The latest cloud infrastructure announcements, technical solutions, and enterprise cloud insights.

Oracle Announces NVIDIA A100 Tensor Core GPU Availability, Startup Program Collaboration, Data Science VMs, and More

Karan Batta
Vice President, Product

Over the last few years, we’ve collaborated with NVIDIA on various offerings, beginning with Pascal generation P100 GPUs and, more recently, Volta generation V100 Tensor Core GPUs. Across our global regions, these hardware-accelerated instances enable some of the most exciting use cases on Oracle Cloud Infrastructure, from oil exploration and DNA sequencing to rendering animated content for media and entertainment.

Today, we’re excited to take the next step by announcing the upcoming availability of the most powerful and newest generation GPUs with NVIDIA A100 Tensor Core GPU instances across Oracle Cloud Infrastructure’s global regions. NVIDIA A100 is the first elastic, multiple-instance GPU that unifies training, inference, high-performance computing (HPC), and analytics. The new NVIDIA A100 will help our customers unlock even more value from their data and innovate faster.

NVIDIA A100 Tensor Core GPU Instances


The new instances will also feature the latest 2nd Gen AMD EPYC processors running at 2.9 GHz with up to 64 physical cores, along with local NVMe for low latency data access for workloads that rely on heavy checkpointing.

Additionally, these GPU instances will be the first on Oracle Cloud Infrastructure to support cluster networking. Cluster networking is our high-speed RDMA interconnect that lets customers run MPI workloads with latencies of under 2 microseconds. In the case of our new NVIDIA A100 GPU offering, these instances will have over 1.6 Tbps of bandwidth!

We’ll write more about this in a separate post, but imagine the possibilities of hundreds of these powerful GPUs in a single cluster, at your fingertips. You can create clusters of GPU bare metal instances for large-scale distributed workloads such as AI training or large-scale HPC simulations such as Computational Fluid Dynamics (CFD) or Crash. And, you can mix and match these instances with existing cluster networking-enabled HPC CPU instances for even more flexibility and choice. This heterogeneous infrastructure opens a whole new set of use cases not possible on other cloud providers today.

These new instances will be available on demand during summertime in our EMEA, APAC, and US regions for US$3.05 per GPU hour, adding to our portfolio of GPU instances that serve every price point and use case today.

Oracle GPU Instance Offerings


More Collaboration with NVIDIA

In addition to our new compute instances, we’re excited about several more updates we’ve been working on with NVIDIA. These updates will grow our ecosystem and give customers many new ways to power their businesses with accelerated compute on Oracle Cloud Infrastructure.

  • NVIDIA Inception plus Oracle for Startups: We also announced today that eligible NVIDIA Inception startups can access Oracle for Startups program benefits. These benefits include up to US$100,000 in value of cloud credits that can be used on Oracle Cloud Infrastructure’s GPU portfolio of offerings such as NVIDIA V100 and the upcoming NVIDIA A100 GPUs. Existing Oracle for Startups members can also get access to Inception program benefits. Read the announcement and get started by applying to Oracle for Startups.

  • Data Science and AI Marketplace VMs: The new NVIDIA A100 GPUs bring enormous benefits to data science and AI workloads. These workloads have an ecosystem of rich open source applications and frameworks. Using the new preconfigured Oracle Marketplace VMs, data scientists can deploy and run their workloads in minutes. You can easily deploy these images from Oracle Cloud Marketplace.

  • Oracle Cloud Infrastructure Data Science: This service enables teams of data scientists to easily build, train, and manage models on Oracle Cloud in a collaborative, managed environment. The service supports Python and JupyterLab. It also supports a variety of the most popular open source packages for deep learning (such as TensorFlow, Keras, PyTorch, and MXNET), machine learning (such as scikit-learn and xgboost), and visualization libraries (such as Plotly, matplotlib). In a future release, data scientists will be able to access the new NVIDIA A100s on the service to speed up large-scale matrix calculations and parallelize large-scale machine learning and deep learning problems. Additionally, Oracle and NVIDIA have collaborated to enable real-time streaming prognostics by accelerating MSET2.

  • GraalVM: Oracle Labs, our cutting-edge research arm, has been working with NVIDIA to integrate CUDA-X libraries into GraalVM, an enterprise application development environment that’s popular with developers. This integration enables data to be efficiently shared between GPUs and GraalVM languages (R, Scala, Java, JavaScript), and allows GPU kernels to be directly launched from those languages.  With projectgrCUDA, enterprise developers can easily integrate GPU-accelerated libraries into existing software stacks.

Our Fight Against COVID-19

Most importantly, with everything else going on in the world today, we’re well aware of the part we need to play in the community. We’re working with various organizations and are involved in projects to help in the fight against COVID-19. We’re supporting Flinders University in their development of a vaccine, supporting researchers using NVIDIA Parabricks on Oracle Cloud Infrastructure, and making NVIDIA Quadro available on our Marketplace to assist users to work from home. You can find out more about what Oracle is doing beyond these projects by visiting our COVID-19 page.

Get Started

You can get started today with our free credits and Always Free services as a next step to get familiar with Oracle Cloud Infrastructure and get access to our new accelerated offerings.

We can’t wait to see what problems researchers will solve and what workloads our customers will run on our portfolio of accelerated offerings.

Stay safe, everyone.

Be the first to comment

Comments ( 0 )
Please enter your name.Please provide a valid email address.Please enter a comment.CAPTCHA challenge response provided was incorrect. Please try again.Captcha