A Multi-Time Scale Morphable Software Milieu for Polymorphous Computing Architectures (PCA) - Composable, Scalable Systems
Abstract
Polymorphous Computing Architectures (PCA) rapidly "morph" (reorganize) software and hardware configurations in order to achieve high performance on computation styles ranging from specialized streaming to general threaded applications; the former is most like a pipeline processor, while the latter is most like a collection of common desktop processors. This project explored a multi-time scale morphable software environment (denoted milieu') whose purpose was to suggest ways to build applications from components of varying scales; these components were to include small and large amounts of code, different algorithms, and different parallelism. The purpose of PCA is to reduce the cost, size, power, and other "footprint" aspects of computing that can be most efficiently done with radically different kinds of computing elements; pre-PCA systems connect disparate parts with low utilization outcomes. This project sought to create a model-based software framework for reactive monitoring, optimization, modeling, resource negotiation and allocation, regeneration, and verification. Model-based software uses principles like "legos" or "tinker toys" to model interactions and roles, and allow a framework in which software is created. This project report addresses the research objectives and key accomplishments, while introducing the technical concepts and approaches developed in this project for the PCA software development. Future work is also recommended in this report.
Document Details
- Document Type
- Technical Report
- Publication Date
- Oct 01, 2004
- Accession Number
- ADA427345
Entities
People
- Anthony Skjellum
- Hong Yuan
- Yoginder S. Dandass