More on the use of stacks in the evaluation of expressions
J. Leslie Keedy · ACM SIGARCH Computer Architecture News · 1979
Since my demonstration by example that the accumulator~stack technique is superior to the pure stack technique[2] was apparently unconvincing [6], I am forced to raise the issue once again, this time without reference to examples, in an attempt to show that the accumulator/stack technique produces more compact code and less instructions in the evaluation of expressions.This technique was described in [2].To keep the issue simple we restrict the comparison initially to assignment statements where the expression on the right hand side uses diadic operators linking immediately addressable operands, and does not contain function calls.We also initially assume fixed length function codes and operand addresses in instructions.Throughout the comparison we shall use the following parameters: f: a: i: S: t: the number of operators in the expression (excluding the assignment operator) the number of operands in the expression (i.e. on the right hand side) the number of intermediate results which cannot immediately be reused the size of an operand address in a machine inst~ruction the size of the function code in a machine instruction