ACADEMIA
New Star-P 2.6 Boosts Application Performance
Interactive Supercomputing, Inc. (ISC) today announced a new version of its Star-P software that improves parallel application performance and expands support for programming languages, operating system and workload management platforms. Star-P 2.6 is a parallel application development platform that lets scientists, engineers and analysts create algorithms and models on their desktops using familiar mathematical tools -- such as MATLAB, Python and R -- and then run them instantly and interactively on parallel computers with little to no modification. Star-P eliminates the need to re-program applications in C, Fortran or MPI in order to run on parallel systems, resulting in huge productivity gains. 
What's new?
-- New code optimization tools for faster app performance
-- Expanded Python support
-- New parallel platform ease-of-use tools
-- R language "preview" client
-- More workload manager support
-- Windows Vista support
Expanding platform support
Star-P 2.6 features a "preview" version of an R language client, expanding its support for Very High Level Languages (VHLL) beyond MATLAB and Python. R is an open source VHLL for statistical computing and graphics that is favored by the life sciences and financial analysis industries. In keeping with ISC's support for open source initiatives, the "preview" R version is designed to ignite further development and refinement working closely with early adopter customers and the open source community.
Star-P 2.6 also extends its support for Python, adding more than 70 new parallel computing functions to its collection of Python mathematical libraries. The new version also adds support for parallel file I/O in Python, as well as the ability to do task parallel computing with SciPy, an open source library of Python programming extensions that support large, multi-dimensional arrays and matrices.
Star-P 2.6 also expands its support for popular cluster workload managers, now integrating with Sun Microsystems's Sun Grid Engine as well as PBSpro. Workload management support enables Star-P to run on a broader set of enterprise clusters in the market. The new version also includes a Star-P desktop client for the Microsoft Vista operating system, in addition to its Linux and Windows XP clients.
Researchers at the Whitehead MIT BioImaging Center, part of MIT's Computational and Systems Biology Initiative (CSBi), are taking advantage of Star-P's expanded platform support to create new biological models that may someday yield new drug discoveries. "Our project includes multi-disciplinary teams of biologists, engineers and computer scientists with different skill levels using a variety of tools, languages and computing platforms to create the best possible computational models in the shortest time," said James G. Evans, assistant director of the Whitehead MIT BioImaging Center. "Star-P's ever expanding platform support increases its versatility and value for our work."
Performance optimization
To boost performance, Star-P 2.6 now has the ability to dynamically load mathematical libraries that are optimized for a specific computing platform. Previous versions were not hardware specific, requiring further fine-tuning of the libraries to optimize performance. The new version automatically recognizes and loads mathematical libraries pre-optimized for a specific hardware platform, resulting in performance gains of 20 to 100 percent.
Star-P 2.6 also features a number of new performance profiling tools for optimizing application performance on parallel systems. A new PPBench tool allows users to diagnose performance problems stemming from hardware and network issues. PPBench is a system-level profiler that lets users see real-time graphical analysis of how their programs execute across processors, memory, cache and network interconnects, allowing them to fine-tune their code structure for better performance. The new version also features a "Dashboard" and "Heartbeat" indicator that provides real-time server, scheduling and computational status of an application running on a parallel system.
"Our goal is to continuously open up and extend Star-P to support all the languages, platforms, tools, libraries and infrastructure important to the technical computing community," said Bill Blake, ISC's CEO. "The new enhancements in Star-P 2.6 will enable customers to tackle much larger problems on their desktops, and solve their problems many times faster using a wider array of programming and computing resources."
What's new?
-- New code optimization tools for faster app performance
-- Expanded Python support
-- New parallel platform ease-of-use tools
-- R language "preview" client
-- More workload manager support
-- Windows Vista support
Expanding platform support
Star-P 2.6 features a "preview" version of an R language client, expanding its support for Very High Level Languages (VHLL) beyond MATLAB and Python. R is an open source VHLL for statistical computing and graphics that is favored by the life sciences and financial analysis industries. In keeping with ISC's support for open source initiatives, the "preview" R version is designed to ignite further development and refinement working closely with early adopter customers and the open source community.
Star-P 2.6 also extends its support for Python, adding more than 70 new parallel computing functions to its collection of Python mathematical libraries. The new version also adds support for parallel file I/O in Python, as well as the ability to do task parallel computing with SciPy, an open source library of Python programming extensions that support large, multi-dimensional arrays and matrices.
Star-P 2.6 also expands its support for popular cluster workload managers, now integrating with Sun Microsystems's Sun Grid Engine as well as PBSpro. Workload management support enables Star-P to run on a broader set of enterprise clusters in the market. The new version also includes a Star-P desktop client for the Microsoft Vista operating system, in addition to its Linux and Windows XP clients.
Researchers at the Whitehead MIT BioImaging Center, part of MIT's Computational and Systems Biology Initiative (CSBi), are taking advantage of Star-P's expanded platform support to create new biological models that may someday yield new drug discoveries. "Our project includes multi-disciplinary teams of biologists, engineers and computer scientists with different skill levels using a variety of tools, languages and computing platforms to create the best possible computational models in the shortest time," said James G. Evans, assistant director of the Whitehead MIT BioImaging Center. "Star-P's ever expanding platform support increases its versatility and value for our work."
Performance optimization
To boost performance, Star-P 2.6 now has the ability to dynamically load mathematical libraries that are optimized for a specific computing platform. Previous versions were not hardware specific, requiring further fine-tuning of the libraries to optimize performance. The new version automatically recognizes and loads mathematical libraries pre-optimized for a specific hardware platform, resulting in performance gains of 20 to 100 percent.
Star-P 2.6 also features a number of new performance profiling tools for optimizing application performance on parallel systems. A new PPBench tool allows users to diagnose performance problems stemming from hardware and network issues. PPBench is a system-level profiler that lets users see real-time graphical analysis of how their programs execute across processors, memory, cache and network interconnects, allowing them to fine-tune their code structure for better performance. The new version also features a "Dashboard" and "Heartbeat" indicator that provides real-time server, scheduling and computational status of an application running on a parallel system.
"Our goal is to continuously open up and extend Star-P to support all the languages, platforms, tools, libraries and infrastructure important to the technical computing community," said Bill Blake, ISC's CEO. "The new enhancements in Star-P 2.6 will enable customers to tackle much larger problems on their desktops, and solve their problems many times faster using a wider array of programming and computing resources." 