This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH] Perform constant folding of math builtins
- From: Gabriel Dos Reis <gdr at integrable-solutions dot net>
- To: Roger Sayle <roger at eyesopen dot com>
- Cc: Brad Lucier <lucier at math dot purdue dot edu>, <gcc-patches at gcc dot gnu dot org>
- Date: 28 Aug 2002 20:33:41 +0200
- Subject: Re: [PATCH] Perform constant folding of math builtins
- Organization: Integrable Solutions
- References: <Pine.LNX.4.33.0208281040550.26671-100000@www.eyesopen.com>
Roger Sayle <roger@eyesopen.com> writes:
| My experience is that many of the FP precision related bugs misposted
| to GNATS, e.g. PR other/5574, are due to misunderstandings of floating
| point arithmetic, rather then the results of dubious optimizations.
Mine differs. We just had, last month, a lenghty investigation of GCC's
inconsistent use of FP registers on x86 in the French speaking
newsgroup news:fr.comp.lang.c++ -- for example VC++ with various
optimization levels gets the computation right (as do many other
compilers on various plateforms with various optimization flags), only
GCC fails to deliver a sensical result.
That isn't to say that you should not consider your proposed
transformations. It is just to say that it isn't just a matter of
misunderstandings (or, if misunderstanding there is , it is on GCC
part, IMHO).
-- Gaby