I did some programming in the 60s using punch cards; RPG and COBOL run on an IBM 360-20.
Write the program on graph paper … send to Data Entry Dept and wait for it to be keypunched … check your cards to be sure they are correct (yes, you learned to read the holes in the cards) … send to the computer room and wait until they have time to run your program through once … get a printout and debug by shuffling your cards and/or creating some new ones … back to Data Entry, etc.
We were hearing rumors that some guys in the lab were putting letters up on a TV screen.
My oldest brother was studying physics at Stanford in the 70’s and they used punch cards. He said a fellow student was walking across the room with a box of punch cards for a program he wanted to run. The guy tripped and the box went flying, punch cards all over. My brother said it was literally the first time he had ever seen a grown man cry.
And if that file were a 64kbp MP3 file, to play back the file you’d have to read the cards at a rate of about 100 cards per second. Would that have been feasible?
I always thought it was standard practice to draw a diagonal line across the top of your deck of cards. This made putting them back in sequence after a fall reasonably easy.
The types of readers used in 70s mainframe environments were surprisingly fast for a mechanical device. I remember operators loading decks into the reader on the Sigma computers from my school days. The reader would suck through a foot or so of cards in a few seconds. Earlier ones were nowhere near as fast.
This computer history site gives a figure of 2000 cards/min for 70s vintage optical readers, which sounds right:
Punch cards came in boxes - the height and width what you’d expect, the length was somewhere around 11 inches. I think I still have some holding index cards; I can check. The biggest program I wrote on them was a compiler for a compiler class which got to about two boxes before I got smart and read them in to a editing program that ran right on the mainframe.
In the very early '70s the MIT SF Society kept the index of its holdings, called Pinkdex, on punch cards. New members wishing to get keys to the library got to punch cards for the new books, and every so often the Librarian got to take all the boxes to the collator at the MIT computing center, where you can sort and then print the cards off-line.
As for efficiency, you used one card per 80 (or 72) character record, even if you needed only half that, so they weren’t extremely efficient, and computations of number of cards for a certain file size is best case - the real number would be a lot higher.
IIRC, the size of the card was equivalent to the size of a dollar bill when Hollerith used punchcards for the U.S. census.
Brings back memories. In college we had to punch our own programs with special machines.
Any typo meant you had to retype that line, the entire 80-character card.
After a while I got to be a wizard and could correct cards by re-fitting the chads. To do this you had to know ascii code and binary and have a lot of patience.
Sort of off topic, but I am still working in a S/360 mainframe environment which has the punchcard constraints still in effect.
I write JCL and COBOL (although not so much recently) which still has positional syntax and limitations. The emulator I use to edit code on the mainframe is 80 characters wide, and only the first 72 show up on during editing. We’ve had instances where characters were inserted past postition 72 and the computer didn’t recognize the characters and failed the program.
In another instance in the last week or so, an IF ENDIF condition wasn’t processed correctly because it started in column 6 instead of column 7.
While I don’t have to deal with th punchcards any more, I still deal with the legacy.
Reading it in is only half of the efficiency problem. Your hardware also needs to be beefy enough to decompress the MP3 into something listenable; I don’t think hardware of that era had near enough CPU power nor RAM to do the job in real time. Perhaps something could be cobbled up to output the audio onto audio tape at slower than real-time, and then the tape could be sped up on playback. But then if you’re using tape, why are you mucking with MP3s…
For most applications, 80-character records were cramped, and thus sprung the Y2K problem.
The other problem with encoding MP3 files on punched cards would be that the card-reader would be so loud that you wouldn’t be able to hear any music in the same room.
Though there were some who played “tunes” on line-printers by suitable arrangements of print output.
Why do I have the feelng that this thread is the the output end, so to speak, of the thread where I asked whether the people of 1983 could connect to my Macbook Pro? We had come to the conclusion that the best way to send data out of the Macbook Pro was to synthesize modem tones in the Pro, send them out the line-out connector, and decode them in the external 1983 hardware. Plus the inverse for line-in.
I didn’t use the chads; I used paper ‘dots’ out of a hole-punch for the confetti. The cards were just to keep the rubber band-powered paddle from hanging up on the lip of the desk when it was opened.
I still have the very first program I ever wrote on punch cards. I think it calculates the first 20 Fibonacci numbers or something like that. Anyway, it was written in FORTRAN in 1976.
Or the slightly different braaaaaapppp of them being shredded in the process.
Back when I first started out as a computer operator our fancy new 360/65 included a combination card punch/card reader (known officially by IBM as a Multi Function Card Machine, but more appropriately by the local IBM SEs as the “MF-ing Card Mangler”). It would occasionally get a metal burr stuck in its innards which would then neatly shred cards in two as they went flying through the machine. By the time we computer operators heard the different sound, ran over to the machine and stopped it several hundred cards would be shredded.
Boy, you just don’t get equipment like that anymore.
