MIT’s digital computers are saturated until the IBM 360 becomes operational

See listDEC 22 196666-FM1-191FM/Deputy Chief, Mission Planning and Analysis DivisionMIT’s digital computers are saturated until the IBM 360 becomes operational

I guess I ought to record the saga of the MIT 360 computer, if only so that it may take it's proper place in history. It is a little adventure which has been going on in the shadow of the more dramatic crises at MIT and is now rising to the surface in it's own right. Although it looked as though there was not much we could do about it except watch in fasci- nated horror, we have taken some steps which I hope will be effective and which might interest you.

As far as I can tell, somewhat less than a year ago MIT foresaw that the two Minneapolis Honeywell 1800 digital computers they are using for space- craft computer program development would be inadequate at some time in the future. In the absence of good programming plans, it's hard to know how they either predicted this would be the case or were able to convince any- one of it. In fact, I guess they did have a bit of trouble since it wasn't until June that they were finally given the go-ahead to procure an IBM 360 by MSC. Installation of this computer has proceeded, along with training of MIT personnel to use it. IBM contracted to supply some important spe- cial programs, which they have apparently delivered on schedule, and MIT has prepared some others to permit use of this facility. But the one which now appears to be the most critical of all is the MAC compiler – a complicated service program which translates symbolic programs into computer language. MAC is the MIT equivalent of IBM's Fortran, except that it is said to be superior. The development of this program was contracted to Hankins – a Boston outfit employing between 40 to 80 people, depending on who tells you. The compiler was to cost about $170,000 and was to be delivered January 15th with a $200 a day penalty clause in the contract for late delivery; Hankins refused an incentive reward for earlier delivery. I have no idea how the January 15 date was selected. But I am certain of one thing – it is about two months too late. The fact is the AS-206 pro- gram development by itself has now saturated the two 1800 digital computers and it's evident the situation is going to become much worse before the IBM 360 becomes operational. Debugging and integration of the AS-207 and AS-208 programs has just begun, and this activity will impose a very heavy load on the digital computer facilities very soon.

Hankins is not going to deliver on schedule. They Estimate a slip of about approximately 6 weeks. Of course, delivery of the MAC compiler around the 1st of March does not by any means guarantee operational capability at that time – it will probably take at least another month to get the system working. Unquestionably the lack of adequate digital computer facilities at MIT during the first months of next year is going to impact delivery of the AS-207/208 programs badly, which is going to be extremely painful to bear.

What can be done about this? Well, MIT has obtained as much programming assistance for Hankins as is worthwhile in an attempt to prevent his delivery schedule from slipping even further, although I sure wouldn't count on it. We are attempting to identify all tasks that could be transferred from MIT to our CDC 3600 simulation facility at MSC. We have also offered to supply our 3600 software packages to MIT when completed about the first of the year for their use on whatever 3600 systems are available to them in the Boston area. These steps may help some, but do not really solve the basic Problem – namely what has happened at Hankins. In spite of recent discussions with them emphasizing the importance of this pacing item, it becomes apparent that management of that organization is immature and unstable – completely unreliable. MIT and MSC have agreed that to maintain an approach depending solely upon Hankins is unacceptable. In our joint search for some way to work around this problem, it has been concluded that the best “backup” course of action is to replace the MIT developed MAC programming system with the IBM Fortran which, of course, exists right now and is said to be working well on the 360. This makes it necessary to completely recode the simulation programs, those defining the environment, spacecraft systems, world earth models, etc., into Fortran. However, since these two compiling systems are so similar, this conversion is not expected to be a very difficult or complex task although it will be time consuming. These programs are very large. Since the CDC 3600 facility here at MSC will also use these MIT simulation programs, we have been routinely outlining and documenting them. It's our intention to supply this material to MIT for the use of their subcontractor, probably IBM, who will be given the task of this conversion. It is anticipated that parts of this simulation will be available for use during the latter part of January and the complete package should be finished in less than two months. Thus it is evident that this is a quicker approach for getting the IBM 360 on- line even if you believe the new Hankins delivery schedule. But I must say I have no faith whatsoever in that organization in spite of MIT's valiant efforts to support them.

I am very pleased to report that key personnel at MIT have given enthusi- astic and wholehearted support to this plan. They have worked very hard to solve the problems and to get IBM on-board and working promptly and efficiently. I understand that if the Fortran approach succeeds as we anticipate it will, MIT will probably not maintain the simulation programs in MAC language for the 360 but will convert over completely to Fortran. Of course, they will have to carry along the MAC simulation system for the 1800 computers as long as we continue to use them at MIT. My personal opinion is that we will still be using them at the end of next year.

Aside from cutting back on the programs and thus reducing the work re- quired, relieving the overloaded digital computer facilities at MIT seems to me to be the only course of action available to improve the delivery schedule for the AS-205 and AS-208 programs. Although it has taken some time to arrive at this solution, I believe it to be the best we can do unless you want to pray that the flight schedule will slip.