The value of optimizing for 0 , as in problem 12.28, depends on the frequency of 0 as a multiplier operand. For the purpose of computing execution times, assume that each microinstruction takes the same amount of time. Compute the number of microinstructions executed for (1) the unmodified routine (from problem 12.27), (2) the modified routine (from problem 12.28 ) with the multiplier equal to 0 , and (3) the modified routine with the multiplier not equal to 0 . Suppose that from empirical measurements of typical multiply operands we find that about 10 percent of the time the multiplier is 0 . Will the modification be beneficial on average?