This paper presents an overview of a study, performed during 1971 and 1972, to design a computer architecture capable of executing a higher-order aerospace programming language directly—without the aid of a compiler or interpreter. The prime motivation for developing such a machine is to reduce system costs, for while hardware logic is becoming much cheaper, software is consuming a greater proportion of total system costs. A tremendous potential savings can be obtained by designing computer hardware that is oriented to aiding the programmer rather than to simplifying the computer designer's job. In this three-phase study, the first phase, preliminary to the actual design, analyzed direct execution tradeoffs for a number of languages. The languages examined were FORTRAN, ALGOL, PL/I, APL, JOVIAL, and SPL. In the second phase, the functional design was produced for a direct-execution system for a subset of Space Programming Language (SPL)/Mark IV. In the third phase, the performance of that system was evaluated by comparing it to a conventional architecture. The study successfully demonstrated that a higher-order aerospace language computer is feasible and efficient.
Two questions perpetually arising within Computer Science Departments are (1) whether or not the programming languages stressed in the academic programs are appropriate and (2) whether the correct emphasis is being placed on one language as opposed to another. In particular, one always wonders if any emerging and promising language is receiving enough attention. Attempts at answering such questions seem to provide quite suitable vehicles for student involvement in curriculum development and so recently at New Mexico State University a combined faculty-graduate student study was made for one important appropriate language, PL/I.