Remarks 

The $FINITEFLOAT directive controls the handling of floating point overflow and underflow, and invalid floating point operations such as division by zero. 

In the {$FINITEFLOAT ON} state, which is the default, the results of floating point calculations are checked, and an exception is raised when there is an overflow, underflow, or invalid operation. In the {$FINITEFLOAT OFF} state, such floating point calculations will return NAN, -INF, or +INF. 

Extra runtime processing is required to check the results of floating point calculations and raise exceptions. If your Delphi code uses floating point operations but does not require strict enforcement of overflow/underflow exceptions, you can turn {$FINITEFLOAT OFF} to get slightly faster runtime execution.