HP Forum Archive 20

[ Return to Index | Top of Index ]

IBM 1620 computer Message #1 Posted by Don Shepherd on 18 Jan 2012, 8:13 p.m. I recently got a book on this scientific computer from the early 1960's. Supposedly, it couldn't add; it used lookup tables to perform addition. At one point it had the code name of "Cadet" which, humorously (I think) stood for: can't add, doesn't even try. Did anyone here ever program or use this computer, and what was it like?

Re: IBM 1620 computer Message #2 Posted by Gerson W. Barbosa on 18 Jan 2012, 8:28 p.m., in response to message #1 by Don Shepherd Ah, but it was soon followed by a far more advanced model: "The 1620 Model II introduced basic ALU hardware for addition and subtraction (making "Can't Add, Doesn't Even Try" no longer applicable) and index registers." http://en.wikipedia.org/wiki/IBM_1620

Re: IBM 1620 computer Message #3 Posted by Don Shepherd on 18 Jan 2012, 8:43 p.m., in response to message #2 by Gerson W. Barbosa Thanks Gerson, but I can't see that page for 3 more hours! I really don't like this 24-hour blackout.

Re: IBM 1620 computer Message #4 Posted by Paul Dale on 18 Jan 2012, 8:45 p.m., in response to message #3 by Don Shepherd Replace the "en" with "m" in the URL :-) Or turn off style sheets and load the page. - Pauli Edited: 18 Jan 2012, 9:09 p.m. after one or more responses were posted

Re: IBM 1620 computer Message #5 Posted by Don Shepherd on 18 Jan 2012, 8:51 p.m., in response to message #4 by Paul Dale Thanks Pauli.

Re: IBM 1620 computer Message #6 Posted by Gerson W. Barbosa on 18 Jan 2012, 9:11 p.m., in response to message #5 by Don Shepherd Don, Notice the Taylor series for sin(x) in the fourth picture (IBM 1620 Model I Level A (prototype), as it appeared in the IBM announcement of the machine.) Regards, Gerson.

Re: IBM 1620 computer Message #7 Posted by Don Shepherd on 19 Jan 2012, 5:10 a.m., in response to message #6 by Gerson W. Barbosa Yeah, I see that. That poor guy trying to thread his paper tape is interesting. Magnetic tape must have seemed like a huge breakthrough!

Re: IBM 1620 computer Message #8 Posted by Dave Shaffer (Arizona) on 18 Jan 2012, 9:51 p.m., in response to message #1 by Don Shepherd Our local college (Juniata College; Huntingdon, Pa.) got one of these in 1963. I was a Senior in high school then and took a calculus course at Juniata (everybody does that these days, or their high school has calculus, but in 1963 the highest high school math was algebra 2, and it helped that your Dad (who 10 years later introduced me to the HP35) was a part-time physics/engineering instructor got you in the door at the college). I was the "curve buster" in that calculus class - on the first exam I got a grade 30 per cent higher than any of the college students! Part of the course was to program the 1620. As I recall, the FORTRAN compiler was a deck of cards - you read it in first, and then read in your FORTRAN deck. A bunch of us was sure that we had programmed things OK, but we kept getting error messages. Turned out, there was a bad card in the compiler. (bugs have been around forever!!) It was S-s-l-l-o-o-w-w! One of the assignments was to calculate pi using various approximation integration over a quarter of a circle (i.e. a lot of square roots). Various members of the class had a different number of slices to use in the approximation (I had inside trapezoidal, some had box approximations, either inside or outside the circle). I had something like 2000 slices, and a run took about 15 minutes of CPU time (you were the only one using the CPU!). My recollection is that I was good to 3 or 4 significant digits. I think any one of our HP calculators would do this now in a second or less!

Re: IBM 1620 computer Message #9 Posted by Steve Leibson on 18 Jan 2012, 10:18 p.m., in response to message #1 by Don Shepherd One of these 1620s has been restored to operation at the Computer History Museum in Mountain View, CA. The magnetic core memory had deteriorated from solder flux and was neither operable nor replaceable. So the restoration team got the smallest semiconductor SRAM they could buy, interfaced it to the 1620, and used a small fraction of the capacity of the chip. Before the overall museum redo about a year ago, you could watch a live demo of the 1620 in operation. I'm not sure they do that any more. And yes, the 1620 did indeed use lookup tables rather than an ALU.

