The IBM 650, one of the first general-purpose digital computers, designed in the early fifties, used decimal digits with bi-quinary representation for reasons discussed here: Why did the IBM 650 use bi-quinary?

The Burroughs 205, nee CEC 30-201, was a computer developed around the same time, using essentially the same technology (vacuum tubes and drum memory) that also calculated with words of ten decimal digits plus sign, i.e. an extremely similar design, as far as I can tell not as a result of either company copying the other, but of convergent design driven by similar technological and market forces.

So I was surprised to read in the manual that it does not use bi-quinary, but BCD.

Why does it differ from the 650 in this regard? Was it simply a case of two teams of engineers weighing the trade-offs and reaching different conclusions, or was there some identifiable technical or business reason for the difference?

  • 1
    I don't have an answer but I wanted to point out that bi-quinary uses 6 bits/digit while BCD only uses 4. Unless you are using more than 6 illegal digits for marks or flags, BCD will save 33% on memory. I'm just guessing that the designers of the earlier machines just never thought of BCD. – DoxyLover Sep 8 '20 at 6:08
  • @DoxyLover Isn't bi-quinary seven bits per digit? You are certainly right about BCD saving resources in that regard. Though BCD is obvious enough that I have difficulty believing the designers of the IBM 650 didn't think of it. I'm inclined to believe the explanations in terms of error checking and possibly simpler arithmetic circuitry. – rwallace Sep 9 '20 at 4:19
  • 1
    You are right, 7 bits per digit. I misread the Wikipedia page. I have a hard time imagining that the trade off of simpler processing circuitry could possibly trump the significantly larger memory requirements and error checking could easily have been done with a parity bit. But then again, I really have no idea of the real design considerations. – DoxyLover Sep 9 '20 at 7:39
  • 1
    @rwallace: Storing a bit requires two inverting gates. Storing a one-of-five selection requires five larger-fan-in gates. Storing four bits requires eight small-fan-in gates, while storing a bi-quinary decimal digit requires two small-fan-in gates and five large-fan-in gates, so the storage cost tradeoff is far from clear. – supercat Feb 22 at 18:22

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.