Announcing the availability of OCI Data Science Oracle Cloud Marketplace images, preconfigured with NVIDIA tool kits

October 19, 2022 | 5 minute read
Sanjay Basu PhD
Senior Director - Gen AI/GPU Cloud Engineering
Text Size 100%:

In 2020, we released the virtual machine (VM) image for Data Science and Machine Learning (ML) services on Oracle Cloud Infrastructure (OCI) to use NVIDIA GPUs to speed up AI application development. Read the blog announcing the release and background information.

Since then, we have released three updates to address security vulnerabilities, targeting NVIDIA P100 and V100 Tensor Core GPUs. Before these updates, features mismatched between this OCI VM image and the OCI Data Science service. It also had no support for NVIDIA Container Runtime. This fiscal year, we have addressed these gaps and released the following new Data Science images on OCI:

  • Oracle Linux Image with NVIDIA GPU drivers

    • Oracle-Linux-8.6-Gen2-GPU-2022.05.31-0

    • Image family: Oracle Linux 8.x

    • Operating system: Oracle Linux

    • Release date: May 31, 2022

  • Ubuntu Linux image with NVIDIA GPU drivers

  • Oracle Linux image for x86 (supporting both Intel and AMD)

Essential features

The images have several popular tools preinstalled for modeling, development, training, and inference.

Operating system, drivers, and other base components:

  • Operating system of your choice: Oracle Linux and Ubuntu Linux

  • NVIDIA drivers, CUDA Toolkit, cuDNN library (when GPU machines are used)

  • Docker, NVIDIA Container Runtime, NVIDIA Container Toolkit

  • Anaconda (“conda”)

  • Git

Authoring tools:

  • Visual Studio Code

  • PyCharm Community Edition

  • Jupyter and JupyterLab

ML framework:

  • PyTorch, TensorFlow, MXnet, scikit-learn

  • pyspark

  • dask

Other notable components:

  • OCI command line interface (CLI)

  • OCI software developer kit (SDK) for Python

  • OCI Accelerated Data Science (ADS) kit (including MLX and AutoML)

Data Science notebook:

  • Data Science Conda environments for x86, including General ML. Install them without the conda zip files on the drive.

  • Basic Python conda environment installed on the base image of the notebook session

  • Example notebooks

  • JupyterLab v2.X

  • Oracle Data Science CLI (Requires setup)


The images have preconfigured conda environments that include NVIDIA GPU drivers, CUDA toolkit, cuDNN drivers, common Python and R integrated development environments (IDEs), Jupyter notebooks, and open source ML and deep learning frameworks. For high-performance shared file systems, BeeGFS client is also preconfigured in the current image.

Computational power, ease-of-use, productivity, and efficiency are at the forefront of why this solution is attractive to data scientists. It allows the users to run machine learning models across all OCI’s GPU-accelerated machines, including machines built on NVIDIA A100 Tensor Core GPUs and NVIDIA A10 Tensor Core GPUs. Oracle’s powerful automation capabilities enable the deployment of hundreds of instances for model experimentation and testing through a single click. No more requests for provisioning. No more maintenance issues or refreshes. Just straight deployment that’s ready-to-go with the right open-source frameworks installed, upgraded, and tested.

How to deploy the images

You can deploy the images from the Oracle Cloud Console or Oracle Cloud Marketplace. You can also provision the images using OCI CLI or SDK, called from stand-alone Terraform script, or through Oracle resource manager: NVIDIA GPU image and X86 image

A screenshot of the Marketplace view of the image.

A screenshot of the image in the Oracle Cloud Console.


Overall solution

You can access the instances through Remote Desktop, SSH, or browser such as Jupyter Hub. You have full access to the instances. If needed, you can adjust configurations and install other frameworks like with any other VM. Maintenance and protection against vulnerabilities of provisioned Data Science VMs is the customer’s responsibility.

The solution includes a reference architecture that deploys a bastion host, training node, inference node, user application VM, and other components on OCI. It uses a region with one availability domain and regional subnets, and the same architecture can be used in a region with multiple availability domains. It even runs on Oracle Linux 7.X GPU version. If using Oracle Ksplice, real-time upgrades can occur without shutting down instances.

Security and data privacy are of utmost concern. OCI VMs for Data Science and AI include a security-first design built for the enterprise. Data sets are encrypted, and data privacy is protected through data minimization and transparency. In other words, Oracle has no insight into data on its cloud and is transparent about where this data is processed and stored.

Finally, data scientists can expand their Compute resources by using autoscaling or stopping the Compute instance when it’s not needed to control costs. The VM even includes basic sample data and code for testing and exploring.

Try for yourself

To learn more about our GPU shapes and try them out, visit the NVIDIA and Oracle Cloud Infrastructure NVIDIA GPU Cloud Platform. To learn more about Oracle’s data science solutions, visit the OCI Data Science page, and follow us on Twitter.

Get your free account for Oracle Cloud Infrastructure and get started today!

Sanjay Basu PhD

Senior Director - Gen AI/GPU Cloud Engineering

Sanjay focuses on the advanced services like Generative AI, Machine-Learning, GPU Engineering, Blockchain, Microservices, Industrial IoT, 5G core along with Cloud Security and Compliance. He has double masters in Computer Science and Systems Design. His PhD was in Organizational Behaviour and Applied Neuroscience. Currently, he is pursuing his second PhD in AI. His focus of research is Retentive Networks.

Previous Post

Oracle Cloud adds NVIDIA chips, software to speed enterprise AI uptake

Aaron Ricadela | 3 min read

Next Post

Accelerating quantum simulations with NVIDIA cuQuantum Appliance on OCI Marketplace

Sanjay Basu PhD | 4 min read