Draft "Unsafe fp optimizations" project description.

dewar@gnat.com dewar@gnat.com
Mon Aug 6 18:32:00 GMT 2001


<<It would also be rather dubious on the SH4, where single precision is
considerably faster than double precision.
>>

Not if you already have computed the double precision result. In other words
CSE could regard a*b as equivalent to double(a)*double(b). I agree this is
an unusual case, but certainly might occur.



More information about the Gcc mailing list