Inaccurate cancellation for a fraction.

Inaccurate cancellation for a fraction.Clip both nominator and denominator to said number of bits. If either of those already have equal or less number of bits used, this method does nothing.

Parameters

nSignificantBits denotes, how many significant binary digits to maintain, in both nominator and denominator.

(8) has an error <1% [1/2^(8-1)] - the largest error occurs with the following pair of values:

binary 1000000011111111111111111111111b/1000000000000000000000000000000b = 1082130431/1073741824 = approx. 1.007812499

A ReduceInaccurate(8) yields 1/1.