From: krishna (krishnendu39_at_gmail.com)
Date: Tue Jul 26 2016 - 00:56:38 CDT
Dear All
We have gpu k20 and I want to use 1gpu and 16 cpu for md run. I'm using
script below. I set cuda path correctly.
#!/bin/bash
#SBATCH --job-name=krishna
#SBATCH --nodes=1
#SBATCH --ntasks-per-node=1
#SBATCH --gres=gpu:1
#SBATCH --begin=now
#SBATCH -e slurm.err
#SBATCH -v
cd $SLURM_SUBMIT_DIR
echo "CWD:" $SLURM_SUBMIT_DIR
echo "NODELIST:" $SLURM_JOB_NODELIST
source /etc/profile
export PATH=/opt/cuda-5.0/bin:$PATH
export
LD_LIBRARY_PATH=/opt/cuda-5.0/lib64:/opt/cuda-5.0/lib:$LD_LIBRARY_PATH
# SET FOR NUMBER OF CORES(THREADS) TO USE
export OMP_NUM_THREADS=16
charmrun +p16 ++local namd2 +setcpuaffinity +idlepoll npt30.inp > npt30.out
But the error message:
namd2: error while loading shared libraries: libcudart.so.4: cannot open
shared object file: No such file or directory
Please help
*********************************************
Regards
Krishnendu Bera
PG Student
Department of Bioinformatics
Central University of South Bihar, India
https://sites.google.com/site/krishnendu2345/
http://krishna-biodiesel.webs.com/
**********************************************
This archive was generated by hypermail 2.1.6 : Sun Dec 31 2017 - 23:20:36 CST