Using AMBER 14 on the GPU and compute clusters

Using the pre-compiled AMBER 14 source

We have a site license for AMBER 14, and it has been pre-compiled for you to use on the GPU and compute clusters. You can find the AMBER 14 manual here.

In order to use AMBER 14, you need to add a few lines to your ~/.bashrc file on the machine you are working on:

volkhan/sinister/dexter

# Adding support for AMBER 14
export AMBERHOME=/home/csw34/amber14
if [ -z "${LD_LIBRARY_PATH}" ]; then
   export LD_LIBRARY_PATH="${AMBERHOME}/lib"
else
   export LD_LIBRARY_PATH="${LD_LIBRARY_PATH}:${AMBERHOME}/lib"
fi
# PATH
export PATH=$AMBERHOME/bin:$PATH

You should then be able to access all of the AMBER 14 binaries (pmemd, ptraj, tleap etc) from anywhere.

GPU cluster (pat)

# Adding support for AMBER 14
export AMBERHOME=/home/csw34/amber14
if [ -z "${LD_LIBRARY_PATH}" ]; then
   export LD_LIBRARY_PATH="${AMBERHOME}/lib"
else
   export LD_LIBRARY_PATH="${LD_LIBRARY_PATH}:${AMBERHOME}/lib"
fi
# PATH
export PATH=$AMBERHOME/bin:$PATH

On the GPU cluster pat, both pmemd and pmemd.cuda have been compiled for you to use. If you are planning to use CUDA, make sure you have loaded the cuda/7.0 module:

module load cuda/7.0

Compiling AMBER 14 for yourself

If you want to compile and use AMBER 14 with special options e.g. MPI, or a different level of CUDA precision, you can find the source code on quartz:

pat:/sharedscratch/csw34/AmberTools14.tar.bz2 
pat:/sharedscratch/csw34/Amber14.tar.bz2

You can find compilation instructions in the AMBER 14 manual. Obviously - please don't distribute the source code beyond the group.