Re: netwinder floating point
Peter Teichmann wrote:
It is probably compiled with FastFPE because it is noticably faster (for
instance with apt and dselect) and there were no known bugs anymore. Would
you please consider reporting what kind of problem you have with sqrt and
FastFPE? Without useful error reports, errors will just persist.
Hmm, bit offtopic here, but why would apt or dselect need floating point
at all? Surely there are no tasks in either of those programs that
would need anything but integer math? Speedup must be somewhere else...