Oct. 23, 2008 | This week The MathWorks announced that its parallel computing products – Parallel Computing Toolbox and MATLAB Distributed Computing Server – now run EGEE grid (Enabling Grids for E-sciencE) as part of the latest update to both products. These tools are now fully integrated with gLite, the middleware used by EGEE, which is Europe’s largest high-performance computing consortium. Prior to this integration, users could not work with MATLAB in a grid environment. Now, EGEE users can access the power of the grid directly from their desktop.
Grid computing has grown rapidly worldwide. According to analyst firm IDC, clusters comprised 65 percent of computing by the end of 2007 up from 20 percent in 2003. There has also growing interest in cloud computing. MathWorks, like many in the tool community, has embraced these trends by making it easier for user to develop applications used on multiple-core machines or multiple machines.
For example, the EGEE – co-funded by the European Commission – is a grid infrastructure with over 70,000 CPUs and access to more than 250 high-performance computing (HPC) centers. EGEE provides users, irrespective of their location, with the large-scale computing power necessary to run their applications.
“The work we’re doing with EGEE is the most recent step in our mission to provide tools and services that support parallel computing for the scientific and engineering community,” said Silvina Grad-Freilich, manager of parallel computing and application deployment marketing at The MathWorks.
Key features in the updated Parallel Computing Toolbox 4.0 include:
- Support for data-parallel and task-parallel application development
- Ability to annotate code segments with parfor (parallel for-loops) and SPMD (single program multiple data) for implementing task- and data-parallel algorithms
- High-level constructs such as distributed arrays, parallel algorithms, and message-passing functions for processing large data sets on multiple processors
- Ability to run four workers locally on a multicore desktop
- Integration with MATLAB Distributed Computing Server for cluster-based applications that use any scheduler or any number of workers
Up to four users can use the toolbox on a multiprocessor computer using a single toolbox license. Scaling beyond that requires licensing MATLAB Distributed Computing Server. In addition, MATLAB applications that use Parallel Computing Toolbox functions can be built into standalone executables and shared software components using MATLAB Compiler, for royalty-free distribution. These executables and libraries can connect to MATLAB Distributed Computing Server workers and perform MATLAB computations on a computer cluster.
MATLAB Distributed Computing Server supports several schedulers: the MathWorks job manager (provided with the product) or any other third-party scheduler such as Platform LSF, Microsoft Windows Compute Cluster Server, Altair PBS Pro, and TORQUE.
----------------------------------
This article first appeared in Bio-IT World’s Predictive Biomedicine newsletter. Click here for a free subscription.