next up previous
Next: Introduction to Matlab Up: Numerical Representation Previous: Double Precision

Rounding

When a number in a register of a floating point unit is copied to memory, it must be placed into either single precision or double precision format. This may require that the number be rounded. The default mode is to round to nearest, and to round to even in the event of a tie.



Russ Miller
Thu Sep 14 13:56:19 EDT 1995