Use the GPU nodes How to Use the GPU Enabled Nodes ARCHIE currently has 8 nodes with attached nVidia M2075 GPU’s. These are Fermi based GPU’s with 448 cores and 6GB RAM. The nodes themselves are identical to the standard parallel nodes with 12 cores and 48GB RAM. 1. Interactive sessions It is possible to obtain an interactive shell (so that the user can execute programs directly on the GPU and run the visual profiler, for example), via the command: qrsh -V -P my-projectID.prj -pe gpu-verbose 1 xterm 2. Using a job script Jobs can be submitted in the usual way via a job script: # # ************* SGE qsub options **************** #Export env variables and keep current working directory #$ -V -cwd #Select parallel environment and number of parallel queue slots (nodes) #$ -pe gpu-verbose 1 -q gpu-low.q #Add ProjectID #$ -P testing.prj #Combine STDOUT/STDERR #$ -j y #Specify output file #$ -o out.$JOB_ID #Request resource reservation (reserve slots on each scheduler run until enough have been #gathered to run the job #$ -R y #Request exclusivity of each node #$ -ac runtime="3hours" # ************** END SGE qsub options ************ ./myCUDAcode This script requests a single GPU node using the low-priority gpu-low.q queue.