| EPSRC Reference: |
GR/J43295/01 |
| Title: |
ALGORITHMS, ARCHITECTURES & MODELS OF COMPUTATION: SIMULATION EXPERIMENTS IN PARALLEL SYSTEMS DESIGN |
| Principal Investigator: |
Ibbett, Professor R |
| Other Investigators: |
|
| Researcher Co-Investigators: |
|
| Project Partners: |
|
| Department: |
Sch of Informatics |
| Organisation: |
University of Edinburgh |
| Scheme: |
Standard Research (Pre-FEC) |
| Starts: |
01 September 1993 |
Ends: |
28 February 1997 |
Value (£): |
234,097
|
| EPSRC Research Topic Classifications: |
| Parallel Computing |
System on Chip |
|
| EPSRC Industrial Sector Classifications: |
|
| Related Grants: |
|
| Panel History: |
|
|
Summary on Grant Application Form |
|
The project is a more component of an ongoing long-term strategy to undertake quantitative evaluation of computer architectures (in terms of both cost-effectiveness and scalability) through the use of an architectural simulation facility. This project will investigate architectures to support the H-PRAM model of parallel computation in a cost-effective and scalable way through the use of a hierarchical architectural design and simulation facility, HASE, being designed and implemented as part of the project.Progress:The project is proceeding well. Both Research Associates are in post and the equipment has been bought.In the area of H-PRAM implementation techniques, work so far has resulted in two papers. The first describes a theoretical analysis of implementing the H-PRAM on a Peano-indexed mesh, appearing in Aizu International Symposium on parallel algorithms/architecture synthesis in March. The other, to be submitted in about a month, deals with practical aspects of harsh functions and memory organisation and their implementations via routing on a Peano-indexed mesh. This work is for an EREW H-PRAM, the weakest variant, and includes the results of some simulation experiments. Impending work will consider the implementation of synchronisation. We have also designed a prototype variant of the FORK programming language, augmented with a construct which captures the notion of partitioning present in the private H-PRAM. We have implemented an interpreter for the language which generates traces of H-PRAM computations for use as simulation input. We now intend to use this to produce simulations of complete stub H-PRAM programs (where stub means consisting of stub computation and communication steps with no specific work performed in the computations). We will then move on to trace real data-dependent programs. The implementation of the HASE infrastructure, using the object oriented database Object-Store, has reached the point where support for the types of complex experiments envisaged in ALAMO is now possible. This support is the key to increasing the productivity of the user of HASE models, making it possible to concentrate on modelling and analysis of results, rather than the mechanics of executing the models. This has actually worked better than we had hoped and it seems that the system developed would form an excellent basis for a project to use objet techniques to define a general candidate modelling environment. This idea is likely to form the basis of a new proposal to EPSRC.
|
|
Key Findings |
|
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
|
|
Potential use in non-academic contexts |
|
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
|
|
Impacts |
|
| Description |
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk |
| Summary |
|
| Date Materialised |
|
|
|
|
Sectors submitted by the Researcher |
|
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
|
| Project URL: |
|
| Further Information: |
|
| Organisation Website: |
http://www.ed.ac.uk |