binary - In 2s complement floating point what is wrong with the leading 11? -
binary - In 2s complement floating point what is wrong with the leading 11? -
i think why 01 because not waste important digits , why 00 not good. not sure problem 11 , why important bits must 10 negative 2s complement float normalised.
the question unclear, believe mellyo asking non-standard signed significand + exponent formats. simplicity, let's consider eight-bit twos-complement significands.
all positive numbers have significands of form b0xxxxxxx
. significand normalized if has total representable precision, normalized positive significands have form b01xxxxxx
, , represent 7 important bits; interpreted integers, these significands have values in range [64,127].
all negative numbers have significands of form b1xxxxxxx
. normalized negative significand? precise negation of normalized positive significand; need able represent significands integer values in range [-127,-64], have next bit patterns:
b10000001 -127 b10000010 -126 ... b10111111 -65 b11000000 -64
note things little weird border cases these formats; -64 has bit pattern b11000000
, may want exclude, , -128 has bit pattern b10000000
may want include, people slide range of significands negative values. ones-complement , sign-magnitude more natural representations floating-point significands reason (because have ranges symmetric around zero).
floating-point binary
Comments
Post a Comment