Wolfram Research has released gridMathematic, a parallel computing solution for dedicated grids or clusters. It can run on any cluster of machines including Mac OS X, Unix, Linux and Windows via TCP/IP connectivity.
Wolfram’s Mathematica allows complex calculations to be carried out on single-processor machines, but gridMathematica can tackle much larger computations using clusters, multiprocessor machines and computing grids. Typical uses include bioinformatics applications, processing and analysis of large data sets, data mining and large computations in physics, mathematics, and the life sciences.
“One of my main motivations for writing this package was to make parallel computing easily available to a wide range of workgroups,” said Roman Maeder, creator of Parallel Computing Toolkit (a component of gridMathematica).
gridMathematica supports all common parallel programming paradigms, such as virtual shared or distributed memory, automatic or explicit scheduling, and concurrency, including synchronization, locking and latency hiding. Other features include machine-independent implementation and parallel functional programming, as well as failure recovery and automatic reassignment of stranded processes case of a system failure. And the benefits of gridMathematica go beyond pure number crunching, according to Theodore Gray, director of User Interfaces at Wolfram Research.
“For Stephen Wolfram’s keynote at Comdex 2002 [this week’s trade show in Las Vegas, Nevada], we were able to create intricate animations of cellular automata in mere hours,” he said. “After writing and testing code on an iMac, we moved it over to a 16-way Xserve cluster running Mac OS X Server, thereby cutting calculation time from days to less than an hour. Performance gains like this open up new opportunities for Macintosh users in science, engineering, and the arts.”
Wolfram’s Web site doesn’t list the pricing for gridMathematica. If you’re interested in buying it, you’ll have to fill out an online information request form.