Problem in cuda for AMBER 14

Following error message appeared when I tried the steps as given below
/data/apps/amber/amber14/cuda/bin/pmemd.cuda: error while loading shared libraries: libcurand.so.6.5: cannot open shared object file: No such file or directory

But, libcurand.so.6.5 is located at “/data/apps/cuda/6.5/lib”

1] The command was “qsub TEST_1_cuda_AMBER14_1.sh” where “qsub TEST_1_cuda_AMBER14_1.sh” consists of
#!/bin/bash
# -N TEST # -q gpu
# -l gpu=1 # -pe gpu-node-cores 6
#$ -m beas
/usr/bin/perl production_cuda_mr.pl

2] Command issued
module load cuda/6.5
module load gcc/4.4.3
module load amber/14-cuda

3]
In bashrc file, I made the changes as
export PATH=/data/apps/cuda/6.5/bin:PATH export LD_LIBRARY_PATH=/data/apps/cuda/6.5/lib:{LD_LIBRARY_PATH}
export AMBERHOME=/data/apps/amber/amber14/cuda
export PATH={PATH}:{AMBERHOME}/bin

I posted this question on AMBER forum but the response didn’t help me to start simulation.

Thank you for help

The reply you received on the AMBER forums was:

http://archive.ambermd.org/201707/0034.html
So evidently you haven’t got all the libs on all the nodes.

Did you follow up on this? What steps did you take to ensure that the required libraries are available on all the nodes of the cluster that you are running your simulation on?

Best I can tell this is not a CUDA question but a machine / software configuration question.