[Beowulf] RE: Compare and contrast MPI implementations

Robert G. Brown rgb at phy.duke.edu
Tue Dec 20 13:53:40 PST 2005


On Tue, 20 Dec 2005, Mark Hahn wrote:

> scheme.  it's rather disappointing to find out how often users compile code
> with something like "gcc *.c -o whatever" and expect it to work well without
> optimization flags...

Of course in the best of all possible universes gcc *.c -o whatever
would produce optimally optimized code totally automagically WITHOUT
flags at least wrt the native compile host... where really the only
tradeoffs a user might want to make by hand are to optimize for length
of program vs performance.

Yeah, yeah, reality is complex and life is hard (and then you die) so
this doesn't usually work.  But it IS the ideal.  After all, anything
you might do to pick flags for optimal performance a computer can do for
you, if you make it smart enough and self-capability-aware enough.

My enduring dream is for a linux computer (and its compilers/libraries)
to become self-capability-aware enough to do better at this sort of
thing than all but the smartest programmers.

    rgb

>
> _______________________________________________
> Beowulf mailing list, Beowulf at beowulf.org
> To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf
>

-- 
Robert G. Brown	                       http://www.phy.duke.edu/~rgb/
Duke University Dept. of Physics, Box 90305
Durham, N.C. 27708-0305
Phone: 1-919-660-2567  Fax: 919-660-2525     email:rgb at phy.duke.edu





More information about the Beowulf mailing list