Abstract :
The methods of performing addition on a parallel arithmetic unit are discussed, and a method based on the reversal of the order of calculation in the conventional counting method is devised. A parallel unit, built at Imperial College and using a packaged construction, has been based on this method. The methods of performing the various arithmetic operations are described and the control for the unit (also built of packages) is outlined. A solution to the problem of detecting and correcting for spill using the standard packages is developed. Estimates of times for operation are given as less than 4 microsec for addition and 1 microsec for shifting, with corresponding times for multiplication and division.