The following slides contain innformation that has been presented in the past concerning different aspects of Mentat.
These slides present an overview of Mentat. They begin by giving a description of the Mentat Approach. Afterword, they talk about existing Mentat Applications and Performances for those applications. The slides then conclude with a brief summary of the slides and some discussion on future work to be done with Mentat
These slides describe how Mentat performs dynamic data dependence detection. They begin with an introduction to the macro dataflow model, the data driven computation model underlying Mentat. Next, they describe the compilation processes and those aspects of the run-time system interface that are used by the compiler. Once the compilation process and the generated code is understood, they focus on the run-time behavior of Mentat objects, including the overhead costs associated with dynamic data dependence detection.
Large meta-systems comprised of a variety of interconnected high-performance architectures are becoming available to researchers. To fully exploit these new systems, software support must be provided that is easy to use, supports large degrees of parallelism in applications code, and manages the complexity of the underlying physical architecture for the user. these slides describe our approach to constructing and exploiting meta-systems. Our approach inherits features of earlier work on parallel processing systems and heterogeneous distributed computing systems. In particular, we build on Mentat, an object-oriented parallel processing system developed at the University of Virginia that provides large amounts of easy-to-use parallelism for MIMD architectures. We have constructed a Metasystems testbed based on Mentat. The testbed permits us to experiment with different ideas and mechanisms for managing both heterogeneity and parallelism. The following slides present our early results from the testbed. This will include a discussion of problem decomposition issues and how they are resolved using user defined call-backs.
These slides present an application template that we have constructed for iterative stencil applications. The template is written in Mentat, a parallel extension of C++. Mentat programs can be executed on a variety of workstation and multicomputer platforms. The structure of the template will be presented, and the "customization" of the template for two applications will be presented, as well as performance results on a network of Sun workstations.
[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 email@example.com, Department of Computer Science, Thornton Hall, University of Virginia, Charlottesville, Virginia 22903.