From: Axel Kohlmeyer (akohlmey_at_cmm.chem.upenn.edu)
Date: Tue Jun 02 2009 - 22:17:01 CDT
On Wed, 2009-06-03 at 10:43 +0800, liupeng012 wrote:
>
>
> HI! everyone!
> now, i have already compiled the namd2.7b1 on infiniband. but i
> find that it is slower than namd2.6.
> 1> build icc/ifort 10.1.022
please first try to build with gcc/g++.
that will make your life much easier.
> 2> build the openmpi
> ./configure --prefix=/hptc_cluster/openmpi-1.2.8-intel
> CC=icc CXX=icpc --with-openib --without-udapl
> --disable-openib-ibcm
> 3> build the charm
> ./build charm++ mpi-linux-x86_64 smp mpicxx -j32 -O2
> -DCMK_OPTIMIZE --no-build-shared
since you build with SMP support, have you checked,
that you are not oversubscribing your nodes?
have you tried to build without?
cheers,
axel.
> --incdir=/opt/intel/cce/10.1.022/include
> --libdir=/opt/intel/cce/10.1.022/lib
> --incdir=/opt/intel/fce/10.1.022/include
> --libdir=/opt/intel/fce/10.1.022/lib
> 4> build the charm
>
> in this route, no fatal errer occure to terminate the
> compileing . but the final performance is very not good. or it is very
> bad.
>
> what's worry?
>
> any help is appreciated!
>
> thanks!
>
> yours liupeng
>
>
> 2009-06-03
>
> ______________________________________________________________________
> liupeng012
>
> ______________________________________________________________________
>
-- ======================================================================= Axel Kohlmeyer akohlmey_at_cmm.chem.upenn.edu http://www.cmm.upenn.edu Center for Molecular Modeling -- University of Pennsylvania Department of Chemistry, 231 S.34th Street, Philadelphia, PA 19104-6323 tel: 1-215-898-1582, fax: 1-215-573-6233, office-tel: 1-215-898-5425 ======================================================================= If you make something idiot-proof, the universe creates a better idiot.
This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:52:53 CST