This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Floating point trouble with x86's extended precision
- From: dewar at gnat dot com (Robert Dewar)
- To: sabre at nondot dot org, wilson at tuliptree dot org
- Cc: gcc at gcc dot gnu dot org, lucier at math dot purdue dot edu
- Date: Fri, 22 Aug 2003 21:49:22 -0400 (EDT)
- Subject: Re: Floating point trouble with x86's extended precision
> Actually, the easiest thing to do is to change the register allocator to
> spill values in their 80-bit form. This at least eliminates the
> unpredictable extra rounding that you get because of register pressure.
> We use this strategy in the LLVM x86 backend, and it seems to work pretty
> well. This obviously is not a complete solution, but it's one easy
> step...
Well this seems clear. AFter all if you *are* using 80-bit fpt, then I
would definitely say it is a bug to do a 64-bit spill. Extra precision
is one thing, random loss of precision is another (and would for example
violate the Ada standard requirements).