ADAPTING LARGE SYSTEMS TO SMALL MACHINES Thomas Stuart Courant Institute of Mathematical Science New York University USING A DISTRIBUTED MINI-COMPUTE~ NETWORK TO AUTOMATE A BIOCHEMICAL LABORATORY* William J. Lennon, Eric L. Baatz Sandra L. Colussi, Kenneth E. Kinnear Jr. Northwestern University Evanston, Illinois 6020] Abstract An evaluation of software available on mini-computers and its adequacy was made. It led to extension of a systems implementation language to two minicomputers. The language LITTLE was chosen for its machine independence and efficiency. Techniques for code generation and storage allocation are recounted with a special emphasis on producing compact code. Key words: Systems implementation languages, LITTLE, Cross-compilers, Machine independence, Code generators, Storage allocation, Honeywell Series 16, PDP-]I. Introduction This research considers a distributed system constructed by networking small computers as an appropriate architecture for the automation of laboratories having an integrated function. The simplicity, inherent reliability, and flexibility of such an architecture can be realized only if the laboratory as a whole and its proposed final operational characteristics are considered from the very beginning. This desire to automate an integrated laboratory forces us to address not only the normal problems associated with instrument automation but also the goal oriented behavior of the laboratory. We have clarified these requirements by extending the accepted characterization of instrument automation and also by characterizing the levels of interaction appropriate for laboratories which must attain successive levels of integration. Because the distributed computer system is capable of supporting either integrated or independent activities and because computing power can be added incrementally as needed, the distributed system must be viewed as a design tool easily as important as those traditionally employed. We have access to the usual tools associated with software and hardware production -- e.g. translators, editors, document generators -- but also the facilities to utilize them simultaneously with production programs in a reliable and flexible manner. We will provide the necessary background material to motivate our design methodology by examining a particular laboratory automation project. The ~oal of the project is to automate a six month manual set of procedures such that the total elapsed time (with automation) averages about two weeks with overlapped experiments being completed nearly every day. We will detail our characterizations of the general problem being addressed and then describe the model we have selected to structure the design process. To do this, we hypothesize the existence of a ma~lu~lly operated laboratory which differs only in speed of operation from the final automated laboratory. By modeling this hypothetical laboratory in tel~ns of traditional manual roles -- i.e. Chemist, Technician, and Instrument Operator -- automation may proceed by tl~e gradual replacement of mammal functions with their automated eounterpazts. Finally, we will describe the system softwa.L~e required to implement the distributed computer system. * q~is research is supported by N.I.h. C, ra~t No. ]-N0] -GM2004 ~-01 -CON1. CORPORATE INTEGRATION TKROUCH SYST]{24S SOFTWARE William M. McKeeman Abstract A software development system is proposed. It is to be used by a manufacturer of a product that contains a minicomputer, and where the minicomputer may have to be programmed differently for each customer. The programming effort yields (1) software tooIs on a powerful central computer, (2) invariant code for the mini, and (3) mini code that can vary under customer demand. The emphasis is on managing the complexity of (3) as the number of product deliveries increases. The techniques of cross-assembly, emulation and real-time event simulation are used to aid the development team. Superimposed on the development tools are driver programs that allow sales and maintenance people to interact with the tools. The result is both enhancement of these non-development efforts and a further reduction in the costs of development. - -

/lp/association-for-computing-machinery/using-a-distributed-mini-computer-network-to-automate-a-biochemical-4sGhaO750t