[Beowulf] Re: Beowulf Digest, Vol 15, Issue 24
mathog at mendel.bio.caltech.edu
Tue May 10 11:59:13 PDT 2005
> I will readily agree that if I were doing it for myself, and I
> time free (as I did in younger, less cynical days), I probably would slog
> through coming up with new code. (And, for that matter I AM developing
> software, on a cluster, at home on my own time, without much
> financial return).
Pretty sure that recompiling the existing code would be cost effective.
It would let you build an executable for the processors you actually
have now (instead of Pentium I's or whatever). That could easily half
the execution time of the same computation on the same hardware.
(Much cheaper than buying 2x as much hardware!) Also compilers have
been improving with time, especially gcc, where the C compiler used
to produce code much slower than the Intel compiler, but lately they've
been pretty close. (YMMV.) Now if your code is written
in Fortran IV and there's 100k lines of it a simple recompile might
not work out too well. If it's at least Fortran 77 the odds are
pretty good that you can recompile it without too much pain.
Where did this code come from? Can you obtain a newer version?
Again, a couple of thousand spent on a newer version of the
same software might prove to be quite cost effective.
The other reason to recompile code from time to time is that the
newer compilers tend to catch more errors than the older ones did.
All that being said, it might be wise to avoid the Intel compilers
when the target is an AMD processor. Intel seems to have put
some effort into making its compilers produce code that runs
worse on AMD CPUs than on its own processors
mathog at caltech.edu
Manager, Sequence Analysis Facility, Biology Division, Caltech
More information about the Beowulf