Abstract

As computing architectures are being implemented in late and post silicon technologies, fault tolerance and concurrent operation are becoming increasingly important. It is already common knowledge that manufacturers are putting two, four or even more cores on a single silicon die to improve computing performance. The proposed architecture far exceeds this number by grouping thousands or even millions of simple reduced instruction set computing (RISC) processors, each of which is capable of a single operation at a time, and to communicate with its eight nearest neighbors. In this architecture, if a single core or cluster of cores have defects at the time of manufacture, or later in the life of the system, it is possible to test and disable them as necessary. A fine-grained architecture of this kind calls for a parallel programming style. One approach to this problem is the use of a parallelizing compiler. Another approach may be to use one of the several application programming interfaces (APIs) available for standard text based programming languages, with some built-in features for parallel programming. This work has generated a solution for creating machine level parallel programs for the massively parallel computer architecture described above using text and graphical means. To support this programming method, an integrated development environment (IDE) and a zero communication latency, register transfer level (RTL) simulator have been developed. Experimental results include the implementation of fundamental data processing algorithms and complex functions.

Library of Congress Subject Headings

Computer architecture--Design; Parallel programming (Computer science); Parallel processing (Electronic computers); Multiprocessors

Publication Date

2010

Document Type

Thesis

Department, Program, or Center

Microelectronic Engineering (KGCOE)

Advisor

Patru, Dorin

Comments

Note: imported from RIT’s Digital Media Library running on DSpace to RIT Scholar Works. Physical copy available through RIT's The Wallace Library at: QA76.9.A73 M87 2010

Campus

RIT – Main Campus

Download

Share

COinS