------- Comment #2 from hubicka at ucw dot cz 2010-02-10 09:26 -------
Subject: Re: New: [4.5 regression] IRA register
preferencing bug
Hi,
note that it is related to way we compute cost through alternatives.
I had very old patch for this
http://www.x86-64.org/pipermail/patches/2001-February/001071.html
that would eventually help to remove the '#' constraints on all FP operations
that are there
to avoid RA from mixing integer and FP classes. Perhaps IRA can now do
something similar
and actually propagate the info?
Honza
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42961
