LIBRA Cluster

To Get Userid in GPU Cluster, Please Click Here

How do I run jobs using GPU processors?

Configuring /Accessing CUDA

CUDA-3.0 is currently installed in all the nodes. The environment variables for CUDA is loaded in the PATH itself

CUDA Example

After login to your home directory,Copy the test directory (NVIDIA_GPU_Computing_SDK) from /usr/local/cuda to your local directory.

cp -R /gpuusers/NVIDIA_GPU_Computing_SDK .

To run a deviceQuery executable, please do the following steps:

  1. Go to the location of executables:

    2. cd NVIDIA_GPU_Computing_SDK/C/bin/linux/release

  2. Batch job:

File Name : job.sh

#! /bin/bash

#PBS -o logfile.log

#PBS -e errorfile.err

#PBS -l cput=02:00:00

#PBS -l select=1:ncpus=1:ngpus=1

cat $PBS_NODEFILE

cd $PBS_O_WORKDIR

./deviceQuery

Submit the job:

qsub

You should be able to get information about CUDA. This information such as number of cores, number of multiprocessors etc might be useful during CUDA programming.

If you there is no path to cuda then, you won't be able to load cuda libraries and encounter an error as follows:

./deviceQuery: error while loading shared libraries: libcudart.so.3: cannot open shared object file: No such file or directory

( Important Note: Please consider using batch submission to minimize the wastage of GPU resources and restrict interactive method for testing purpose only.)

AMBER

Gromacs

Namd