Hello Don,
BCS wrote:
Hello Don,
How would that work? Can a system store ints big-endian and floats
little-endian?
Not sure. My limited understanding is that with on PowerPC, Altivec
can
use big-endian floats, even when the PPC is set to little-endian mode.
Could be wrong, though.
In any case, for 80-bit reals, there is no int which is the same size.
So how tightly can we define it without losing to much? If we define a union
where all the members are of the same size to, overplayed exactly and requiter
the to consistent use of big/little-endian, would that make anything horribly
inefficient? Even if that rules out PPC vector ops, I don't think that would
be a big issue.
--
... <IXOYE><
