[LUGSB] Re: INTs as BYTEs in MPI

[Jeff Sipek]

On Mon, May 16, 2005 at 05:21:12PM -0400, Lucas Carey wrote:
> I guess I'll go with the struct and Pack/Unpack it. 
> The send/recv condensation is not premature optimization -- that is
> definetly one of the many things needed to scale to large numbers
> of nodes.

It is not iff you are finished implementing functionality and now are
fixing up bottle necks. Premature optimization is more of a "this
function will be called a lot so, I'll do some hack-foo".

Just my 2 cents.

Jeff.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
Url : http://www.fsl.cs.sunysb.edu/pipermail/lugsb/attachments/20050516/84d0e367/attachment.bin