Personal Supercomputing Made Easier

By DE Editors

The MathWorks (Natick, MA) announced the availability of version 2 ofits high-productivity programming tool, the Distributed ComputingToolbox. The Distributed Computing Toolbox 2 makes it convenient forengineers, researchers, and scientists to develop and run distributedapplications using MATLAB, the company’s flag technical computingsystem. In addition to making personal supercomputing power readilyavailable at your desktop, the Distributed Computing Toolbox 2 alsoserves as a high-level technical language for supercomputingspecialists.

The Distributed Computing Toolboxworks with the MATLAB Distributed Computing Engine, available separately, to give you the ability to execute MATLAB algorithms and Simulinkmodels in a computer cluster. You prototype and developapplications in the MATLAB environment, and you then use the DistributedComputing Toolbox to divide the applications into independent tasks.

The combination of attributes and functionality within theDistributed Computing Toolbox2 simplifies the development of distributed computing applications,which, in turn, enables domain experts to take on more formidablecomputing challenges than was previously practical. Its newinterprocess communication capabilities facilitate the distribution andexecution of parallel algorithms in a compute cluster. Further, theDistributed Computing Toolbox now supports third-party schedulers, suchas LSF fromPlatform Computing.

Third-party scheduler support means that you can use the generic APIprovided with the Distributed Computing Toolbox as well as integrateMathWorks’s distributed computing tools into your existing distributedcomputing environments. This capability lets you leverage a scheduler’sunique capabilities, such as support for batch jobs. It also means thatyou can continue to take advantage of the interactive workflowssupported in version 1 by The MathWorks job manager in the MATLABDistributed Computing Engine.

The Distributed Computing Toolbox 2’s interprocess communicationenhancements lets you execute parallel applications that are dividedinto interdependent tasks. Version 2’s communication functions arebased on MPI (message passing interface), the industry-wide protocolfor communication in a parallel program. Because Distributed ComputingToolbox 2 runs on any hardware that MATLAB runs on, you can redeployyour parallel applications on new hardware or operating systems withoutretooling the entire application.

The Distributed Computing Toolbox supports the full MATLAB language,most every MathWorks’s product, and all supported MATLAB platforms. TheMATLAB Distributed Computing Engine can run in either homogeneous orheterogeneous clusters.  Both the Distributed Computing Toolboxand MATLAB Distributed Computing Engine are available immediately.Pricing remains unchanged from earlier versions: $1000 US for theDistributed Computing Toolbox 2 and $6,000 Distributed Computing Engine.

Forcomplete details, including data sheets and quick lessons on how youcan develop your own distributed MATLAB applications, go to theDistributed Computing Toolbox information page by clicking here.