Kriterien für die Auswahl von Elektronischen Rechenanlagen für Biomedizinische Forschungsinstitute

Abstract

Computers are now a recognized tool in biomedical research. They are used for the evaluation of data on one hand and on the other hand for data acquisition and control of experiments. Based on our experience, some suggestions concerning the structure of a mini-computer system suitable for a research laboratory are made. According to the two major classes of application, two sets or requirements arise. We argue that it is effective to use this system for data reduction and evaluation because a large percentage of tasks require program development or at least specific input data handling. Therefore, we call for a multi-user time-sharing system which should be equipped with a set of commands that enable the user to perform easily standard tasks such as editing compilation, loading and execution of programs. Some figures are given on memory size, disk file system and other peripherals. In the field of data acquisition and experiment control electrical and logical characteristics of the computer system are of greater importance than in data evaluation. The transient nature of experimental set-ups requires a high degree of adaptability of the systems hardware in regard to changing interface problems. We, therefore, suggest supplying a number of standard interfaces such as RS 232, “paper tape reader” and “paper tape punch”. These connections will generally not suffice so that for the development of custom interfaces it will be necessary to have easy access to the I/O-bus of the CPU, for example by predesigned logic which yields a set of simplified data and timing signals. The structure of the operating system for data acquisition is also discussed. We experienced that a single-user system in a basically non-realtime environment with modification for data acquisition via direct memory access or interrupt requests is sufficient in many cases and should be preferred to a complicated large-scale realtime system. But for some applications, especially for the purpose of control, such a realtime system should be available. Among several other requirements, the demand for the possibility of upgrading processor and memories is stressed. Hard- and software compatibility should be provided to a great extent in case of upgrading. We further argue for mixed hardware and own software development in order to build a suitable system at a reasonable cost. This concept is illustrated by discussing our own system which is basically a PRIME 300 multi-user system equipped with two slave systems. The two slave systems are mainly used for I/0 tasks. The experimental hardware is connected to the system by means of standard interfaces (RS 232, parallel I/0), by modified standard interfaces and by hardware developments of our own. Care was taken to provide for a comfortable analog I/0 system with a communication network to allow access to the computer from all laboratories. The standard operating systems were modified to incorporate the slave systems and the experimental equipment. In conclusion, we suggest that it is necessary to have an intimate knowledge of the hard- and software of the computer system in order to use all the resources of the system in the changing applications. This requires a hard-and software documentation that covers the whole system in great detail.