GPU computation

Application specific recommendations for GPU computations


To enable GPU acceleration of your code, 2 conditions need to be met:
  1. 1.
    You need to run your application on a GPU enabled node. By default applications on Nuvolos run on nodes that do not have a GPU card integrated, however with a single click you can scale your applications to run on GPU accelerated nodes.
  2. 2.
    You need to make sure the applications are configured to use a GPU. The documentation below mostly addresses the configurations needs to be done for applications to be able to use a GPU once it's available.
The NVIDIA device drivers will be loaded in all GPU supported images once a GPU node is started on Nuvolos. However depending on the image type additional components (e.g. CUDA toolkit) might need to be installed via conda.
If you launch a GPU accelerated node on Nuvolos, the nvidia-smi tool will be available from the command line / terminal. You can use this to check the load on the given GPU to verify how effectively the code is leveraging the accelerator.
Please find below some examples on how to get started with GPU computations on Nuvolos or consult directly the relevant machine learning library documentation. If you require additional support, please reach out to our team directly.


We recommend to install the appropriate cudatoolkit from conda that is compatible with the target library and then use a conda packaged version of the library that can leverage the cudatoolkit.


For example, to run pytorch with on Nuvolos, simply run:
conda install pytorch torchvision torchaudio cudatoolkit=10.2 -c pytorch


To run tensorflow with GPU acceleration, install via conda:
conda install -c anaconda tensorflow-gpu==2.4.1
pip install gast==0.3.3


With Machine Learning (CUDA enabled) Rstudio images you can run GPU computations on GPU accelerated nodes. These images have the CUDA runtime / toolkit installed as well.


We recommend to use the pre-built experimental binary to get started with XGBoost and R. In a terminal on a GPU node:
# define version used - update if needed
# download binary
# Install dependencies
R -q -e "install.packages(c('data.table', 'jsonlite'))"
# Install XGBoost
R CMD INSTALL ./xgboost_r_gpu_linux_${XGBOOST_VERSION}.tar.gz
You can test the code via the following example program:

Tensorflow / Keras

You can use Tensorflow with GPU acceleration, by following our Tensorflow installation guide and selecting to install version = "gpu" when installing Tensorflow.