Float enables programs to use the the floating point instructions of VFP enabled processors. It defaults to the FPA instructions when VFP is not available. The VFP instructions do not extend to trancendental functions such as Sine and Exponential so these are provided with fast routines.