Running GPU Jobs#

Running GPU Batch Jobs#

Requesting GPU resources#

You can learn more about the GPUs available on M3 by referring to our other GPU documentation, including our GPU look up tables. Note that desktop GPUs are not available for sbatch job submission or smux jobs, and you will need to use our compute GPUs.

When requesting a Tesla V100 GPU, you need to specify --partition=m3g

#SBATCH --gres=gpu:V100:1
#SBATCH --account=nq46
#SBATCH --partition=m3g

When requesting a Tesla T4 or A40 GPU, you need to specify --partition=gpu

#SBATCH --gres=gpu:T4:1
#SBATCH --account=nq46
#SBATCH --partition=gpu

#SBATCH --gres=gpu:A40:1
#SBATCH --account=nq46
#SBATCH --partition=gpu

Sample GPU Slurm scripts#

To submit a job, if you need 1 node with 3 cores and 1 GPU, then the Slurm submission script should look like:

#!/bin/bash
#SBATCH --job-name=MyJob
#SBATCH --account=nq46
#SBATCH --time=01:00:00
#SBATCH --ntasks=3
#SBATCH --cpus-per-task=1
#SBATCH --gres=gpu:1
#SBATCH --partition=m3h

If you need 6 nodes with 4 cpu cores and 2 GPUs on each node, then the Slurm submission script should look like:

#!/bin/bash
#SBATCH --job-name=MyJob
#SBATCH --account=nq46
#SBATCH --time=01:00:00
#SBATCH --ntasks=24
#SBATCH --ntasks-per-node=4
#SBATCH --cpus-per-task=1
#SBATCH --gres=gpu:2
#SBATCH --partition=m3g

Compiling your own CUDA or OpenCL codes for use on M3#

M3 has been configured to allow CUDA (or OpenCL) applications to be compiled (device independent code ONLY) on the Login node (no GPUs installed) for execution on a Compute node (with GPU).

../../_images/massive-gpu-structure.png

Login nodes can compile some of CUDA (or OpenCL) source code (device independent code ONLY) but cannot run it

Compute nodes can compile all CUDA (or OpenCL) source code as well as execute it.

We strongly suggest you compile your code on a compute node. To do that, you need to use an smux session to gain access to a compute node

smux new-session --gres=gpu:1 --partition=m3h

Once your interactive session has begun, load the cuda module

module load cuda

To check the GPU device information

nvidia-smi
deviceQuery

Then you should be able to compile the GPU code. Once compilation has run to completion, without error, you can execute your GPU code.

Attention

If you attempt to run any CUDA (or OpenCL) application (compiled executable) on the Login node, no CUDA device found error may be reported. This is because no CUDA-enabled GPUs are installed on the Login node. You must run GPU code on a compute node.