> Hi Fellows,
>
> Quadravox/imagecraft, etc. are _planing_ to add 64 bits floats support to
> their compilers.
> With gcc this can be done _right away_ (say, I'll upload fp double support
> within an hour). But...
>
> I doubt this is necessary for the controller with 2K ram only. 64 bit
doubles
> will eat lots of ram and will be slow.
>
> Then also, all FP library (functions) must be rewritten.
>
> I doubt I can do this 'flagged'.
>
> Lets vote. Who wants doubles 64 bits ieee854 ?
> Who does not?
>
> ~d
>

I would not imagine that 64-bit doubles are a real issue for most msp430
users.  If they are quick and easy for you to implement, then maybe it is
worth it.  However, it is important that most users get plain 32-bit floats
throughout unless they specifically request 64-bit double support, as it
would make the generated code much bigger and slower.  As far as I recall,
there are some gcc flags to force all "doubles" to have the same precision
as normal floats - is it possible to make these the defaults for the
msp430-gcc port?  An alternative way to let people get high precision if
they really need it, but not to get it accidently when they don't, might be
to make "double" 32-bit, and "long double" 64-bit ?

mvh.
David



Reply via email to