Re: Problems with Hardware, Languages, and Compilers

Herman Rubin <hrubin@stat.purdue.edu> wrote:
>...
> Hardware on the "cutting edge" can use a table lookup to start
> reciprocal or reciprocal square root, and iterative methods using
> only a few multiplications and additions to get accurate values for
> these. The same cannot be done for square root.
>
>The compilers available did manage to do x/y as x*(1/y), but as square
>root is recognized and reciprocal square root is not, it was found
>quite difficult to manage the problem on a routine basis.


Now I am lost. If the compiler can handle x/y as x*(1/y), why on earth
cannot it handle sqrt(x) as x*(1/sqrt(x))?


In both cases, there is the same slight inefficiency, and the problem of
getting last-bit accuracy (or even near last-bit accuracy). But I assert
that the problems are almost identical for the two cases.




Nick Maclaren,
University of Cambridge Computer Laboratory,
New Museums Site, Pembroke Street, Cambridge CB2 3QG, England.
Email: nmm1@cam.ac.uk
Tel.: +44 1223 334761 Fax: +44 1223 334679
--

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