On Thu, 13 Mar 2003, Jeff Squyres wrote: > On Wed, 12 Mar 2003, Mattson, Timothy G wrote: > > > [snipped] > > compilers will be in beta release soon. Hence, at least for icc, there > > may be no need to worry about separate mpi or PVM rpm's (note: this > > assumes that there are not big performance advantages to using icc to > > build pvm or mpi). > > I think that's the big issue -- performance. > > For example, there can be large performance differences between gcc 2.95 > and 3.2. I'm assuming that icc, being a vendor-supported compiler, will > have even better performance (although I have no empirical data to back > that up). This becomes critical for numerical libraries, not just MPI and > PVM libraries.
Yeah, but as long as icc and gcc are link compatible (which is what Tim says is going to happen), it doesn't matter. The KAI team didn't make any difference for LAM, so it's pretty doubtful that the Intel team will do anything for LAM. User apps, yes. But if icc and gcc are link-compatible, who cares? Compile LAM with gcc ship it. Let the user compile his app with icc, and the world is good. Brian -- Brian Barrett LAM/MPI developer and all around nice guy Have a LAM/MPI day: http://www.lam-mpi.org/ ------------------------------------------------------- This SF.net email is sponsored by:Crypto Challenge is now open! Get cracking and register here for some mind boggling fun and the chance of winning an Apple iPod: http://ads.sourceforge.net/cgi-bin/redirect.pl?thaw0031en _______________________________________________ Oscar-devel mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/oscar-devel
