Card Games (Score 1)

Cards could also store the object code in "binary" form. Ordinarily the card held one character per column, but in binary mode each row of each column (about 12 rows if I remember correctly) was a bit, and either had a hole punched in it or not. You didn't punch in the binary yourself, you wrote the program as characters and then when the job ran, the computer (if you asked it to) would write the object code onto cards in binary format. Binary card decks were considerably smaller than character cards. It also saved compilation time, just load, link, and go. You could even get the machine executable on cards, that was more bulky because the cards had to include bits for all the allocated storage. (This was Fortran, no dynamic memory allocation in those days, all the ram for all the variables was allocated at compile time.)

  The great drawback of binary cards was that they were unmodifiable, but when running the same program a number of times that didn't matter. Some flexibility was possible because you could have several versions of some subroutines which you could swap in and out. The main problem with binary was the card reader, reading binary cards was much slower and more noisy because of the greater number of holes per cards. The error rate was higher too, some "user" card readers were almost unusable since they continually misread cards. In those computer centers you gave the binary deck to the operators because their card reader was much better.

Another thing about card punch machines was how noisy they were, like manual typewriters but worse. They didn't just hit paper with a key, they punched a hole in card stock, truly high impact printing. Even though the card punch was bolted to the table it was on, the table would shake.