This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
Re: optimization/1762: treatment of NaNs violates IEEE 754
- From: rth at gcc dot gnu dot org
- To: echristo at gcc dot gnu dot org, gcc-bugs at gcc dot gnu dot org, gcc-prs at gcc dot gnu dot org, reichelt at igpm dot rwth-aachen dot de, rth at gcc dot gnu dot org
- Date: 21 Feb 2002 01:28:02 -0000
- Subject: Re: optimization/1762: treatment of NaNs violates IEEE 754
- Reply-to: rth at gcc dot gnu dot org, echristo at gcc dot gnu dot org, gcc-bugs at gcc dot gnu dot org, gcc-prs at gcc dot gnu dot org, reichelt at igpm dot rwth-aachen dot de, rth at gcc dot gnu dot org, gcc-gnats at gcc dot gnu dot org
Synopsis: treatment of NaNs violates IEEE 754
Responsible-Changed-From-To: rth->echristo
Responsible-Changed-By: rth
Responsible-Changed-When: Wed Feb 20 17:28:02 2002
Responsible-Changed-Why:
I can't currently reproduce this with gcc 3.1, but I can't get 3.1 to generate movf.d at all, which is also broken.
Eric, once conditional move is unhorked, I suspect the
problem here is that FP conditional move is not symmetric.
I shouldn't have thought that would be a problem once the
non-reversible comparison is reduced to a boolean. Nor do
I seem to be able to produce the same failure on Alpha,
which has the fcmov insn patterned the same way.
http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=1762