klimbal@riacs.edu (Phil Klimbal) (03/01/90)
Many of the "Realtime Unix" systems provide what I would characterize as "soft realtime" performance (lets say 1 millisecond dispatch latency). What would I do with such a system? How many people actually want all the functionality of a complete Unix system for their time critical applications? --Phil Klimbal klimbal@riacs.edu
kokody2@me.utoronto.ca (Gerry Kokodyniak) (03/01/90)
Phil Klimbal (klimbal@riacs.edu) in <2057@hydra.riacs.edu> writes: > Many of the "Realtime Unix" systems provide what I would > characterize as "soft realtime" performance (lets say 1 > millisecond dispatch latency). What would I do with such > a system? How many people actually want all the functionality > of a complete Unix system for their time critical applications? > > --Phil Klimbal > klimbal@riacs.edu Realtime Unix becomes very useful in data acquisition/control environments. An example is say one wants to control several aspects of an experiment and take data at the same time. In an MSDOS environment one would have to integrate the different pid loops, data acquisition, etc etc into one program and one would have to integrate hardware to handle such things as function generation. If one wanted to add or subtract control/data acquisition modules, the timing of the program would be seriously affected and one would have to rework the program to optimize to the desired timing. On a real-time unix system one could have a child process that is a function generator, another child process acquiring data, another child process controlling one parameter etc etc. With the addition/subtraction of a process, the timing aspects remain the same. One also has the advantages of the development environment of a unix system. Included are networking capabilities thru tcp/ip aloowing one to remotely monitor/control experiments, etc etc. There is also the advantage that it is a unix system in that many machines in universities are unix machines running network file servers such as nfs where one could share resources over the net and have a single type of operating system instead of transfering from MSDOS to unix etc etc. It also appears that NASA will be using a realtime UNIX system (i.e. LYNX OS - a Sys V R 3.2 derivative with BSD features) for use on the space station (for whenever it comes into being). The beauty of the system is that many hardware dependent functions can be taken over with software modules. This becomes more possible with faster and faster micros such as the 80486, 88000, 68040 etc machines. This alows one more flexibility in applying/changing these functions. So in my opinion, I think many different people would have use of such operating systems. There are many uses that probably haven't even been considered to this point that have been hardware dependent till now. Gerry Kokodyniak Gerry Kokodyniak, Ph.D. Student Department of Mechanical Engineering kokody2@me.toronto.edu University of Toronto kokody2@me.utoronto.ca Structural Integrity Fatigue and kokody2@ME.UTORONTO.BITNET Fracture Research Laboratory {linus,allegra,decvax,floyd}!utcsri!me!kokody2 (416) 978-6853
mike@lynx.uucp (Mike Bunnell) (03/02/90)
> >[... >It also appears that NASA will be using a realtime UNIX system (i.e. LYNX OS >- a Sys V R 3.2 derivative with BSD features) for use on the space station >(for whenever it comes into being). >..] > Thank you for mentioning LynxOS. LynxOS is NOT, however, a derivative of SYS VR3 (yuck) or any other operating system. LynxOS was written from scratch as an operating system for running hard real-time applications. Actually, we origionally chose the Berekely 4.2 system call interface (working signals, select, symbolic links, sockets, job control, long file names etc.) System V compatiblity was added later as were 4.3 stuff and POSIX. It is true that you can run System V executables, but that is because everyone has picked System V for there binary coding standards and there is more off-the-shelf software available for system V. However when we compile GNU stuff or X-Windows we make the BSD versions.
m5@lynx.uucp (Mike McNally) (03/02/90)
mike@lynx.uucp (Mike Bunnell) writes: >>It also appears that NASA will be using a realtime UNIX system (i.e. LYNX OS >Actually, we origionally chose . . . >System V compatiblity was added . . . >everyone has picked System V for there binary . . . We'll be adding the "ispell" spelling checker to our standard distribution real soon now... -- Mike McNally Lynx Real-Time Systems uucp: {voder,athsys}!lynx!m5 phone: 408 370 2233 Where equal mind and contest equal, go.