BillW@Sri-Kl.ARPA (03/23/84)
From: William "Chops" Westfield <BillW@Sri-Kl.ARPA> I wasnt going to get into this argument, but I give up. Here goes: 1) What is "State of the Art" "SotA" does NOT mean 'current exciting topics in computer research' In this case, the "art" is PERSONAL computers, and "SotA" is definately 8086/68000/16032 architectures. You can get rid of the 16032 if you make the art "low cost/student affordable personal computers", and the only reason the 68K is still there is becasue of MacIntosh. 2) Remember that there are universitys out there whose current "computing environment" is horendous. My freshman programming course (fortran) was done on cards (the "honors" class I took (PL/1) got to use terminals!). That was in 77. Things are a little better at U of Penn now. Note that this is a relatively large, expensive, Ivy league school. I hate to think where smaller shools might be. (not every place is a Stanford or berkeley that gets nice fat discounts because Computer manufacturers tend to get nice things back from them...) 3) Computerizing the university does not just mean giving personal computers to computer science or even engineering students/faculty! I know Stanford gave a bunch of IBM PCs to HUMANITIES type people, and Penn has Apollo workstations for linguistics, and there are all these poor economics and other business people whose idea of a computer course is "how to use the linear regression package". None of these CARE whether their microprocessor chip is elegant or not. They only see the final applications. This is really the kind of person MacIntosh was designed for; the CS people will get them first so that they can write the linear regression package, the Mac(intosh)symna (hmm.), the circuit simulators, and so on. And my god! Think of all those people who still lug typewriters to school with them... We all live in an ivory and silicon tower. We know and understand and like computers. We dont have to worry about running up to many charges, and not being able to finish our final assignments (at least to the point of spending all this time sending silly messages back and forth!). We think the joys of using a computer are related to networking and graphics and parallel processors and RISC and GaAs - We forget the simple joy of having to type ONLY the corrections to the latest term paper. Many places, giving the students ANY kind of personal computer power will GREATLY improve the situation (eg: Syracuse used to give ALL students time on a DEC10, to use as they wished... This much "Personal" computing power wasnt available at Penn until several years later, and then only to engineering students, and other places probably NOTHING exists...) Sigh. BillW
ron@Brl-Tgr.ARPA (03/23/84)
From: Ron Natalie <ron@Brl-Tgr.ARPA> Yes, what you describe is what we refer to as the Library approach to computer science. It's how it works here at the Laboratory. Everyone gets a terminal and enough computer time to keep them happy. Johns Hopkins University used to be like that. Used to be that just by asking the right people you could get as much free time as you wanted on either the DEC-10 (TOPS-10) or an 11/45 (UNIX) depending on your preferences. Hopkins produced some really nice software for both machines outside of any of the normal (and scarce) computer science curriculum. Lots of us became the UNIX gurus this way (and we put out a few TOPS-10 gurus as well). Unfortunately a policy change several years ago essentially locked up the academic computer resources by attaching price tags to everything and abolishing the gratis "I want to learn about the computer" accounts. Too bad, the interesting advancements Hopkins was making has all but died out and they aren't graduating too many of the shining star computer gurus anymore. There latest attempt to make up for this was to can plans to network the university together and launch into doing everything with a bunch of PC350's that DEC unloaded on them at bargain prices. They are now beginning to realize that theyare having a distribution problem with a hundred micros. -Ron
BILLW@Sri-Kl.ARPA (03/27/84)
