Re: floating point, was inlining + optimization = nuisance bugs

wclodius@aol.com (Wclodius) wrote:


> Toon Moene <toon@moene.indiv.nluug.nl> wrote:


> > Of course, the compiler _could_ remedy this by slowing down the
> > whole computation such that every floating point intermediate was
> > stored in memory (just singling out XNEW is too hard), <snip>


> Is it too hard in general, or would it require a significant
> restructuring of GCC in particular? I would expect this to be fairly
> easy to do, for example, in a compiler that uses single structural
> assignment for its intermediate representation.


Well, now that you point it out to me, I should have added "IMDatedO"
to that phrase - all I know about compilers is written up in the 1977
edition of the Dragon book ;-) SSA has been discussed shortly on the
egcs list when it was very young (the list, that is), but it never got
a follow-up.


Perhaps just storing both sides of an (in)equality operator in memory
before the comparison would already help.
--
Toon Moene (mailto:toon@moene.indiv.nluug.nl)
Saturnushof 14, 3738 XG Maartensdijk, The Netherlands
Phone: +31 346 214290; Fax: +31 346 214286
g77 Support: mailto:fortran@gnu.org; egcs: mailto:egcs-bugs@cygnus.com

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