scotth@hercules.UUCP (Scott Herzinger) (03/19/84)
There have been several opinions expressed as to whether educational organizations should invest in proven technology or jump ahead and buy state-of-the-art. This is, I acknowledge, somewhat oversimplified and inaccurate, but I think you know what I mean, given the example systems being compared from each category: 1- IBM PC (or clone), a well-integrated system of proven, accepted, and supported components. 2- 16032 or 68000 Unix system. This is a class of systems, generally similar in capability and technology. Call this class "state-of-the-art", though, we really mean something like "state-of-the-available-art." This note is not intended to convince anyone either way on this question. I'd just like to mention one criterion to include in the decision-making process. The ability of the educational organization to support a computer science program varies widely. Large, state-funded universities tend to be able to support broader, and more advanced programs. Smaller, especially private, colleges find it much more difficult to provide an adequate program. This is especially true of private liberal arts schools without heavy financial support of industry. Larger institutions can often provide an in-house computer support and service team of hardware and software gurus, consultants, etc. The smaller organizations usually cannot afford to do this. Rather they rely on faculty expertise, talented students that pass through the program, etc.. The problem with this arrangement is that the experts have other, higher priority committments, and support is inconsistent at best, simply not there at worst. The smaller institutions, then, to survive require computing equipment and service to be very reliable, and in the event of failure, rely on vendors and outside repair organizations for maintenance. Systems must be easy to learn and apply, because there is often a low level of institution-local consulting support. The moral of the story is, unless you have a lot of spendable money and a strong local organization with the expertise to help you with your problems (and you will have many) you better invest in equipment that is proven, accepted by a broad user community, and supported by a healthy, available service organization. And by all means, by service contracts. They're the best way to keep your equipment up. FYI: The statements above are not related to my position with my employer, and are my opinion only. I formed these opinions during several years as operations manager of a small academic computing center that supported instructional computing needs of a private liberal arts college on the campus of a small university in the Pacific Northwest. There were about 1350 undergraduate students in liberal arts, about 425 instructional computing users each semester, with about 150 students in computing classes each term. I recieved my BS from this college in 1982. Representatives from similar organizations seeking advice or interested in discussing these issues may contact me. I also continue to discuss instructional computing with the cs dept chairman at my alma mater. Perhaps we all could establish a forum for discussing the unique requirements of these small institutions? Anyone interested? -- Scott Herzinger; Logic Design Systems; Tektronix; Inc. uucp: {ucbvax,decvax,pur-ee,cbosg,ihnss}!tektronix!teklds!scotth CSnet: scotth@tek ARPAnet: scotth.tek@rand-relay --
ron@brl-vgr.ARPA (Ron Natalie <ron>) (03/21/84)
What makes you claim that the 68000 is any more "state-of-the-art" than the 8088 in the PC. Both are relatively mundane. If you are actually going to use the paging 16032 you are getting closer to the state of the art. -Ron
kds@intelca.UUCP (Ken Shoemaker) (03/22/84)
Hey, this is a really great discussion, make me laugh, "state of the art?" What do you mean? As computer science goes, I would scarcely say that the 68000 or the 16000 is "state of the art." I have heard it said that they are great implementations of architectural thinking that has been around since the 60s. Just because it is in a microprocessor does not make it state of the art in an architectural sense. When I think of state of the art, I think of parallel architectures that solve old problems in new ways, not just higher integration implementations of the same old thing. Of course if you are talking about semiconductor technology....... And another thing.....someone (excuse me, I didn't pay any attention to who) mentioned that he wanted to know whether someone with experience with a VAX was more employable than someone who used a 68000 running Unix or an IBM/PC running Unix. I notice that he didn't include the VAX running Unix (was it just an oversight?) BUT I would think that given these three as options, it could really scarcely make much difference, the real differentiation is how they are used, and what you want to do with your life! If you want to be a site manager of a large site, then perhaps working with the VAX as a system administrator/ operator type might help, but if you want to do things like compiler/ application development, or even real world type work, then really you are using the computer for practice. And the line really should be drawn between lumping the VAX on one side, and the 68000/PC on the other, not putting the VAX with the 68000...not that I'm trying to defend the PC, but lets be real in terms of differences in the systems.....hey, this was fun, can we do it again sometime? -- Ken Shoemaker, Intel, Santa Clara, Ca. {pur-ee,hplabs,ucbvax!amd70,ogcvax!omsvax}!intelca!kds
jvz@sdcsvax.UUCP (03/23/84)
