Minutes of the SIPB Meeting of 23 June 1980 The meeting was called to order at 19:30 by William M. York. In attendance were ZRM, RK, WMY, RMP, J9, ACW, GMP, JSL, c|, ALAN, RDM. Late were: SEL, DGHP, Naha, BEE, HGA, and RLL. The minutes of the meeting of 16 June were read and accepted as corrected. *** USER INTERRUPT *** * * Marc K. Freedman wants $300 to maintain a data base * for the Macroengineering Research Group. He will * write an RDMS front end (learning PL/1 in the * process). He is being paid by UROP, and has a note * from them to the effect that they cannot pay for * computer time. $100 was autoed already, and we will * defer further allocation until we find out how he's * doing with the first hundred. * *** TPURRETNI RESU *** Treasurer's Report: We've spent roughly $7000 so far this month, according to system mail from the absent treasurer. Chairman's Report: The chairman continued to try unsuccessfully to reach Jean Bonney about the Center's policies concerning non-Multics use of the Spinwriter. The chairman spoke to Norma McGavern from the UROP office, about relations between UROP and the SIPB. UROP is no longer funding students for computer time. Bill Roberts talked to Wes' secretary about whether the Center could back him in his efforts to get a room to teach COBOL to CETA people in. RK suggested to him that he speak to his supervisor, and maybe his supervisor would vouch for him. Another possibility would be to go to ESP, and they could back him. ALso, the SIPB could write a letter to Scheduling and try to unwedge things. Office Report: There is still an office, in case we didn't know. We found out who CWG(?)L is. RLL put the Commands and Active Functions out of user's sight. Censure. Computer Services: The line printer, card reader/punch, and printer controller for the new Multics configuration arrived, but were not signed for, so they were taken away again. They will be brought back tomorrow. They are the first scraps we have seen of the new hardware. We are told that the rest has been shipped. It is really coming, really. We estimate installation within a week of delivery, in the optimal case. We don't want to think about the pessimal case. All 370 applications have been being taken automatically to the meeting. It was moved and seconded that this not be the case, and that 370 applications be made autoable. This would be done in order to help users who must use the 370, and who at present must wait until Monday to get started. The auto limit would be $200, just like on Multics. Signatures in blood would still be extorted from our 370 users. The motion passed. Should SIPB Secede and Get Their Own Computer?: Should the SIPB work toward the goal of acquiring its own computer, thus breaking their dependence on the IPS? This question is timely because it is rumored that Honeywell will soon announce moderately priced Multics mainframes that may very well fall within our (extended) financial capabilities. This is still a probabilistic question, since Honeywell seems nowhere near making a final decision. A system has been announced, the DPS 8/20, manufactured in Japan, that will run GCOS. The question is whether Honeywell will produce the mythical DPS 8/20 M, which will run Multics. A DPS 8 processor, 1MW memory, a 501 disk, with 2 spindles, holding 250K Multics records, 2 tape drives (800/1600 BPI), and a FNP with 32 asynchronous lines minus modems --- all this would cost about $556,133. This is GMP's calculation, and is an optimistic estimate. Maintenance would be $2336 per month. This would be a GCOS system, and is only an approximation to the cost of the hypothetical Multics system. Alternatively we could get a 3 year lease at $15883 / month. We should remember that the capacity of this system would only be a fifth to a fourth of the capacity of the Multics we are used to. This means only fifteen simultaneous users. An immense amount of this cost is the front end, at $160K. This cost could perhaps be saved by using an 11-based Palterplexer. This would reduce the cost for the entire system to $395,723, maintenance to $1482 / month, leasing on a 3 year basis $11,235 / month. We are MIT. We can get hardware at GSA prices, and we don't know what percentage savings this means. But on top of this we save another 13% as part of the terms of the Honeywell / MIT contract. We may be able to save more by making special arrangements with Honeywell. So, one question is, is this system worth the money? Do we get enough computrons to the dollar to make it worth consideration? (SEL and DGHP came in during this explanation from GMP.) We should remember that the DPS 8/20 cannot be configured with more than 1M of memory. Also, it can only be upgraded to the 8/44 level, which is only a 50% enhancement, at which point you hit a dead end as far as enhancement goes. As far as software prices go, you get the system tape with your processor. This is a completely bare system, with no user commands. The next thing you get is the executable library (>sss). These are all of the ordinary commands. These would cost a four-digit figure per month. We don't know what this figure would be, as Honeywell itself has not yet decided. The software must be leased and cannot be bought. Then