Low-cost, Reliable Software Uses COTS, Runs on Multiple Operating Systems: enParallel Inc. will introduce their ePX software, a low-cost solution for accelerating applications utilizing Graphics Processor Unit (GPU) technology at SC08, the supercomputing conference in Austin. Using readily available commercial off-the-shelf (COTS) hardware, enParallel Desktop SuperComputing (DSC) technology implements a true supercomputing parallel processing model on GPU Arrays (GPA). In booth 3012, enParallel will demonstrate the ePX software on both a laptop (Mac OS) and a Windows-based desktop. enParallel’s proprietary software enables the use of scatter-gather, CPU/GPU process pipelining and advanced GPA resource scheduling, effectively turning a standard laptop or desktop computer into a full-fledged supercomputer, the company said. “Depending on the specific hardware configuration and set of algorithms, scientific applications can run over 1000 times faster.” says Curtis Kent, President and CEO of enParallel, Inc. “It’s important to note this benefit is extended to complete applications and not just specific algorithms.” ePX software will run on any laptop or desktop with the supporting GPA hardware configuration and with any standard operating system like Windows XP, Vista, Linux, and Mac OS. In standard configuration, the software will drive arrays of 2-4 GPUs, and can be scaled upwards for custom hardware solutions. The company is currently engaged in performance testing on other operating systems, including Unix. “enParallel’s desktop supercomputing technology effectively creates a new class of supercomputers part and parcel of an overall evolutionary trend for PCs.” says Dr. Glenn-Anderson, CTO of enParallel, Inc. “At first, PCs supplanted workstations in a host of high performance applications. Now, with the advent of this exciting new GPU technology, we’ve created a desktop PC that can rival the performance of distributed and cluster supercomputing systems and at far lower cost.” enParallel’s proprietary software technology employs reusable code frameworks, making it faster for software engineers to develop and create programs, shaving months off development time and increasing both the performance and efficiency of applications code. In addition, as hardware technology advances or specific algorithms change, the software code-base does not become obsolete. enParallel products are available in two forms: (1) custom solutions in the form of specific accelerated applications, and (2) generic applications toolboxes and function libraries for MATLAB, SciLab, SOFA and RLab math-modeling applications. Prices start around $9,000.00 per user.