Back to Home Page
IBM-Experienceof Ed Thelen - January through August of 1967
Table of Contents:
- Leaving GE
- WOW - these IBM guys are SHARP!
- Leaving IBM
- Short Stories
- E-mail from Tom VanVleck about Andy Kinslow - Nov 7, 2010
- Epilog of 360 Mod 67
A group of about 10 of us left General Electric Computer Department at the same time - January of 1967.
I was getting worried about the ability of G.E. Computer Department to compete
- they said that the Computer Department was losing $100 million per year -
- and I didn't want to get stuck in low-tech Phoenix if G.E. shut their doors.
(In 1970, GE sold that operation to Honeywell ;-))
My boss (Bob Decker) had related a story of G.E. shutting down their ?Light-bulb? factory in Ohio - click -
Folks had gone home Friday with paychecks, and Monday the doors were locked - the plant closed. My boss was a bouncy guy, and had wound up in Phoenix, G.E. Computer Department, boss of software test and diagnostics, with only a coal mining and accounting background. (One of his tales was that the day he made journeyman coal miner, the tunnel he was in collapsed, and Bob successfully ran for his life. - He decided that coal mining was not for him - never went underground again.)
In any case I was interviewing around, and got several job offers - what glory dayze - hard for me to imagine now -
Then G.E. had decided to ramp up production for the GE-625, hired a lot of people, found life was tough, and just before Christmas, laid-off the new hires - I was irritated -
So a couple of months later, a number of friends decided to leave G.E.
In fact we held the same going away party - about 9 were going to California to set up their own computer related company (timesharing software) and me - I was invited to go along to do a FORTRAN compiler for them - but I also had a job offer from IBM in New York, to help on their TSS (Time Sharing System) effort. Manny Lemas, from G.E., had joined IBM a few months previous, and said the water was fine at IBM - and some good solid work needed to be done. (I also had a job offer from Control Data in Minnesota, but that didn't seem so exciting. What glory days!!)
Being a bit chicken about helping start up a software company, I *finally* (after the going away party) decided to head for IBM's 360 TSS (Time Sharing System) effort in IBM Advanced System Development (Mohansic Labs) in Yorktown Heights, New York.
WOW - these IBM guys are SHARP!
So our family left sunny warm Phoenix, and arrived in the snow banks of Yorktown Heights :-((
Friends - I like to think that I'm as sharp as the average bear. Learned that I wasn't brilliant while quite young, but I can hang in there most of the time. I'm an incrementalist, but can usually move things along.
Friends, those guys at IBM were *SHARP* - it was a thrill to associate with them. They would tell adventures of conning customers into getting extra equipment. My boss's boss (Andy Kinslow, thanks Peter A. Goodwin for reminding me) had recently received a $9,000 cash award (about half of a nice house at the time) for developing a timesharing system to run on a 7094. One of the guys had done the DASDI software for the new disk packs that were just coming out - and all stood in awe of him. I felt that I would really have to hustle to stay up with these guys.
I was put under the wing of a contract employee - This guy was your basic slug - what he was doing in this crowd I never did figure. After a couple of months I was doing the work of both of us, and I was trying to get this turkey fired! My boss said that was not an option - "cope".
We were part of a 45 person group doing performance analysis and providing suggestions for performance enhancements of the TSS (see above) effort that was dragging in schedule and performance.
"Slug" and I were taking instruction traces of the IBM 360 Mod 67 doing an assembly of a few lines of 360 code. Our goal was to suggest groupings of subroutines so that there would be fewer "page faults" requiring access to the swapping drum which cost a lot of "wall clock" time. Basically, try to get more through-put from the machine by positioning the modules of this task in a more optimum manner.
"Slug" and I (actually I) designed and wrote a FORTRAN routine to take the address in the Program Counter and data access addresses from the trace tapes, take the module locations from the loadmap tape, identify which subroutines/modules and data were being accessed, and make a report suggesting position optimizations.
This was a wonderful task, well defined, I could understand it quickly, very doable, and I felt that I could make a definite contribution quickly - and get recognized and get even more interesting job assignments. I even found a bug in the FORTRAN F - I put in lots of comment cards about interfaces and things, and FORTRAN F got lost and did quirky things if there were more than 19 comment cards in a row. Basically, I was having a blast, full access to the documents of the TSS technical reports, code, design documents - the whole tomato :-))
The wife was unhappy with the neighbors, they had been there for years and regarded her as a passing phase. And she wasn't too happy with our rental house (a somewhat winterized summer cabin that I had picked while she was coming from Phoenix). Then she got really unhappy - wanted to go back to sisters and family in Minnesota. Well, I had dragged her all over the U.S. - I guess it was my turn to get dragged.
So I called Control Data and asked if the job offer was still open - They said "Sure - when and where do you want the truck?" No kidding - computer's glory daze :-))
So I busted my butt to get my project all tidied up so that I could resign and not leave a mess. My sleepy looking boss - who didn't seem quite up to the rest of the staff - called me in one day and said that I was doing an outstanding job. I thanked him - then he popped me with this - "Are you getting ready to resign?"