This is starting to make a lot of sense. Consider a university that currently has no free computing cycles for general "free" use by students for other than specific computer oriented courses (lets see: document preperation, communications, recreation, calculator type things, etc (I remeber in my electon optics class, the prof told us to go off and describe the characteristic of a Pierce electron gun, and that since CAD was big, why didnt we all use a computer to do it? This was an EE class, where the only required computing experience was a intro fortran class. Sigh. I went off and used macsyma. Other lucky students went downstairs and used the CP/M systems they had been taught in another class. What a fiasco!)). Now then. Lets consider the foolowing two alternatives, each to support 1000 students: DECSystem 2060, running tops20 or tops10. This may actually support up to around 5000 students, depending on how the load is distributed. SRI-KL has about 1000 users that do just these kind of things, and it is noticable slow durring prime time, but empty durring grave shift. Wharton (Tops10), had about 5000 accounts, and was impossible durring the day. It all depends on memory, disk, etc, etc. TOPS20 will handle about 60 users at once, tops10 perhaps 80. You could also use 2 vaxen (30 users), I think the prices will work out about the same... Hardware costs (CPU, MEMORY, DISK, 96 terminal/dialup modems) ~500,000 Systems programmer. 20,000/year maintanance. 20,000/year operations staff. min wage*3shifts ------ ------- total a lot! (these are typical off the wall estimate by a hacker that prefers not to think about such things.) ============================================================== option 2: 1000 Macintoshs. Hardware costs (100 systems for CS dept, student user room, software development, etc.) 100,000 liason/sales person 30,000/year. ------ ----- total A lot less. Note that by selling the Macs for slighly over their cost, a universtity can pay for the above expenses.... Of course, this requires the students to shell out ~1K each, but it saves them about 1K over buying their own systems...) BillW
ron@Brl-Tgr.ARPA (03/28/84)
From: Ron Natalie <ron@Brl-Tgr.ARPA> C'mon, let's be realistic with your cost analysis (although I feel that some universities are not). Even if you buy (or are given) a hundred micros someone does end up supporting them if things are going to work. In addition you are going to have to deal with the shopping cart full of floppy problem of data transfer or you will need some kind of networking strategy. Balancing it all out it would seem much more convenient not to use the cheapest computers possible. While, I think it is a good point to have students have their own personal computers for scratch homework computations, text processing, and maybe a little computer literacy of their own, I don't feel that this takes the place of having (excuse me) a real computer for these students to learn on. I am currently planning on micro systems for my brother and sister to take with them to college. One is entering law school and needs word processing more than anything else, and the other is starting undergraduate liberal arts studies and would need a more diverse mixture. The exposure, when you are using your little personal machine is minimal. There is no interaction with the academic community. What is there to be learned, other than being an expert on the particular MICRO that you happen to have purchased? A group of supported micros networked is better, but I really don't think the technology is that far along. The software base on a midsized (VAX) academic computer is typically more diverse than a micro can afford to support. Our machine, for instance, had LISP, APL, FORTRAN, PASCAL, and C languages, all different, all important, and all necessary for their particular applications. We have software from many inexpensive sources that gives a diversity which a single user just couldn't assemble on his own. My initial statement (about library style computing) was a comment that universities are tending to cut off the free computer access to students that many used to have. This is partly because university administrations are now jealously considering computer time as a university resource to be charged for, etc...when just the opposite should be true. The access should be increasing because of the declining price of the technology. Now that they have seen that computers are a good thing, and not some academic toy, they are getting very stingy. -Ron
jbf@ccieng5.UUCP (Jens Bernhard Fiederer) (03/30/84)
While I am no anti-PC flamer, the statement that 8088-8086 (etc.) is state of the art, and the only reason 68K is STILL there is because of the Macintosh, is confused. The 68K is state of the art, and the only reason the 8088 still can be used in the same paragraph without too many giggles is because of the IBM-PC. The 16K is somewhat more advanced than the state of the art(in micros). jbf -- Reachable as ....allegra![rayssd,rlgvax]!ccieng5!jbf Or just address to 'native of the night' and trust in the forces of evil.