After having read what I believe are the majority of articles concerning the best computer for a university to purchase, I have found some lacking information. Namely: What is the purpose of the computer system (what will it teach)? There are many teaching areas which could use computer systems, but they don't all require the same system. For example, the areas below place entirely different demands on computing facilities, and not even your best microprocessing system will satisfy all of them. - computer literacy (word processing, accounting, engineering analysis, etc.) - teaching beginning programming techniques in an information systems department - teaching beginning programming techniques in a computer science department - advanced compiler writing and code optimization - graphics research - computer architecture - real-time computing - multi-processing operating system construction
mason@utcsrgv.UUCP (Dave Mason) (03/24/84)
**** Response to Ken Shoemaker, Intel From your net address, your built-in biases are fairly obvious (doesn't of course mean that they are real, just obvious), but, YOU CAN'T BE SERIOUS! I guess it depends on what you mean by cutting edge of computer architecture: if one were relating 16 and 32 bit micros to mainframe architecture I suggest a pairing something like: 68000 - S/360, 16032 - VAX, 8086 - DG Nova; and while the Nova solves a lot of peoples problems, few would claim that its architecture is as powerful as that of a VAX, or even S/360. You are right of course that these are all fairly conventional VonNeuman architectures, but that doesn't mean that they are equal (or even close) from an architectural viewpoint. The problem with real cutting edge architecture like transputers, Illiac IVs, and (maybe) 432s is that the software technology is such that we can't use them (Occam perhaps to the contrary someday). This does not mean we should all go back to PDP-8s, Novas, and their ilk. We have made real progress down one branch of the computer architecture tree, and we should recognize that and take advantage of it. This means that we (particularly in academia) must not view IBM-PCs and other architectural dinosaurs as other than interim solutions or power supplies and keyboards for micro boards with modern architectures. And we must make it clear to holders of purse strings that that is how we view it. -- Usenet: {dalcs dciem garfield musocs qucis sask titan trigraph ubc-vision utzoo watmath allegra cornell decvax decwrl ihnp4 uw-beaver} !utcsrgv!mason Dave Mason, U. Toronto CSRG CSNET: mason@Toronto ARPA: mason%Toronto@CSNet-Relay
kds@intelca.UUCP (Ken Shoemaker) (03/26/84)
Out of the frying pan...although one may think that I have lots of built-in bias in this area, I think people maybe mistook my point concerning this whole can of worms, so I will take one response and try to clarify what I was saying, perhaps if people want to flame at me *personally* they should do it using mail, rather than bothering the entire net... >if one were relating 16 and 32 bit micros to mainframe >architecture I suggest a pairing something like: 68000 - S/360, 16032 - >VAX, 8086 - DG Nova; and while the Nova solves a lot of peoples problems, You missed the point, how many VAXes do you have that have <40M of secondary storage? How many micros are there that can handle 100 terminals. What I was talking about was not the processor architecture per se, but rather the system architecture. >The problem with real cutting edge architecture like transputers, >Illiac IVs, and (maybe) 432s is that the software technology is such >that we can't use them (Occam perhaps to the contrary someday). This Precisely, if universities are supposed to "explore the cutting edge" then it should be their charter to make these products useable, or even to suggest and implement different, more novel architectures, and then to make them usable. But what are people suggesting now? Go and buy some more VAXes, for which the *hard* problems have been solved and customize the things for the local environment. Now this is fine if people realize that what they are doing is buying a tool, maybe to be used to teach people what a computer is or to explore other problems. If this is the purpose of the computer, however, I would suggest that you would do much better buying a pre-made environment with lots of available support. Users are going to get very frustrated if they try to use a development system in a production environment! >This means that we (particularly in academia) must not view IBM-PCs >and other architectural dinosaurs as other than interim solutions or It is my feeling that *any* computer is an interim solution, all the more reason to get one that comes up with the least amount of effort or grief. Also, if you look at most modern computers as complex instruction set computers, and are really in to RISC as the trend of the future, then one could say that 8086s, 68000s, 16000s, Z80000s etc, are all architectural dinosaurs. Anyway, lest we forget, most people, including University Computing centers, view computers as tools to solve problems, not as works of art, subject to interpretation as to their beauty. If that is what you want to do with computers, great, but don't try to impose that kind of incomplete, unsupported environment on *users* If you want to develop an environment, you have a lot of work and documentation to do, so you better get started right now, and not waste so much time reading the net (using code, most probably, written somewhere else on an operating system written somewhere else on a computer you bought as a package (for a lot of money, lest we forget, letting someone else solve our problems costs a lot of money)). toodles..................... -- Ken Shoemaker, Intel, Santa Clara, Ca. {pur-ee,hplabs,ucbvax!amd70,ogcvax!omsvax}!intelca!kds
