Servers#
The web-based servers run on the cluster, and users can access them directly from their browsers. We offer several Jupyter Lab-based apps as well as RStudio.
Jupyter Lab#
This app launch the Jupyter Lab with specific Python version and libraries, based on these Anaconda versions:
- Anaconda3 2024.06-1, python 3.12.4
- Anaconda3 2024.02-1, python 3.11.7
- Anaconda3 2023.07-2, python 3.11.4
- Anaconda3 2023.03-1, python 3.10.9
You can also load other modules in your environment, for instance CUDA related modules.
ML/AI#
Jupyter Lab + Tensorflow, PyTorch and some other common ML libraries:
- Python/3.10.8-GCCcore-12.2.0
- JupyterLab/4.0.3-GCCcore-12.2.0
- CUDA/12.2.0
- cuDNN/8.9.2.26-CUDA-12.2.0
- genomap/1.3.6-TF2.15.0-Torch2.2
- PyTorch/2.2.0-Py3.10.8
- TensorFlow/2.15.0-torch2.2-Py310
PyTorch#
Jupyter Lab + different PyTorch versions:
- PyTorch-2.2, CUDA-12.3, NGC Container, python 3.10.12
- PyTorch-2.1, CUDA-11.7, Anaconda3 2023.03-1, python 3.10.9
- PyTorch-1.12.1, CUDA-11.7, python 3.10.4
The NGC container is optimized for use with NVIDIA GPUs, and contains the following software for GPU acceleration:
- CUDA
- cuBLAS
- NVIDIA cuDNN
- NVIDIA NCCL (optimized for NVLink)
- RAPIDS
- NVIDIA Data Loading Library (DALI)
- TensorRT
- Torch-TensorRT
RStudio#
Rstudio server with specific R and R packages from the Rocker project. Currently we ship:
- R 4.3.2 + Rstudio + Geospatial packages (rocker-4.3.2-geospatial)
- R 4.3.2 + Rstudio (rocker-4.3.2-rstudio)
- R 4.2.1 + Rstudio + Geospatial packages (rocker-4.2.1-geospatial)
TensorFlow#
Jupyter Lab + TensorFlow. Available versions are:
- Tensorflow-2.17, CUDA-12.6, NGC 24.12-tf2-py3, python 3.12.3
The NVIDIA TensorFlow Container is optimized for use with NVIDIA GPUs, and contains the following software for GPU acceleration:
- CUDA
- cuBLAS
- NVIDIA cuDNN
- NVIDIA NCCL (optimized for NVLink)
- RAPIDS
- NVIDIA Data Loading Library (DALI)
- TensorRT
- TensorFlow with TensorRT (TF-TRT)
We have included additional packages and extensions to enhance the user experience. These include:
Jupyterlab-nvdashboard
andnvitop
for GPU monitoring,Jupyter-resource-usage
for CPU and RAM monitoring,Jedi
for auto-completion.