CUDA - nesi/applications GitHub Wiki

Table of Contents

Details

Description: CUDA (formerly Compute Unified Device Architecture) is a parallel

Example SLURM jobs

example

Job description (download)
#!/bin/bash
# CUDA SubmitScript
##########################################################################
#SBATCH -J CUDA_JOB
#SBATCH --time=01:00:00     # Walltime
#SBATCH -A uoa99999         # Project Account
#SBATCH --mem-per-cpu=2048  # memory/cpu (in MB)
#SBATCH --cpus-per-task=2   # 2 OpenMP Threads
#SBATCH --ntasks=1
#SBATCH --gres=gpu:2
##########################################################################
###  Load the Enviroment Modules 
source /etc/profile
module load CUDA/5.5
##########################################################################
###  The files will be allocated in the shared FS ($SCRATCH_DIR)
cd $SCRATCH_DIR
##########################################################################
###  Run the Parallel Program
env | grep SLURM 
srun env

mpi

Job description (download)
#!/bin/bash
#SBATCH -J GPU_JOB
#SBATCH --time=01:00:00     # Walltime
#SBATCH -A uoa99999         # Project Account
#SBATCH --ntasks=4          # number of tasks
#SBATCH --ntasks-per-node=2 # number of tasks per node
#SBATCH --mem-per-cpu=8132  # memory/cpu (in MB)
#SBATCH --cpus-per-task=4   # 4 OpenMP Threads
#SBATCH --gres=gpu:2        # GPUs per node
#SBATCH -C kepler
source /etc/profile
srun binary_cuda_mpi

⚠️ **GitHub.com Fallback** ⚠️