phipps@fortune.UUCP (Clay Phipps) (03/31/84)
[Note: this was withdrawn soon after it was originally posted, because I thought that most of this had already said; my mail indicates that such is not the case, so I am reposting it.] I would think that IBM PCs would be more than adequate for many undergraduate computing courses, if you had enough of them (PCs). They are certainly adequate for "intro to programming" classes. I doubt that any of the programs assigned in the "intermediate programming" classes that I assisted with (at the Univ. of Maryland in 1978) occupied more than the equivalent of 64KB of memory on the UNIVAC 1100 machines in use at the time. These programs were either punched on cards or typed on DECwriter IIIs, by the way. If faculties still want to teach Pascal as a first language, as I believe the UofMd now does, I should point out that you can get a really huge program into a 128KB IBM PC running the UCSD p-System; I have one interactive program [on my IBM PC] that occupies 2400 lines of source code in over a dozen (15 now ?) separate compilation units, organized consistently with information hiding principles. You could fit a small compiler project into that kind of space. How much code do you expect a student to be able to write for one of many classes in a single semester, anyhow ? Furthermore, you can let students hack around with toy operating systems in assembler and crash only their own machines rather than some expensive multi-user machine like a pdp-11/40 or 45 that has to run stand-alone for such a course (to allow crashes). All you architectural chauvinists should keep in mind that "computer architecture" is often a junior or senior-level course in many schools, so the students won't know the difference between 8088 and 16032, anyhow. The liberal arts people just want to write their papers and maintain modest data files, I suspect. A Z-80 can be used for that, so an IBM PC should suffice. Maybe someone can even write a humane word-processor or editor for them, or scrounge some Macintoshes. Computer science people need to write papers, too. I would bet that most still use typewriters for this; I watched the sun rise over a typewriter more times than I like to remember. Once more powerful and less expensive machines become available, retire the PCs to the student union for use in writing term papers or playing "Zork"-alike games. If we can give undersupported undergraduates an alternative to punch cards and typewriters now, then DO IT NOW ! The freshmen of today may have dropped out in disgust by the time sexy 16032 or 68020 machines become available at the right prices. -- Clay Phipps -- {amd70 cbosgd decwrl!amd70 harpo hplabs!hpda ihnp4 megatest nsc oliveb sri-unix ucbvax!amd70 varian allegra} !fortune!phipps
phipps@fortune.UUCP (Clay Phipps) (04/03/84)
I would think that IBM PCs would be more than adequate for many undergraduate computing courses, if you had enough of them (PCs). They are certainly adequate for "intro to programming" classes. I doubt that any of the programs assigned in the "intermediate programming" classes that I assisted with (at the Univ. of Maryland in 1978) occupied more than the equivalent of 64KB of memory on the UNIVAC 1100 machines in use at the time. These programs were either punched on cards or typed on DECwriter IIIs, by the way. If faculties still want to teach Pascal as a first language, as I believe the UofMd now does, I should point out that you can get a really huge program into a 128KB IBM PC running the UCSD p-System; I have one interactive program that occupies 2400 lines of source code in over a dozen (15 now ?) separate compilation units, organized consistently with information hiding principles. You could fit a small compiler project into that kind of space. How much code do you expect a student to be able to write for one of many classes in a single semester, anyhow ? Furthermore, you can let students hack around with toy operating systems in assembler and crash only their own machines rather than some expensive multi-user machine like a pdp-11/40 or 45 that has to run stand-alone for such a course (to allow crashes). All you architectural chauvinists should keep in mind that "computer architecture" is often a junior or senior-level course in many schools, so the students won't know the difference between 8088 and 16032, anyhow. The liberal arts people just want to write their papers and maintain modest data files, I suspect. A Z-80 can be used for that, so an IBM PC should suffice. Maybe someone can even write a humane word-processor or editor for them, or scrounge some Macintoshes. Computer science people need to write papers, too. I would bet that most still use typewriters for this; I watched the sun rise over a typewriter more times than I like to remember. The faculy will probably be reading better papers as a result of widespread computer usage. Once more powerful and less expensive machines become available, retire the PCs to the student union for use in writing term papers or playing "Zork"-alike games. If we can give undersupported undergraduates an alternative to punch cards and typewriters now, then DO IT NOW ! The freshmen of today may have dropped out in disgust by the time sexy 16032 or 68020 machines become available at the right prices. [Comments, umcp-cs!mark & co. ?] -- Clay Phipps -- {allegra amd70 cbosgd decwrl!amd70 harpo hplabs!hpda ihnp4 megatest nsc oliveb sri-unix ucbvax!amd70 varian} !fortune!phipps