Aerospace HOL Computer. Volume II. Direct Execution Tradeoffs.
Final rept. Jun 71-Nov 72,
LOGICON INC SAN PEDRO CA
Pagination or Media Count:
This volume presents an analysis of higher order language computer architectures. These architectures execute an HOL directly without the aid of a compiler. The implications of directly implementing six scientifically oriented HOLs FORTRAN, ALGOL, JOVIAL, APL, and SPL and PLI are discussed, making use of a method which formally specifies the syntax and semantics of these languages. It was found that the gross number of features in a language had the most influence on the degree of implementation difficulty, with PLI and SPLMark IV being the two largest languages and FORTRAN IV and ALGOL 60 the smallest. Ramifications of preprocessor implementation were analyzed, the preprocessor being an assemblerlike software tool for converting the HOL source program to its internal memory form. Preprocessorarchitecture tradeoffs were discussed and a set of recommended proprocessor transformations was prepared. These included removal of blanks and comments from the source program, and the conversion of symbolic names to machine addresses. Problems designing and programming the preprocessor were analyzed. The Appendices in this volume contain a description of the SPLMARK IV subset for which an architecture has been designed, an analysis of optimal storage retrieval schemes, and an introduction to the transduction equation concept used in designing an architecture for SPL. Author
- Computer Programming and Software
- Computer Hardware