MPI/Pro optimizes time to solution for parallel processing applications. MPI/Pro works to retain system scalability for applications, while balancing performance criteria (latency versus overhead) and resource utilization to optimize application time to solution.

Increased Performance and Quality: MPI/Pro strives to minimize application runtime, while balancing performance criteria and resource utilization. MPI/Pro supports overlapping of communication and computation to increase overall performance. It outperforms freeware MPI versions on average by 10-20% (some as high as 50%) on most customer applications.

Complete MPI-2 Support: MPI/Pro has all of the functionality of MPI-2. These include all of MPI 1.2 plus one-sided communication, dynamic process management, parallel file I/O and extended collective operations.
MPI/Pro in Production Cluster Computing: "We use MPI/Pro message passing middleware with commercial applications and fracture mechanics codes that run in parallel on 256-processor Windows High Performance Computing clusters. MPI/Pro is a highly reliable, commercially supported product that provides our programmers with a fully compliant implementation of the MPI standard in all areas, including error handling. The engineers at Verari Systems continually update their products to use the latest Microsoft technologies such as .NET."
Gerd Heber, PhD, Research Associate
Cornell Theory Center

Quality and Performance: "With its excellent performance, support products and access to engineers, MPI/Pro was the natural choice for West End Capital to implement its business critical parallel computation environment. MPI/Pro helped us to set new standards for real-time evaluation of risk within our Fixed Income Arbitrage Fund, making use of a large scale parallel Monte Carlo simulation of financial markets."
Mark Byrne, Chairman and CEO
West End Capital Management

  MPI/Pro Data Sheet

VSI/Pro is optimized to take advantage of single instruction, multi-data (SIMD) operations on the processors that support Altivec™ and SSE. These are the popular G4, G5, Intel® Pentium® 4 and AMD Opteron™ processors that are common to both embedded and cluster computing environments. VSI/Pro provides a portable, open standard, high performance out-of-the-box solution that accelerates the development of signal and image processing applications and virtually eliminates the complexities of hand tuning for special hardware/operating system configurations.

The library supports a wide range of signal processing and linear system functions including 1D, 2D and 3D FFTs, 1-D and 2-D convolutions and correlations, both FIR and IIR filters and windowing functions (Hanning, Chebyshev, Blackman and Kaiser).

VSI/Pro provides excellent performance optimization for the complete range of DSP applications on PowerPC®/Altivec (both G4 and G5) and x86/SSE architectures (both 32-bit and 64-bit). As a rule of thumb, our highly optimized SIMD (Altivec/SSE) implementations provide a factor of six to eight times performance boost compared to non-SIMD implementations.

VSI/Pro supports the following operating systems on the G4 processor: VxWorks, LynxOS™ 4, Mercury™ OS, Linux, Mac OS X and OSE. On the latest G5/PPC970 based systems, both Mac OS X and Linux are supported. In addition, libraries are available for both Linux and LynxOS 4 on Intel Pentium 4 and for Linux on AMD Opteron-based platforms.

  VSI/Pro Data Sheet

The VSIPL Standard API (Application Programmer’s Interface) consolidated and streamlined existing mathematical libraries and defined standard functions for scientific and engineering computing. Verari Systems, the leading middleware developer for high performance computing, is an active member of the VSIPL Forum.

VSIPL’s Image Processing API has been designed for optimal performance whether the target application is for an embedded platform, a workstation or a cluster. Some important features of the VSIPL Image API include lowering the development costs for embedded systems by performing optimization “under the hood” and reducing the binary executable size. It also provides great programming flexibility through carefully designed abstract data types as well as providing portability for various kinds of embedded systems.

VSI/Pro Image currently provides the functionality covered by the VSIPL Image Processing API (Version 0.2), and the VSI/Pro CoreLite library, which contains the basic building blocks for signal and vector processing and highly optimized FFT/Convolution/Correlation operations. VSI/Pro Image also offers extended functionalities to better exploit legacy code whenever possible and better transition from legacy code to compliance with the VSIPL standard.

VSI/Pro Image provides excellent performance optimization for different kinds of image processing applications on PowerPC®/Altivec™ architectures. For example, for 2D Fir filter operations, at a size of 3136 x 3136, VSI/Pro Image achieves 4993.7 MOPS on an 800MHz G4 machine. In general, VSI/Pro Image’s highly optimized Altivec implementation provides up to a factor of 20 to 30 times performance boost compared to non-Altivec implementations. Also, the integrated VSI/Pro CoreLite library enables the developer to exploit the computationally intensive signal and vector processing kernels. For example, at size 1K and 2K, its FFT functions can achieve 1.58GFlops on a 400MHz G4 machine.

Verari Systems also provides other Signal/Vector/Image/Linear Algebra libraries for the developer who wants to exploit every cycle and bit of their hardware. Verari Systems mathematical library flagship, the VSI/Pro CoreLite and Core releases, provides the cutting edge performance on PowerPC architecture, the well-done portability on different embedded systems, and the smart APIs for compliance to the VSIPL standard, which is a well developed and recognized research effort in the DoD community.

  VSI/Pro Image Data Sheet