Title

Architectural approaches to the design of LISP oriented reduced instruction set machines

Date of Award

1987

Availability

Article

Degree Name

Doctor of Philosophy (Ph.D.)

Department

Electrical and Computer Engineering

First Committee Member

Borivoje P. Furht, Committee Chair

Abstract

In this study The Reduced Instruction Set Computers (RISC), architectural concept was applied to the Artificial Intellience (AI) applications, namely, the List Processing language LISP. The study investigates the most important aspects of the RISC architecture (i.e., Instruction Set, Call/Return Mechanism, and Pipeline Organization). These architectural features are, then, tailored to fit the nature the LISP language.For the first time, to our knowledge, a software tool has been developed in this thesis to experiment with the dynamic execution of many programs that were determined to represent the LISP language applications in artificial intelligence. This tool collected data relating to the execution of machine language instructions. The most frequently used machine instructions were found, and an instruction set for a new RISC oriented LISP was proposed based on these results. A complete analysis of the frequency of memory and register usage by LISP programs, as well as the frequency of different addressing modes, has been presented in this thesis. It was found that the Call/Return operations consumed more than 50% of the total execution time of LISP programs due to the recursive nature of LISP. This factor prompted the study and development of an efficient Call/Return mechanism designed especially for LISP execution. An Overlapping Register Windows (ORW) scheme was introduced, and a replacement strategy was found as a result of simulating the ORW.

Keywords

Artificial Intelligence; Computer Science

Link to Full Text

http://access.library.miami.edu/login?url=http://gateway.proquest.com/openurl?url_ver=Z39.88-2004&rft_val_fmt=info:ofi/fmt:kev:mtx:dissertation&res_dat=xri:pqdiss&rft_dat=xri:pqdiss:8813674