Adam Ferrari, Adrian Filipi-Martin, and Soumya Viswanathan, The NAS Parallel Benchmark Kernels in MPL, September 12, 1995.
Abstract: The Numerical Aerodynamic Simulation (NAS) Parallel Benchmarks are a set of algorithmically specified benchmarks indicative of the computation and communication needs of typical large-scale aerodynamics problems. Although a great deal of work has been done with respect to implementing the NAS Parallel Benchmark suite on high-end vector supercomputers, multiprocessors, and multicomputers, only recently has the possibility of running such demanding applications on workstation clusters begun to be explored. We implemented a subset of the NAS benchmarks using the Mentat Programming Language, and ran performance tests on a cluster of workstations using the Mentat system. We compared our performance results to previous NAS benchmark tests using the Parallel Virtual Machine (PVM) system in a similar hardware environment. We found that due to algorithmic improvements, efficient communications provided by the Mentat system, and low introduced overheads even at the higher level of programming abstraction provided by Mentat, we observed significantly improved performance in a number of cases.
Andrew S. Grimshaw, W. Timothy Strayer, Padmini Narayan, The Good News About Dynamic Object-Oriented Parallel Processing, December 17, 1992.
John F. Karpovich, Matthew Judd, W. Timothy Strayer, Andrew S. Grimshaw, A Parallel Object-Oriented Framework for Stencil Algorithms, January 27, 1993.
Abstract: We present an object-oriented framework for constructing parallel implementations of stencil algorithms. This framework simplifies the development process by encapsulating the common aspects of stencil algorithms in a base stencil class so that application-specific derived classes can be easily defined via inheritance and overloading. In addition, the stencil base class contains mechanisms for parallel execution. The result is a high-performance, parallel, application-specific stencil class. We present the design rationale for the base class and illustrate the derivation process by defining two sub-classes, an image convolution class and a PDE solver. The classes have been implemented in Mentat, an object-oriented parallel programming system that is available on a variety of platforms. Performance results are given for a network of Sun SPARCstation IPCs.
Laurie MacCallum, Andrew S. Grimshaw, Linear Algebra Library for Mentat Applications, August 06, 1993.
Padmini Narayan, Sherry Smoot, Ambar Sarkar, Emily West, Andrew Grimshaw, Timothy Strayer, Portability and Performance: Mentat Applications on Diverse Architectures, July 22, 1992.
[Overview] [Documnetation] [Mentat Team] [Distribution] [Sample Programs] [Slides] [New Features]
If you have any questions about Mentat please feel free to e-mail us at firstname.lastname@example.org, Department of Computer Science, Thornton Hall, University of Virginia, Charlottesville, Virginia 22903.