Compute – GPU Resources Available

The IPPP have several systems with GPUs available for use, they are currently only available via SSH.

ip3-gpu1 (Reserved)

This system is currently only available via ssh for specific IPPP members. Please get in touch.

Coming Soon in late-2024.

• Intel Xeon Gold 6248 (48 Core/96Threads) @3Ghz
• 768GB RAM
• 2TB Scratch
• 4x NVIDIA Tesla A100 80GB (Reserved)
• 4x NVIDIA Tesla A100 80GB (IPPP)
• Compute Profile 8.0 (Cuda 11+)

ip3-gpu2

This system is only available via ssh for all IPPP members.

• Intel Xeon SIlver 4216 (32Core/64Threads) @2.1Ghz
• 192GB RAM
• 1TB Scratch
• 2x NVIDIA Tesla A100 80GB
• Compute Profile 8.0 (Cuda 11+)

ip3-gpu3

This system is only available via ssh for all IPPP members.

• Intel Xeon SIlver 4216 (32Core/64Threads) @2.1Ghz
• 192GB RAM
• 1TB Scratch
• 2x NVIDIA Tesla V100 32GB
• Compute Profile 7.0 (Cuda 9+)

ip3-gpu4

This system is only available via ssh for all IPPP members.

• Intel Xeon SIlver 4216 (32Core/64Threads) @2.1Ghz
• 192GB RAM
• 1TB Scratch
• 2x NVIDIA Tesla V100 16GB
• Compute Profile 7.0 (Cuda 9+)

Other GPU

We have some other GPU resources available for specific members of the IPPP; just get in touch.

• Intel Xeon Gold 6126 (4Core/8Threads) @2.6Ghz
• 42GB RAM
• NVIDIA Tesla P100 16GB
• Compute Profile 6.0 (Cuda 8+)

CUDA Support

Please see our CUDA article for more details.

Slurm SBATCH

This section is for future use.

Add the following to your SBATCH file to use the GPUs

#SBATCH --partition=gpu

or exclude particular GPUs with

#SBATCH --exclude=ip3-gpu[2-3,5]

You can select a specific GPU type with the following:

#SBATCH --gres=gpu:p100
#SBATCH --gres=gpu:v100
#SBATCH --gres=gpu:a100