IBM 1401 compiles and runs FORTRAN II

[jimdempseyatthecove]

Next time you complain about your program compile times take a look a this:

https://www.youtube.com/watch?v=uFQ3sajIdaM

FWIW the first Fortran programs I wrote were in Fortran II, I also wrote in Fortran-D, these were run on a Digital Equipment Corporation PDP8-I and PDP8-L minicomputers. A year later (I think) we updated to Fortran IV. The PDP8-I had card reader, the PDP8-L had paper tape (10cps).

Jim Dempsey

[jimdempseyatthecove]

Historical note.

Before I get flamed over the use of FORTRAN II as opposed to Fortran II please bear in mind that the 1401 (and PDP8-I/L) alphanumeric character set with 6 bits. IOW no lower case. The PDP8 series was a 12 bit CPU and could represent two characters per word or three 8-bit characters in a pair of words. Early programming used 6-bit, later programming used 6 or 8 bit.

Jim Dempsey

[Steve_Lionel]

Anyone who flames you over "FORTRAN II" is wrong.

When I was in college one of my projects was to write a 1401 emulator in IBM 360 assembler, and to have it run a program. Word marks! Decimal arithmetic! Fun!

[andrew_4619]

Great video! thanks :-)

[rase]

When I started my programming career on an IBM 1401 (machine instructions using SPS and Autocoder) a formula translator was available for that computer. It was called Mathcoder and created by staff members of IBM Germany. Colleagues at the local office of IBM Germany told me that the language accepted was completely identical to FORTRAN II. But the creators were not allowed to call the program a FORTRAN compiler, because the parent company had the right of ownership for the FORTRAN language. Two years later I used the official FORTRAN compiler for the 1401 for special project with a lot of mathematics. The small address space, the low operating speed and the lack of floating point operations were a serious restriction for routine applications using the FORTRAN language for the IBM 1401, but it worked.

[jimdempseyatthecove]

The particular interesting thing about the link I provided is it wasn't a benchmark with spec's kind of demo with FLOPS. It illustrated a complete example of what happened in a data center to compile and run a Fortran program (including the OOPS of missing blank card, forgetting to poke a button, parity error, etc...). What it did not show (illustrate), is that the typical interaction programmer/engineer/student had with this machine... they didn't.

The programmer/engineer/student usually had access to only the card punch/duplicator, or access to a staff of people that used the card punch/duplicator. After careful preparation of a program (pencil and paper), you would commit it to punched cards. You then by hand, visually walked through the card deck to make the initial debugging and/or edits to be made (hand notes on paper or scribbles on cards). This then required a trip back to the card punch/duplicator (or submission to staff member to edit your card deck). Only when you were satisfied that your program should be correct, would you then submit your card deck to the data center (possibly inclusive of a deck of data cards, or a note indicating which stack of data cards to use from those kept at the data center, or mag tape reel ID). Usually the data center had a customer service center window where you would hand someone your card decks. Then you returned to your office to work on something else.

Depending on how busy the data center was, if you were lucky, you may get the results back the same day or the following day. Oops, I should say you will get something back: either a successful run with results data (printout/card deck/mag tape), or an indication a programming error (printout, original program deck, data deck/mag tape)....

Looking at this today, it was like an episode of Monty Python's Ministry of Silly Walks

Jim Dempsey

[cryptogram]

First computer I used was in high school, back in 1975.  This was a 1960's vintage IBM 1620 with 20,000 digit magnetic core memory and we programmed it using FORTRAN II.  

Was taking calculus at the time, and one of the problems on a take home test referred back to the fundamentals of integration using a Reimann

sum with a delta x approaching 0.  You were meant to jump right to the answer, which was ln (2), but I thought it would be fun to program this

into the old 1620 with 10,000 slices and see how long it would take, which turned out to be a solid 40 minutes.

I showed this to my father and he ran it through the mighty (for its time)  IBM 360 mainframe computer at ORNL and it took about  15 seconds.