comes unbundled or separately priced software. This includes Fortran, Mail, Emacs, etc., as well as the RJE software which we use presently to run the Spinwriter. We hope that we could piggyback on some comprehensive software contract that MIT already has with Honeywell. We are not, however, familiar with the terms of these agreements. If we cannot do this, we can expect to pay on the order of $10K to $20K per month. When you buy the software, you get the source. It should be mentioned here that there are other possibilities in terms of what hardware we should get. We should investigate these other possibilities in detail. These possibilities include VAX/VMS, VAX/UNIX, 11/70/UNIX, 20/TWENEX, Primus, Data General Eagle, and any other systems that can run 40 users maximum for $0.5M list price or less. IBM should be given a chance to speak as well. MTS is an outside possibility. The two questions we need to ask of manufacturers are, "What would it cost us to run 40 users?", and "How many users can we run for $500K?". One thing to keep in mind is that the Committee for the Future seemed to say that the SIPB would get a computer in the Future, and we want some control over which. Issues over whether or not we want a computer at all will be deferred to a future meeting, since we have already spent 45 minutes discussing this. The organization is congratulated on its restraint. Naha and BEE came in at the end of this discussion. Telecommunications Report: c| talked to Sylvia Something at the Student Center Library, and told her that we really do want the terminal room carpeted. There is storage space available at the Library for us to keep our equipment during the carpeting. JSL talked to Leo Ryan about moving our terminals into the IPC terminal room for the duration of the carpeting. Leo Ryan doesn't know as usual. We have not found out about the modifications to the ADM 3A. Dave Moon is the person who knows this sort of thing. Publications Report: J9 will read the new Emacs handout and suggest modifications to Bernie. Dormitory terminal-room documentation should be compared with the ideal list on the blackboard and be augmented to the extent that this is needed, by buying manuals for the places that need them. This needs to be done this week so that we can buy manuals on this year's budget. Get the revision number of the manual and a list of addenda, when you make the documentation inspection. We will certainly need MPM commands manuals for each dormitory (8 of them). GMP will order them. R/O Report: HGA has yet to start making arrangements for the tour. The Freshman letter must be printed and sent to the right people this week. ACW will do it. Naha has to write a letter to MIT Telecommunications Office for two telephones for the Midway. Other: IBM entered the word processing market. Their entry has a terminal, two floppy drives, a 50K word dictionary, and a Selectric printer, with communications capabilities to IBM mainframes. User interface is a complete unknown. List price is -- get this -- $8000. Phil Earnhardt would like to send us some Coop money to maintain contact. But maybe we could save him the money by putting him on the SIPB project. The consensus seems to be that he would be an asset to SIPB project and should be added to it. RDM will be running some benchmarks on Multics. RLL came in at this point. Applications: The Graduate Student Council wants $100 to continue to run their accounting program on Multics. They have a budget of $30K per year. Deferred, with the strong suggestion that they find an alternate source of funding, or at least send a representative to our next meeting. Robert A. Proctor wants $100 to type up a paper on some combinatorial aspects of Bruhat orders. He has had $300 before for typing math papers. Given. Attention: Alan Strong has used none of the $100 allocated him for hacking thesis. Ali Ahmadi had $80 added to his Coop account. Jane Martini had $75 added to her Coop project. Total appropriations: $1205 without Coop Autoed: $1105 Student: $1005 370: $100 To Meeting: $100, all Student. Coop: $155 Concourse will be allowed to buy a tape on our account, for use with their recently acquired 11. Other Other: Be careful when you lock your bike. Moon and Penny locked their bikes to a signpost and had their rear wheels stolen. DEC installed their first 11/44 at LL-ASG last week. It runs. J9's car got its inspection sticker. RK is looking for a part for his car -- a hydraulic frobulator pulley main drive belt type thingie frob. Any leads will be welcome. The Concourse/LOGO 11/45 is doing all right. It is running RT11 as an interim measure. They hope to be bringing up the TV system shortly. If anyone knows of a company named Per Data (Peripheral Data), pointers will be useful, as the 11 has a tape drive from them, and no one can make it work. WMY, SEL, and RM and DGH P have moved into their summer accommodations. An all-night hacking session for a last-ditch annual report will be held Saturday night and will last forever. RDM had a run-in with a taxi. The meeting was adjourned at 2215. Minutes taken and submitted by Allan C. Wechsler ;; Local Modes: ;; Mode:TEXT ;; LISP ) Hack:1 ;; Auto Fill Mode: ;; Fill Column:70 ;; Fill Prefix: ;; END: