SOURCE: Interactive Supercomputing

November 20, 2007 12:46 ET

Interactive Supercomputing Wins HPC Challenge Awards Competition

Star-P for Python Awarded Most Productive Commercial Implementation Honors

WALTHAM, MA--(Marketwire - November 20, 2007) - Interactive Supercomputing Inc. (ISC) has won the annual High Performance Computing (HPC) Challenge Benchmarks Award Competition at this year's SC07 conference. It is the second honor bestowed upon the company at SC07, having also won the HPCwire Editors' Choice Award for Best Price/Performance High Performance Computing (HPC) Software.

Sponsored by The DARPA High Productivity Computing Systems (HPCS) Program and the IDC analyst group, and judged by a committee led by Jack Dongarra (University of Tennessee and Oak Ridge National Lab) and Jeremy Kepner (MIT Lincoln Lab), the competition's goal is to focus the HPC community's attention on developing HPC hardware and software capabilities that boost the productivity of using HPC systems.

ISC won the competition's Most Productive Commercial Implementation category for the ability of its Star-P® software to dramatically accelerate the creation of Python versions of the HPC Challenge benchmarks. The benchmarks -- which included an FFT, a linear solve, a matrix multiply, and a random access problem -- were written in Python and Star-P, and run on a 128-processor at the San Diego Supercomputer Center. According to the competition's rules, winners are judged based on the most "elegant" implementation of four or more of the HPC Challenge benchmarks. The award is weighted 50 percent on computational performance and 50 percent on code elegance, clarity and size. The ISC submission required just four lines of different source code to parallelize three of the benchmark kernels across a 128-core cluster, with the performance of the HPL benchmark achieving 57 percent of peak. This combination of simplicity and performance is unique in the HPC landscape.

"Winning this award underscores how user productivity has become as important to HPC as raw processor performance," said Sudarshan Raghunathan, member of the technical staff at ISC. "Python has recently emerged as a powerful and flexible language for rapidly developing numerical applications, and Star-P extends its benefits into the high performance computing domain. We are very excited about the HPC Challenge award, as it illustrates that Python and the Star-P programming model can deliver a strong balance of performance and productivity for computationally intensive applications."

Star-P is a parallel application development platform that enables users to code algorithms and models on their desktops using familiar mathematical software packages such as MATLAB, Python and R and run them instantly and interactively on large scale multi-core clusters. Star-P eliminates the need to re-program applications in C, C++, Fortran or MPI in order to run on high performance computing servers. Reprogramming algorithms created in VHLLs to parallel versions without Star-P can take months to complete for large, complex problems, so Star-P yields dramatic improvements in parallel programming productivity while significantly cutting the labor costs of intensive research projects.

About Interactive Supercomputing

Interactive Supercomputing (ISC) launched in 2004 to commercialize Star-P, an interactive parallel computing platform. With automatic parallelization and interactive execution of existing desktop simulation applications, Star-P merges two previously distinct environments -- desktop computers and high performance servers -- into one. Based in Waltham, Mass., the privately held company markets Star-P for a range of biomedical, financial, and government laboratory research applications.

Contact Information

  • Contacts:
    Ilya Mirman
    Interactive Supercomputing

    Michelle Dillon
    Beaupre & Co. Public Relations
    Email Contact