cem@intelca.UUCP (Chuck McManis) (04/03/84)
I agree with Ron on most parts of his message, especially about Universities finding out that not only will students pay to use those non-prime time hours but will out bid other students for accounts. When the university has a 30XX that is bored stiff at 9:00 at night because it just finished expending .000000001 CPU seconds in figuring out the days accounting info, should be given something to do. Also not a few of these students develop useful utilities that often become essential parts of the system. (This I believe was the original motivation of Hackers, ie to emulate the gods, Systems Programmers.) On the continuing flame of mainframes vs minis vs micros. Network technology and microcomputers are becoming sufficiently advanced that you can run programs through the network as though they were on your own disk, send mail to other users and basically never know there was a network. Eventually someone will write an executive that will make n micros on a 10Mbit network look like 1 virtual machine with n terminals. It will however be very fault tolerant and should exhibit only marginal degredation as the number of users increases. Also on the network should be various "export" processors that are adept at particular functions, like array processors, or extra high speed data crunchers. Then there may be no way to tell if you are on a micro or mini. Lets propose a standard of computing like the original standard for AI, if when you use a micro you can't tell whether you are using a mainframe micro, or mini. Then for all intents and purposes you are using a mainframe. (highest common denominator) --Chuck
mason@utcsrgv.UUCP (Dave Mason) (04/05/84)
Ryerson Polytechnical Institute is trying to do exactly what ron@Brl-Tgr
suggests: treating micros as a library resource. Every student in the
institute has paid an extra $25 for the last 2 years to buy a microcomputer
facility which is in the library. We have 60 IBM-PCs, 25 Apples, 25 CP/M
machines (Zenith-89s and NEC-8000s) more or less connected to 2 20 meg
winchesters over Omninet (less right now, more when I finish the network
install) with our own network software.
This having been paid for out of student funds other than tuition is
not allowed to be used for teaching, but is a student resource. Students
use it for word processing, computer literacy, learning to program,
game playing. We had our doubts in the first year, but it seems to be
well used now. I can answer technical questions, and can direct anyone
who is interested in administration type questions to the proper person.
This is one type of computer resource that I feel is needed. Here the
need is UNQUESTIONABLY for existing software, and note that we have
the machines that support 90+% of the available software. This is a
very important kind of resource to have available for non-computer types
(as well as the computer types). At Ryerson we have programs including
Fashion, Photo Arts, Radio-Television, Theatre, Business, Engineering,
as well as Business Computer and Applied Computer Science.
Only about 10% of the students are in one of those last 2 programs.
HOWEVER, that is not what is required in a Computer Science program.
I have already expressed my opinions on the 8086 vs. 68000 vs. 16032
question, so suffice it to say that I think computer science students
need computer systems available with sufficient power and elegance that
they can attack and solve interesting, real, problems, and operating
systems and development software that make that possible. It is
difficult (though perhaps not impossible) to do that on processors
with archaic architectures.
--
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-Relaymp@ganehd.UUCP (Scott Barman @ Univ. of Ga.) (04/08/84)
>I agree with Ron on most parts of his message, especially about Universities >finding out that not only will students pay to use those non-prime time hours >but will out bid other students for accounts. When the university has a 30XX >that is bored stiff at 9:00 at night because it just finished expending >.000000001 CPU seconds in figuring out the days accounting info, should be >given something to do. Also not a few of these students develop useful >utilities that often become essential parts of the system. (This I believe >was the original motivation of Hackers, ie to emulate the gods, Systems >Programmers.) This would be great if the University's computer center has their collective acts together! At the University of Georgia, they are charging over $700/cpu hour for an IBM 3081. When you come down to it, how many of these "hackers" would go broke trying to hack? Very many, and quick! There are other problems that all boil down to that the computer centers must be willing to do this. From my understanding (from what I have seen), they shun the idea of students running rampant on the same system as all the accounting/billing/scheduling is done. I agree, it would be a great idea (along with the others mentioned), however, I do not believe in charging the students to use a university resource. This is like the library, it should be included with tuition! Scott Barman ..!akgua!ganehd!mp
seaburg@uiucdcs.UUCP (seaburg ) (04/08/84)
#R:sri-arpa:-23200:uiucdcs:10400127:000:307 uiucdcs!seaburg Apr 7 23:53:00 1984 We at the U of Illinois have our Cyber 174 facility available for any student to use, during low usage hours of course. However, the system ain't real friendly, and most students don't even know about it. Perhaps someday we'll get something the average Liberal Arts (sorry) person won't be afraid to use.
ron@Brl-Tgr.ARPA (04/09/84)
From: Ron Natalie <ron@Brl-Tgr.ARPA> While what Ryerson is doing is commendable, I still believe that basing a computer science education on micros is not yet a good idea for reasons stated in other letters to the net. -Ron s