A hybrid MIMD/DF compiler for parallel processing

Nakhaee, N., 1992. A hybrid MIMD/DF compiler for parallel processing. PhD, Nottingham Trent University.

10290201.pdf - Published version

Download (29MB) | Preview


A new parallel detection algorithm is devised based on the automatic construction and execution of Petri nets for sequential source programs. The algorithm forms part of a hybrid data-flow and MIMD compiler written in POP-11 and accepts Pascal-S source code.

During the compilation process a Petri net model of the input program is constructed. Execution of the resulting net generates a multi-layered code to reveal the full parallelism inherent in the source program. Each layer consists of several independent parallel statements, which are statically allocated to available processing nodes. The allocator optimizes the communication overhead by using a novel static load balancing technique. Both medium and fine grain parallelism are exploited. Fine grain parallelism is implemented by introducing the co-processor concept.

The implementation offers several other novel features including table-driven analysers (potentially adaptable for different source languages), an algorithm for manipulating symbol tables, and combining parallelism detection and scheduling to eliminate the multiple assignment problem.

Full description of all algorithms including many examples is provided.

Item Type: Thesis
Creators: Nakhaee, N.
Date: 1992
ISBN: 9781369324501
Rights: This copy of the thesis has been supplied on condition that anyone who consults it is understood to recognise that its copyright rests with its author and that no quotation from the thesis and no information derived from it may be published without the author’s prior written consent.
Divisions: Schools > School of Science and Technology
Record created by: Linda Sullivan
Date Added: 16 Jun 2021 13:37
Last Modified: 12 Oct 2023 11:07
URI: https://irep.ntu.ac.uk/id/eprint/43091

Actions (login required)

Edit View Edit View


Views per month over past year


Downloads per month over past year