I nearly dropped my teeth - how the $%^& had he figured that out?? All the short timers I knew about sloughed off before leaving. I suddenly figured my boss was a heck of a lot more savvy than I had given him credit for. I told him "yes" - I was going to give 2 weeks notice Friday. We talked for a spell about why - and possible opportunities with IBM in Minnesota. I didn't want to go to the IBM ( ?tape drive? ) operation in Rochester, Minnesota, and I didn't trust the new IBM ad campaign that advertised a free IBM Software Engineer with each new system - that sounded too good to last (and was!!).
So, in a couple of weeks (August 1967) we were moving to Minnesota and wife's sisters and Control Data 6600s :-))
I always looked back at my IBM experience with great fondness
- and that is partly why I'm hanging with some IBM guys restoring an an old IBM 1401.
We arrived in Yorktown Heights. I had heard that IBM had dress code - so before reporting for work, I went down town and bought a blue suit. When I reported for work, I seemed the only one in the building wearing a suit. OK - I wore the suit only on Sundays - good thing - The suit was poorly sewn, most seams failed in a few months !!
IBM was definitely different:
- Trying to play catch-up, I of course worked through many dinners. Being exempt, I of course didn't get overtime, BUT IBM paid a stipend for dinner if you worked through it.
- You put in your time card, with claim for meals, in on Friday afternoon - and received your pay and meal allowance the next Monday. IBM did payroll over the weekend !!! Talk about instant reward !! No wonder the sales force was so aggressive - INSTANT REWARD :-))
I was impressed !!
- One wintery day, I arrived at work - and the building was quite cool - the building heater had failed during the night. But there were electric heaters plugged in and heating under all the desks. Apparently about 3:00 A.M. IBM Building Facilities had waked up local hardware store managers, purchased all the available electric heaters, and had the situation handled as best possible by 8:00 AM. No time was lost - work pressed on!
I was very impressed !!
The 360 TSS - Time Sharing System had 8 or 10 1401s grinding out their lives printing dump tapes - I presume mostly crash dumps :-((
for hours on end.
(The "click" was the magnetic tape drive reading one short record, the "bang" was the printer hammers printing one line.)
I hated walking near them, thinking what a waste of machinery and time.
They should write block tapes and the 1401s should go
So I bugged my boss.
I said I could put a tapeblock back end on the 360 dump and with no 1401 experience figured I could increase the productivity of both the 360 (Mod 67) dump and the 1401s by a factor of maybe 5.
But he figured I had more important fish to fry, and certainly the 1401s were cheeeep (and he was not a boat rocker.)
When I was in BIG computers - GE, IBM, CDC - only IBM seemed to pay attention to the rest of the world.
When I was at the ASDD Mohansic System Laboratory in Yorktown Heights,
we had ( ?Friday? ) weekly all hands meetings of
- what/how we were doing?
- what the competition was doing, several times with corporate competitive analysis staff talking.
Not all was portrayed as wonderful and rosy
- they were quite specific about perceived faults and actions necessary to address the problems.
One of the problems that I thought I saw was not addressed -
- IBM had managers bid on jobs - sometimes with distressing results.
- - - Kind of a lair's poker - bid realistically, and your group had no work -
At the time, the IBM Time Sharing System was not working well
- well, barely processing one user at a time - frequent crashes, horrible performance.
Speaking of liar's poker, Bernie Galler, at the University of Michigan, also with a 360 Mod 67, was claiming good results with 4 concurrent users. We had no idea how such claims could be valid - he was most likely running our code - like what else?
However, when "we" were asked by outsiders how things were going
- "we" responded "Ask Bernie Galler at Michigan".
Decades later I found that Michigan was running different (their own) code for a time sharing system.
E-mail from Tom VanVleck about Andy Kinslow - Nov 7, 2010
I have heard of Andy Kinslow; he wrote a paper called "The Time-Sharing Monitor System" that was in the 1964 SJCC. I will ask Corby [ Fernando J. Corbató ] if Kinslow contributed to CTSS.
My impression was that he did not, and that the Kinslow system was an internal IBM demo project that was ditched in favor of the 360 focus, that essentially did the same thing as CTSS with additional hardware help. CTSS was demo'd in 1961 and had the essential hardware RPQ in late 1962 or early 63.
The paper is online, in the ACM digital library, costs money, I have not bought it.
Google search for Andy finds that he died in 2008.
Epilog of 360 Mod 67 TSS - added 2015
The TSS effort stayed sick (inspite of my brief best efforts).
wikipedia.org TSS operating_system says in Feb 1968, IBM decommitted from TSS, relented in 1969, and terminated the effort in 1971.
Bernie Galler of the University of Michigan says in this oral history - page 24 - (local copy) that
- The 360 mod 67 was in response to design requests from Michigan, different from MIT's design requests.
- IBM's first TSS decommittment was Jan 1967 - I think Bernie was a year off.
If you have comments or suggestions, Send e-mail to Ed Thelen
Back to Home Page