Hybrid Computer Systems Guest Editor's Introduction Robert M. Howe University of Michigan Today's increasing emphasis on simulation of continuous dynamic systems lends a special signi- ficance to this issue of Computer. Hybrid computers, with their tremendous computing speed resulting from the all-parallel configuration of their analog subsystem, are particularly suited for dynamic system simulation. Indeed, back in the 1950's analog computers were the only game in town when it came to real-time simulation of flight vehicles. Digital computers of that era were simply too slow in the numerical solution of differential equations. The first generation of hybrid computer systems appeared in the 1960's, when digital computers were combined with analog computers. Digital computers provided storage and logic capability along with increased arithmetic power and precision to augment the continuous integration capability and speed of analog computers. The resulting hybrid computers were capable of expanded real- time operation and even permitted faster-than-real- time simulation of dynamic systems-a highly desirable feature allowing cost-effective iterative solutions of optimization problems, partial dif- ferential equations, and dynamic stochastic prob- lems. As second- and third-generation hybrid com- puter systems have evolved, the digital computer has been used more and more for high-speed setup July 1976 and checkout of the analog subsystem as well as the performance of specialized computational tasks not easily accomplished on the analog, such as function generation. The five papers in this issue emphasize both present and future directions of hybrid computa- tion. The first paper, by Paul Landauer of Elec- tronic Associates, Inc., is largely tutorial and de- scribes typical current hybrid systems, how they operate, and where they are applied. Cost effective- ness of hybrid computers versus all-digital com- puters in solving dynamic problems is emphasized in this paper. The second paper, by Dr. Harriett Rigas of Washington State University and currently on leave at the National Science Foundation, deals specifically with hybrid computer software systems, including a discussion of hybrid compilers. These compilers use a source program in a Fortran or simulation-language format, automatically trans- lating this into a hybrid object code, including analog component assignments and patching lists. Such compilers greatly increase the ease of pro- gramming of hybrid computers. Next a paper by Landauer and myself presents a quantitative method of comparing the speed of hybrid and all-digital computers. It is based on 13
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Hybrid ComputerSystemsGuest Editor's IntroductionRobert M. HoweUniversity of Michigan
Today's increasing emphasis on simulation ofcontinuous dynamic systems lends a special signi-ficance to this issue of Computer. Hybrid computers,with their tremendous computing speed resultingfrom the all-parallel configuration of their analogsubsystem, are particularly suited for dynamicsystem simulation. Indeed, back in the 1950'sanalog computers were the only game in town whenit came to real-time simulation of flight vehicles.Digital computers of that era were simply too slowin the numerical solution of differential equations.The first generation of hybrid computer systems
appeared in the 1960's, when digital computerswere combined with analog computers. Digitalcomputers provided storage and logic capabilityalong with increased arithmetic power and precisionto augment the continuous integration capabilityand speed of analog computers. The resultinghybrid computers were capable of expanded real-time operation and even permitted faster-than-real-time simulation of dynamic systems-a highlydesirable feature allowing cost-effective iterativesolutions of optimization problems, partial dif-ferential equations, and dynamic stochastic prob-lems. As second- and third-generation hybrid com-
puter systems have evolved, the digital computerhas been used more and more for high-speed setup
July 1976
and checkout of the analog subsystem as well asthe performance of specialized computational tasksnot easily accomplished on the analog, such asfunction generation.The five papers in this issue emphasize both
present and future directions of hybrid computa-tion. The first paper, by Paul Landauer of Elec-tronic Associates, Inc., is largely tutorial and de-scribes typical current hybrid systems, how theyoperate, and where they are applied. Cost effective-ness of hybrid computers versus all-digital com-puters in solving dynamic problems is emphasizedin this paper.The second paper, by Dr. Harriett Rigas of
