP machine with the same number of CPUs. Price for the Cluster Toolkit 3.0 is $749. MPI Library, Math Kernel Library Cluster Edition, and Trace Analyzer and Collector are priced individually at $499 each. INTEL CORP., Santa Clara, CA. (800) 538-3373.
Electronic Design China 
