Re: Q: division vs multiplication

martens@cis.ohio-state.edu (Jeff Martens) writes:
>You have to be careful here, though, because a right arithmetic
>shift of an odd negative integer will not give the same result as
>division by 2. For example:
>
> -1 div 2 is 0
> ashr(-1) is -1


Doesn't this thread recycle every six months or so? The first statement
is only a convention, true on some but not all processors, and not a very
good one at that (for many types of signal processing, defining division
as truncate-toward-zero is a mistake, as it introduces a bias in the data:
uniformly distributed integers are no longer uniformly distributed after
division, instead 0 becomes more likely than other values). It's not even
a standard in many cases: C does not promise that -1/2 is 0; it could also
be -1. C only promises consistency between "/" and "%". If your programs
assume -1/2 is 0, then they are non-portable.




--
-- Joe Buck <jbuck@synopsys.com> (not speaking for Synopsys, Inc)
Phone: +1 415 694 1729
--

Return to the comp.compilers page.
Search the comp.compilers archives again.