A GPU-cluster with 4 nodes and 8 Tesla Fermi 2070 GPU is available for user operation at LRZ, each node is equipped with Intel Xeon based 8-way (L5630) and 2 Tesla Fermi 2070. Note: Interactive jobs are not supported on this system. The resource manager SLURM is used to allocate a free GPUs or CPUs for your job. Usage: First log in to the LRZ linux-Cluster : login and Security Once you've logged into the LRZ Cluster, you should be able to access the node: lxlogin_gpu via: ssh lxlogin_gpu Here's an example jobscript to submit to GPU-cluster using 2 GPUs: cat slurm-gpu-job.sh #!/bin/sh #SBATCH --clusters=gpgpu #SBATCH --nodelist=lxgp1 #or lxgp2..4 #SBATCH --gres=gpu:2 #SBATCH --ntasks=2 #SBATCH -e sample_script_err.log #SBATCH -o sample_script_out.log #SBATCH --job-name=jobname #SBATCH --get-user-env #SBATCH --export=NONE #SBATCH --time=08:00:00 source /etc/profile.d/modules.sh cd mydir ./myprogram Next, submit the job using the sbatch command. sbatch slurm-gpu-job.sh To see your job running type. squeue --clusters=gpgpu -u your-userID Use the scontrol command to get detailed information about the job. scontrol show job job-ID If you have any problems on the GPU-Cluster, please contact HPC support via https://servicedesk.lrz.de/plainsubmit/ This information is also available on our web server http://www.lrz-muenchen.de/services/compute/aktuell/ali4601/ Momme Allalen