Parallel Computing Toolkit Provides Inexpensive Computing
Solution with High
February 7, 2000--With the release of Parallel Computing Toolkit, Wolfram
Research officially introduces parallel computing support for Mathematica.
Parallel Computing Toolkit for Mathematica makes parallel programming easily
affordable to users with access to either a multiprocessor machine or a
network of heterogeneous machines--without requiring dedicated parallel
hardware. Parallel Computing Toolkit can take advantage of existing
Mathematica kernels on all supported operating systems--including Unix,
Linux, Windows, and Macintosh--connected through TCP/IP, thus enabling users
to use existing hardware and
Mathematica licenses to create low-cost "virtual parallel computers."
Parallel Computing Toolkit 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 of Parallel Computing Toolkit include
machine-independent implementation, parallel functional programming, and
failure recovery and automatic reassignment of stranded processes in the
event of a system failure.
Parallel Computing Toolkit implements many parallel programming primitives
and includes high-level commands for parallel execution of operations such
as animation, plotting, and matrix manipulation. It comes with numerous
examples demonstrating many popular new programming approaches such as
parallel Monte Carlo simulation, visualization, searching, and optimization.
Because Parallel Computing Toolkit also provides the Mathematica source code
for all high-level commands, these operations and examples can serve as
templates for building additional parallel programs.
Users can benefit by employing Parallel Computing Toolkit in a number of
ways. The easiest, and often surprisingly effective, method is simply for users to
wrap a command such as ParallelEvaluate around their code. Parallel
Computing Toolkit then assigns individual calculations automatically to free
processes across the network. With slightly more effort, advanced users of
Parallel Computing Toolkit can optimize performance even more by using
explicit scheduling of their calculations. Like all parallel computing
environments, Parallel Computing Toolkit shows the best performance
improvements for inherently parallel computations like many list and matrix
operations or for repetitive operations like Monte Carlo simulations.
Engineers, scientists, and analysts will find Parallel Computing Toolkit to
be an ideal tool for performing the large-scale computations often involved
in the product-design and problem-solving process. Says Roman Maeder,
creator of Parallel Computing Toolkit, "One of my key motivations for
writing this package was to finally make serious parallel computing truly
accessible to a wide range of workgroups, labs, and classrooms." Parallel
Computing Toolkit is written entirely in the platform-independent
Mathematica programming language, giving users access to all of Mathematica's
legendary numeric, symbolic, and graphic parallel-computation capabilities.
For more on Parallel Computing Toolkit, visit