Search

A language concept for concurrent processes without common variables is introduced. These processes communicate and synchronize by means of procedure calls and guarded regions. This concept is proposed for real-time applications controlled by microcomputer networks with distributed storage. The paper gives several examples of distributed processes and shows that they include procedures, coroutines, classes, monitors, processes, semaphores, buffers, path expressions, and input/output as special cases.

General equations are derived for estimating the maximum possible utilization of main storage partitions, CPU and I/O devices under different conditions in an idealized CPU-I/O overlap model of multiprogrammed computer systems. The equations are directly applicable to any configuration consisting of sets of identical CPU's, I/O processors, main storage partitions and user tasks. Examples are provided to illustrate the use of the equations to compute effective processing time per record and expected timesharing response time under both balanced and unbalanced resource utilization conditions.

Determining a policy for efficient allocation and utilization of a set of disk drives with differing operational characteristics is examined using analytical techniques. Using standard queueing theory, each disk drive is characterized by a queueing model with service time of a disk drive represented by the probability density function of the sum of two uniform distributions. Total response time of the set of disk models is then minimized under varying load conditions. The results indicate that faster devices should have higher utilization factors and that the number of different device types utilized tends to decrease with decreasing load. Specific examples using 2314 and 3330 combinations are examined.

Using readily available data on CPU characteristics—main memory size, cache memory size, number of channels, and machine cycle time—it is possible to predict relative CPU performance for a wide range of machines. Statistical analyses indicate that these characteristics explain virtually all the variance in relative performance.

This paper presents the background and algorithms for masking the rotational latency of a disk or drum. It discusses the anticipatory input and output of blocks of data to buffer and primary memories for a mono-programmed computer system. A basic permutation algorithm and several variations are given. Because of the anticipatory nature of the I/O scheduling, these algorithms are restricted to classes of programs with predictable behavior. While the methods are not restricted to numerical computations, matrix and partial differential equation methods are typical examples of their use. It is shown that latency may be masked using a small amount of buffer memory. The methods discussed are independent of the overall size of the data base being considered.

Approximate queueing models for internal parallel processing by individual programs in a multiprogrammed system are developed in this paper. The solution technique is developed by network decomposition. The models are formulated in terms of CPU:I/O and I/O:I/O overlap and applied to the analysis of these problems. The percentage performance improvement from CPU:I/O overlap is found to be greatest for systems which are in approximate CPU:I/O utilization balance and for low degrees of multiprogramming. The percentage improvement from I/O:I/O overlap is found to be greatest for systems in which the I/O system is more utilized than the CPU.

An overview of the current and future positions with respect to operating system performance is given. While a great deal of information and a large number of models for subsystems have been developed, gaps still exist in our knowledge. Because of the severe interactions between the various subsystems of an operating system, an overall model of the total system must be developed to be able to analyze and design the performance aspects of an operating system Although such total system designs are exceptional today, it is projected that they will become increasingly more common and necessary in the near future. Such a design philosophy will clearly have a severe impact on the way we go about modularizing operating and computer systems.

The ALGOL 60 language as first defined made no explicit reference to input and output processes. Such processes appeared to be quite dependent on the computer used, and so it was difficult to obtain agreement on those matters. As time has passed, a great many ALGOL compilers have come into use, and each compiler has incorporated some input-output facilities. Experience has shown that such facilities can be introduced in a manner which is compatible and consistent with the ALGOL language, and which (more importantly) is almost completely machine-independent. However, the existing implementations have taken many different approaches to the subject, and this has hampered the interchange of programs between installations. The ACM ALGOL committee has carefully studied the various proposals in an attempt to define a set of conventions for doing input and output which would be suitable for use on most computers. The present report constitutes the recommendations of that committee.

