NAMD 3.0b7 - SMP+CUDA (single node)

Webpage

http://www.ks.uiuc.edu/Research/namd/

Version

3.0b7

Build Environment

  • GCC 8.5.0
  • Intel MKL 2024.1
  • CUDA 12.4 Update 1

Files Required

  • NAMD_3.0b7_Source.tar.gz
    • tcl and tcl-threaded are obtained from http://www.ks.uiuc.edu/Research/namd/libraries
    • fftw of MKL is used

Build Procedure

#!/bin/sh

VERSION=3.0b7
CHARM_VERSION=7.0.0
WORKDIR=/gwork/users/${USER}
SOURCEDIR=/home/users/${USER}/Software/NAMD/${VERSION}
NAME=NAMD_${VERSION}_Source

TARBALL=${SOURCEDIR}/${NAME}.tar.gz

LIBURL=http://www.ks.uiuc.edu/Research/namd/libraries
TCL=tcl8.5.9-linux-x86_64
TCL_URL=${LIBURL}/${TCL}.tar.gz
TCL_THREADED=tcl8.5.9-linux-x86_64-threaded
TCL_THREADED_URL=${LIBURL}/${TCL_THREADED}.tar.gz

TARBALL_TCL=${SOURCEDIR}/${TCL}.tar.gz
TARBALL_TCL_THREADED=${SOURCEDIR}/${TCL_THREADED}.tar.gz

PARALLEL=12

#------------------------------------------------------------------
umask 0022

export LANG=""
export LC_ALL=C

module -s purge
module -s load mkl/2024.1
module -s load cuda/12.4u1

cd ${WORKDIR}
if [ -d ${NAME} ]; then
 mv ${NAME} namd_erase
 rm -rf namd_erase &
fi

tar zxf ${TARBALL}
cd ${NAME}
tar xf charm-${CHARM_VERSION}.tar

cd charm-v${CHARM_VERSION}

export CC=gcc
export CXX=g++
export F90=gfortran
export F77=gfortran

./build charm++ multicore-linux-x86_64 gcc \
       --no-build-shared --with-production -j${PARALLEL}
cd ../

tar zxf ${TARBALL_TCL}
mv ${TCL} tcl
tar zxf ${TARBALL_TCL_THREADED}
mv ${TCL_THREADED} tcl-threaded

./config Linux-x86_64-g++ \
        --charm-arch multicore-linux-x86_64-gcc \
        --with-mkl \
        --with-python \
        --with-single-node-cuda

cd Linux-x86_64-g++

make -j${PARALLEL}
make release
# install contents of Linux-x86_64-g++/NAMD_3.0b7_Linux-x86_64-multicore-CUDA.tar.gz into /apl/namd/3.0b7-smp-cuda

Notes

  • Failed to build with gcc11, 12, and 13 (syntax error).
    • This is probably due to the implementation of new features and bug fixes.
  • Gcc10 can build the executable without problem. However, the performance of GPU runs is slightly worse than gcc8 version.
    • (For the performance of CPU part, gcc10 version is better than gcc8 one.)
    • The overall performance is comparable or very slightly worse than the previous build (3.0b6).
View PDF