Abstract. Some methods being used to speed up floating-point computation in Maple are described. Specifically, we discuss the evaluation of the elementary functions square root and logarithm. We also describe the correct rounding of these evaluations. The key efficiency improvement is called progressive precision increase.