Developing for the Microsoft Windows HPC Platform
< Go Back
This white paper reviews the tools and technologies that make the Windows HPC Server 2008 high-performance computing solution a highly productive development platform for parallel computing. Modern computing power allows us to solve complex problems that might otherwise be insurmountable. Developing applications that take advantage of scaled-out computing resources presents unique challenges. Microsoft established the Parallel Computing Initiative (PCI)—which includes programming models, languages, tools, platforms, and the Windows operating system—to increase innovation in parallelism.
Windows HPC Server 2008 (HPCS) presents a set of programming models that simplify previously daunting computing problems and thus creates new business opportunities. The main programming models available to HPC developers on the Windows HPC platform are task-based parametric sweep, message passing via the Microsoft Message Passing Interface (MPI) library, and cluster-based service oriented architectures or Cluster SOA. This white paper describes each model, including recommended uses and programming samples.
HPCS forms the core of a growing ecosystem of Microsoft and partner tools and technologies that make Windows a powerful and productive distributed computing platform. These tools and technologies can help you to integrate Windows cluster computing into your organization and your applications in a way that suits your situation.
HPCS job submission can be integrated into a business process from Microsoft SharePoint Products and Technologies, Microsoft Office Excel, custom scripts, scientific computing environments, cross-platform Web service clients, or applications built on the Microsoft .NET Framework. Jobs can be submitted to HPCS directly from the Mathematica and MATLAB environments, and many open-source libraries have also been ported to the Windows HPC Server 2008 platform. In addition to job submission and monitoring tools, the number of Microsoft and partner tools for distributed application quality, profiling, and debugging has increased.
