FP negation optimization

Robert Dewar dewar@gnat.com
Mon Feb 2 02:13:00 GMT 2004


> I agree that this optimization is "wrong" if you expect IEEE
> semantics to be observed. Interestingly, Ada requires that this
> be done right, to be checked!

I checked, and Ada gets it right at -O2, not sure what the
story is, to be investigated.



More information about the Gcc mailing list