Design, implementation, building and evaluation of a Torus Double Transitive Closure Network of Transputers
Frazao Sosa, Jose Ignacio
Kodres, Uno R.
Adams, Richard A.
MetadataShow full item record
Currently the design of highly parallel supercomputers is one of the most challenging problems in engineering. The purpose of this thesis is to describe how the problem was approached in the design, implementation and building of a Torus Double Transitive Closure Network of Microprocessors, using the T414 Transputer as the basic Unit of Computation. Also compares the performance of the evolved model, from one Transputer to the final stage of sixteen Transputers running in parallel. All the programs and examples presented in this thesis were implemented in the OCCAM2 Programming Language, using the Transputer Development System, D700c, BETA 2.0 release March 1987 compiler version.
Approved for public release; distribution is unlimited
Showing items related by title, author, creator and subject.
Design, implementation and evaluation of an abstract programming and communications interface for a network of transputers Bryant, Gregory R. (Monterey, California. Naval Postgraduate School, 1988-06);This thesis presents the design, implementation and evaluation of two abstracted programming and communication interfaces for developing distributed programs on a network of Transputers. One interface uses a shared memory ...
Design, implementation, and evaluation of a virtual shared memory system in a multi-transputer network. Hart, Simon J. (1987-12);This thesis presents the design, implementation, and evaluation of a virtual shared memory in a multi-Transputer network. The thesis explores the Transputer hardware implementation model and highlights the important ...
Dundar, Cem Ali (Monterey, California. Naval Postgraduate School, 1994-03);Line-of-sight (LOS) calculation for the Janus combat simulation model is critical to the processes being simulated and impacts the run speed (ratio of game time to real time), since it may be the single most computationally ...