This thesis is concerned with parallel computer architecture. It surveys the many and varied ways in which parallelism can manifest itself in computers and particularly the role played by the connections between processors.
Having set a foundation for reasoning about parallel computers the transputer and its role in this market are surveyed. Several important parallel computers not based on the transputer are also described in order to put the transputer into perspective. Based on these observations a new computer taxonomy is proposed.
This is the 2nd version of this thesis.
After submission, the 1st version of the thesis was scrutinized by the examiners for 13 months after which they produced the following list of 3 deficiencies that made the thesis unacceptable, hence it was failed.
For the on-line version of the thesis I have returned the references to the end of each chapter.
Some time after submitting this thesis and self-publishing it on the web, I have been contacted by several primary sources, original researchers and original programmers who have generously taken time to expand on details I had not discovered in the literature. If I had/properly assessed this information at the time I would had presented some parts differently.
Although web pages can be readily updated, I feel it would give a false impression that my thesis was more accurate and complete then it was. At the same time I don't want the inaccuracies in the thesis to perpetuate so I have added several blocks of 'Corrections & Addendum' in the relevant sections.