Dear all,
I have just checked in an updated configure_amber script into the AMBER 11
tree. This now attempts to detect the version of MKL one is using and links
in the sequential version of it. This avoids all of the problems associated
with having OMP_NUM_THREADS/=1 when running sander in serial or parallel.
Since version 10 of MKL Intel have started parallelizing all of the vector
functions using openmp which of course when running in parallel with MPI can
lead to a performance disaster.
I have attempted to test this with a number of versions of MKL, statically
linked, dynamically linked, em64t, ia32, ia64 etc. However, the degrees of
freedom are huge and I can't possibly test all combinations so if you notice
any build / linking problems with the latest amber11 related to MKL please
let me know.
Additionally this update means you need to remake your config_amber.h file
by rerunning configure_amber.
All the best
Ross
/\
\/
|\oss Walker
| Assistant Research Professor |
| San Diego Supercomputer Center |
| Tel: +1 858 822 0854 | EMail:- ross.rosswalker.co.uk |
|
http://www.rosswalker.co.uk | PGP Key available on request |
Note: Electronic Mail is not secure, has no guarantee of delivery, may not
be read every day, and should not be used for urgent or sensitive issues.
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Fri Mar 20 2009 - 01:17:40 PDT
