Correction to “logical” arithmetric on computers with two’s complement binary arithmetic

In reference [1] algorithms for performing arithmetic with unsigned two's complement operands were described. The scheme for division was implemented in a set of multiple-precision floating-point arithmetic routines [2]. User experience with those routines showed that there is one case where the algorithm fails [3]. We give here a modification to the algorithm which eliminates the error condition. Equations are numbered beginning with (20), so that we may refer to equations in the original paper as well.

View this article in the ACM Digital Library.