Wayne State University

AIM HIGHER

Wayne State University

High Performance Computing Services

NVIDIA Tesla

HOW TO PREPARE TO DEVELOP AND COMPILE CODE FOR THE NVIDIA TESLA

Follow these steps to develop and compile code for the Nvidia Tesla graphic processing units.

Log on to the Grid.

Start an interactive job, specifying the amount of GPUs that you want. In this example we submit to the smt development queue, dsmtq, which has NVIDIA Tesla cards. We ask for 1 GPU. Example:

qsub -I -q dsmtq -l ngpus=1

NOTE: The dsmtq only has a 12 hour walltime.

Input: qsub -I -q dsmtq -l ngpus=1

Load the cuda-6.0 module. Example:

module load cuda-6.0

Input: module load cuda-6.0

You can find the path for the modules. Example:

which nvcc

which nvcc

The path is shown.

Output: /wsu/apps/compilers/nvidia/cuda-6.0/bin/nvcc
Computer Center Services * 5925 Woodward Avenue Detroit, Michigan 48202