Hi
Because i will use DSP to do fixed point
encoding, and DSP always does 16 bits fixed point integer arithmatics. So, for
data greater than int or less than 2^(-32), i just try to use mantissa and
exponent to represent them, the result is slow! i just want to know how to
optimize it or
On Tue, Oct 16, 2001 at 04:30:36PM +0800, caihong wrote:
Hi
Because i will use DSP to do fixed point encoding, and DSP always
does 16 bits fixed point integer arithmatics. So, for data greater
than int or less than 2^(-32), i just try to use mantissa and
exponent to represent them,