This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: What is acceptable for -ffast-math? A numerical viewpoint


>>Do you actively oppose to this proposal after rereading it?

Laurent you should reread your procedures, yes, you mentioned performance
in step 2, but in step 3, you said that the optimzation would be accepted
if no difficulties were found, but you did not make it clear that lack
of performance advantage data would be considered a fatal flaw, in fact
you don't even really hint of that critical step.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]