Washington State University and currently onleave at the National Science Foundation, dealsspecifically with hybrid computer software systems,including a discussion of hybrid compilers. Thesecompilers use a source program in a Fortran orsimulation-language format, automatically trans-lating this into a hybrid object code, includinganalog component assignments and patching lists.Such compilers greatly increase the ease of pro-gramming of hybrid computers.Next a paper by Landauer and myself presents
a quantitative method of comparing the speed ofhybrid and all-digital computers. It is based on
13
calculating the digital operations per second equiva-lent to various analog components such as integra-tors, summers, multipliers, etc. The total digitaloperations per second equivalent to typical currenthybrid systems is shown to be upwards of severalhundred million operations per second. This givessuch hybrid computers a one- to three-order ofmagnitude speed advantage over the full spectrumof scientific digital computers.Despite the improvements in hybrid computer
system performance over the past decade, thecomputing load for dynamic system simulation hasshifted more and more to all-digital systems. Thisis because the hybrid advantages of high computingspeed and good man-machine interaction are offsetby the disadvantages of programming complexity,and lengthy problem debugging and turnaroundtimes. In recent years the dramatic cost decreasesand speed increases in digital computers, along withthe development of easy-to-use compilers such assimulation languages, have accelerated the move-ment toward all-digital solution of dynamicproblems.Hence, hybrids have languished, and the general-
purpose hybrid computer market has been in anessentially no-growth posture for a number of years.For this reason the manufacturers of hybrid equip-ment have lacked the resources to develop a new,
Proceedings of the Special Symposiumon Advanced Hybrid Computing, July23-24, 1 975.
27 papers examining various aspects ofadvanced hybrid computing systems.208 pages. Price: $20.80
Use order form on p. 88; specify R75-328.14
fourth generation of hybrid computer systems thattake full advantage of the present-day componenttechnology which has been so important in thedevelopment of the latest general-purpose digitalcomputers.
In recognition of this, two years ago the U.S.Army Materiel Command let study contracts toeach of the several hybrid manufacturers to deter-mine the optimum hardware and software con-figuration for the next generation of hybrid com-puters. The last two papers in this issue indicatethe future direction of hybrid computer systemsand are in part based on results of the AMC study.One paper is by Rubin and Mawson of EAI and
the other is by Fadden and Graber of AppliedDynamics International. Both papers emphasizethe importance of taking full advantage of currentand projected integrated-circuit component technol-ogy in development of next-generation machines,components such as high-performance IC opera-tional amplifiers, monolithic DAC's and ADC's,low-cost solid-state digital memory chips, and LSIanalog switch arrays. Use of these switch chipswill allow replacement of the hand-patched problemboards on current hybrids with automatic elec-tronic patching. This in turn will permit problemturnaround in milliseconds, and as a result, largehybrid computers will be opened up to many moreusers operating on a time-shared basis throughterminals, both local and remote. The low cost,small size, and low power of the IC componentsin t-he next-generation hybrid will mean that muchlarger hybrid computers can be assembled at costssimilar to today's machines, with correspondingincreases in the cost effectiveness of hybrid com-puters. The availability of electronic patching willalso make more effective the implementation ofhybrid compilers using a simulation language orother source program, so that users with no parti-cular hybrid expertise can easily program thesenew hybrid computers. R
Robert M. Howe is professor and chairman_ of the Department of Aerospace Engineering
at the University of Michigan, Ann Arbor,where he has been involved with teaching
_w g and research in analog and hybrid computers,automatic control, and flight simulation. Hehas been a consultant to numerous com-panies and is a founder of Applied Dy-namics, Inc. (now Applied Dynamics Inter-national).
Howe was the first national chairman of Simulation Councils,Inc. (now SCS), has been a member of several USAF advisorypanels, and has published over 30 scientific articles and onebook. He received his BSEE from Caltech, his AB in physicsfrom Oberlin College, his MS in physics from the Universityof Michigan, and his Ph. D in physics from MIT. He is amember of the IEEE Computer Society, Tau Beta Pi, PhiBeta Kappa, Sigma Xi, AIAA, SCS, and the BiomedicalEngineering Society.
COMPUTER
Related Documents