As experience with use of on-line operating systems has grown, the need to share information among system users has become increasingly apparent. Many contemporary systems permit some degree of sharing. Usually, sharing is accomplished by allowing several users to share data via input and output of information stored in files kept in secondary storage. Through the use of segmentation, however, Multics provides direct hardware addressing by user and system programs of all information, independent of its physical storage location. Information is stored in segments each of which is potentially sharable and carries its own independent attributes of size and access privilege. Here, the design and implementation considerations of segmentation and sharing in Multics are first discussed under the assumption that all information resides in a large, segmented main memory. Since the size of main memory on contemporary systems is rather limited, it is then shown how the Multics software achieves the effect of a large segmented main memory through the use of the Honeywell 645 segmentation and paging hardware.

A system developed at Harvard for graphically inputting and outputting nonstandard characters on a computer is printed. In principle, the system can deal with any orthography, although at present it is limited to 4000 Chinese characters and some mathematical symbols. New characters can be added to the repertoire of the system by graphical input on a display scope. Text inputting is accomplished via a display scope or a Rand Tablet. The organization and operation of the current system are described, and a discussion of the relative merits of such a system is given. Illustrations of the computer input and output of Chinese characters are included.

List processing systems should be designed to facilitate production of large programs to manipulate large complex symbolic data stores. This paper presents an overview of a number of system features which the author feels are important to improve the productivity of programmers working in such domains. A systems view is taken, rather than focusing just on language features, since algorithms must be not only coded in a language form, but debugged, modified, made efficient, and run on data. Because of this general framework, the requirements specified are applicable to the design of advanced programming systems for a wide range of applications. Three aspects of programming systems are highlighted: good interactive facilities, programmable control structures, and sophisticated data communication mechanisms. Interactive features are described to facilitate program composition, entry, testing, debugging, editing, optimization, and packaging. Implementation of a generalized environment structure model specified would allow programming of various control regimes including multiprocesses, coroutines and backtracking. Alternative methods of procedure invocation required include invocation by pattern and by monitoring condition. The need for extended data forms, storage management, and extensibility are stressed, as is the duality of data retrieval and function evaluation. Syntax directed input and output of data would facilitate use of complex data stores.

TENEX is a new time sharing system implemented on a DEC PDP-10 augmented by special paging hardware developed at BBN. This report specifies a set of goals which are important for any time sharing system. It describes how the TENEX design and implementation achieve these goals. These include specifications for a powerful multiprocess large memory virtual machine, intimate terminal interaction, comprehensive uniform file and I/O capabilities, and clean flexible system structure. Although the implementation described here required some compromise to achieve a system operational within six months of hardware checkout, TENEX has met its major goals and provided reliable service at several sites and through the ARPA network.

A statement which is similar to the NAMELIST statement of FORTRAN IV has been incorporated in the FORTRAN 63 compiler. The FORTRAN 63 implementation allows a greater flexibility and simplicity than the FORTRAN IV feature. The Hollerith names, the location, the made and the dimensions of a variable can be discovered by means of standard FORTRAN statements. Methods of using this information are illustrated in relation to general purpose data directed input and output routines; some other uses such as matrix manipulation are discussed.

Presented in this paper is a technique for determining the relative effects of the internal speed of the computer and the speed of the input-output units upon the overall speed of the system. Equations are derived which permit the determination of these effects from hardware usage measurements.

The author feels that the interrupt feature which is available in most modern computers is a potent source of programming pitfalls and errors, and that it therefore may heavily contribute to the unreliability of programs making use of it. A programming scheme is presented which avoids the concept of the interrupt and permits the specification of concurrent (or pseudoconcurrent) activities in a supposedly more perspicuous manner. It is intended to serve as a basis for the construction of operating systems, which are prime examples of programs with concurrent activities. The scheme includes a set of basic instructions for the generation, termination, and synchronization of parallel processes. A set of routines representing these instructions and thereby simulating a hypothetical machine organization has been implemented and tested on the IBM System/360. Two programs using these instructions, written in PL360, are presented.