Re: IBM 1620 computer Message #10 Posted by Don Shepherd on 19 Jan 2012, 5:22 a.m., in response to message #9 by Steve Leibson Thanks Dave and Steve, very interesting stuff. If I ever travel to California I'll have to see that museum. So, there was a bad card in the Fortran compiler card deck you had to read in each time. It's amazing how far we have come in this industry. I graduated from high school in 1968, and no schools around here (Louisville KY) had any type of computers then; the closest they got was big old mechanical accounting machines. Upon graduation I enrolled in ECPI, Electronic Computer Programming Institute, and learned IBM 360 assembler, RPG, and COBOL. ECPI couldn't afford the monthly rent on an IBM 360, so our instructor took our card decks to his day job, programmer at Blue Cross Blue Shield, and compiled our work. I was so proud of my first printout from that machine, it said "no diagnostics". I asked the instructor what that meant, and he said "it means you had no errors, congratulations". Great memories. Edited: 19 Jan 2012, 5:24 a.m.

Re: IBM 1620 computer Message #11 Posted by Eric Smith on 19 Jan 2012, 5:31 p.m., in response to message #10 by Don Shepherd Well, it means that you had no detected errors. :-)

Re: IBM 1620 computer Message #12 Posted by Don Shepherd on 19 Jan 2012, 6:12 p.m., in response to message #11 by Eric Smith What? How could my program contain an error if it got a clean compile? : ) I have a nagging feeling that more than a few members of the current crop of programmers may believe that today.

Re: IBM 1620 computer Message #13 Posted by Mike Morrow on 19 Jan 2012, 9:32 p.m., in response to message #10 by Don Shepherd Quote: I enrolled in ECPI, Electronic Computer Programming Institute, and learned IBM 360 assembler, RPG, and COBOL. What about JCL??? You couldn't do much on the System 360/370 without some familiarity with Job Control Language, which for some reason almost everybody hated. There were bumper stickers available that read "Honk If You Love JCL". Edited: 20 Jan 2012, 4:07 p.m. after one or more responses were posted

Re: IBM 1620 computer Message #14 Posted by Don Shepherd on 19 Jan 2012, 10:03 p.m., in response to message #13 by Mike Morrow Mike, I don't remember learning a lot about JCL from that early course at ECPI. In looking back at my old notes (which amazingly I have after 44 years!), this is about the only thing I could find. As I recall, our instructor took our card decks to his Blue Cross Blue Shield office and ran an assembler run to produce our listings one time, but that was probably all. He undoubtedly had to get permission from his boss to do that. I'm sure those simple assemblies didn't eat up very much CPU time. I never really worked on any 360 or 370 system in a job, but I did take a course at the community college in the Washington DC area on JCL specifically, years later. I much preferred Univac's Runstream to IBM's JCL, as I think anyone would.

Re: IBM 1620 computer Message #15 Posted by Mike Morrow on 20 Jan 2012, 4:51 p.m., in response to message #14 by Don Shepherd Quote: I much preferred Univac's Runstream to IBM's JCL, as I think anyone would. I very much agree, Don. Forty years ago I spent many enjoyable hours with the Georgia Tech (GT) Univac 1108 system with Exec 8 Level 27 as the OS. All of my programming that mattered was on the 1108, using some ASR-33 demand access and a lot of batch runs using hollerith cards punched on IBM 026 or 029 punch machines. I did limited programming in Algol on the GT Burroughs B5500 with its TSMCP OS. It was a very intriguing stack machine mainframe that documentation described as "Polish Notation" at the machine code level (which users did not utilize). It had a push-down stack with elements arranged FIFO from the top of stack. That is pretty much really RPN from the early 1960s, long before any HP calculator used it. GT had an IBM System/360 for administrative use, so there was no access to that. I had my first JCL experience on the University of Arkansas System/370 in 1980. JCL was entertaining to play with, with the proper references, but I never found any benefit to its complexity compared to the Univac 1108 Exec 8 environment. I loved Exec 8 (although Level 31 which replaced Level 27 at GT seemed a bit squirrelly to me). The era of student mainframe use and the early days of electronic calculators...now those were the good old days. :-)

Re: IBM 1620 computer Message #16 Posted by Marc Ferrer (France) on 19 Jan 2012, 10:34 a.m., in response to message #9 by Steve Leibson Hello, Having read that the 1620 uses lookup tables in its ALU makes me think about HP stuff : the 9810/9820/9830 CPU has 2 256x4 bipolar ROMs dedicated to BCD arithmetic (from US patent 3,839,630)... quite a similar system. Regards, Marc

Re: IBM 1620 computer Message #17 Posted by Eric Smith on 20 Jan 2012, 2:40 a.m., in response to message #16 by Marc Ferrer (France) Yes, but on the 1620 the table was in core, so you could easily change it. For instance, if you wanted to do a bunch of octal arithmetic. Of course, the next user would be really upset if you didn't restore it to the normal table.

Re: IBM 1620 computer Message #18 Posted by Marc Ferrer (France) on 20 Jan 2012, 5:26 a.m., in response to message #17 by Eric Smith Very interesting... it must have been a delight for some facetious programmers, on April 1st :-)

[ Return to Index | Top of Index ]

Go back to the main exhibit hall