Oracle Cloud is racing ahead at NVIDIA GTC 2022

March 16, 2022 | 4 minute read
Andrew Butterfield
Principal Product Manager at Oracle - GPU & HPC
Text Size 100%:

Oracle Cloud Infrastructure (OCI)’s investment in building the most innovative and best price-performant cloud has enabled our customers to take on greater challenges that require advanced AI and computing capabilities.

Customers like Aleph Alpha, Children's Medical Research Institute (CMRI), and University of Oxford have seen AI innovations and tremendous performance gains with no increased cost when moving their most demanding workloads to OCI. Applications such as traditional high-performance computing (HPC) physics simulations, predictive analytics, and large-scale AI training, such as natural language processing (NLP), computer vision, and anomaly detection, thrive on our NVIDIA GPUs like the A100 Tensor Core combined with our bare metal servers and high-speed, low-latency cluster networking.

As a diamond sponsor of the NVIDIA GPU Technology Conference (GTC) held March 21–24, 2022, Oracle is sharing what some of our customers are doing. We’re hosting several on-demand sessions that showcase how customers are using OCI to address their challenges.

The world is loud

With social media and online news sources, the publication of ideas and responses is tremendously voluminous. Analysts and innovators that are monitoring social news and sentiment must lean into machine learning to interpret and scale up language models and new capabilities, such as alternative data platforms. Through state-of-the-art summarization systems and optimized architectural choices, we can build better models that allow posts, news, and sentiment to be processed with better accuracy, precision, and understanding of context.
Accuracy for natural language processing is still a challenge that can only be achieved with the support of the largest and most capable cloud infrastructure. Startups, such as Latent Space, and researchers from University of Oxford and University of Michigan have achieved success by using OCI, which integrates bare metal NVIDIA GPUs and high-speed, low-latency RDMA ROCE v2 cluster networking.

Enabling machine learning at the edge

Machine learning internet of things (IoT) use cases involve thousands of sensor signals, which require heavy demand for cloud resources. One challenge for all cloud companies who seek to deal with big data use cases is peak memory utilization that scales nonlinearly with the number of sensors and autonomously sizing cloud shapes properly before the program run. This process is complicated. Anomaly detection MSET2 technology, combined with NVIDIA A100 Tensor Core GPUs, has given developers the tools that they need to take on the largest machine learning challenges.

Physics simulation 700 times faster

Simulating the physics and flow of particles as they move through space is a common HPC application. This type of simulation allows researchers to fine-tune efficiencies and model how molecules interreact with each other in drug discovery research. This application requires compute-intensive capabilities and has traditionally been performed on CPU-based HPC clusters. With the advances of NVIDIA A100 Tensor Core GPUs, many traditional HPC workloads are seeing massive performance improvements. ESSS Rocky and UC Davis have both seen incredible performance by optimizing their applications to take advantage of the NVIDIA GPUs running on the high-performing Oracle Cloud Infrastructure.

OCI at GTC

To learn more about the work being done on these topics, join Oracle at NVIDIA GTC 2022, where Oracle is hosting four virtual sessions. All sessions this year are free of charge. Register today!

S42537 Autonomous Memory Sizing Formularization for Cloud-based IoT ML Customers
 Presented by:

  • Guang Wang (Presenter), Oracle Labs
  • Jason Ding (Copresenter), OCI
  • Kenny Gross, Oracle Labs
  • Prasad Ballingam, OCI
  • Syed Fahad Allam Shah, OCI

S42539 Accelerate NVIDIA GPU Simulations with HPC on Oracle Cloud
Presented by:

  • David Chen, master principal cloud architect, HPC OCI   
  • Vinicius Daroz, CAE applications specialist, Engineering Simulation and Scientific Software (ESSS)
  • Igor Vorobyov, Ph.D., assistant professor School of Medicine, University of California, Davis

S42629 AI Innovations: Three Case Studies on Successful Natural Language Processing
Presented by:

  • Dr. Burcin Kaplanoglu, VP, Oracle Innovation Lab
  • Dr. Adam Saunders, University of Oxford
  • Dr. Gatis Mikelsons, University of Oxford
  • Shuyang Cao, Researcher, University of Michigan
  • Darryl Barnhart, Cofounder, Latent Space

S41955 Arm’s performance and impact on HPC in the cloud
Presented by:

  • Kevin Jorissen

Racing ahead with HPC on OCI

For more information on how ESSS optimized their 200 million particle simulation for GPU workloads, see ESSS Rocky’s 200 million particle simulation on Oracle Cloud pushes DEM analysis to a new height.
 
For more information on OCI and NVIDIA, visit the NVIDIA and Oracle Cloud Infrastructure GPU Cloud Platform page.

Andrew Butterfield

Principal Product Manager at Oracle - GPU & HPC

Andrew Butterfield is the Product Manager for Oracle Cloud Infrastructure’s GPU and HPC offerings. He drives the product development, product launch, as well as the AI and HPC strategy.


Previous Post

All systems operational: Announcing an improved OCI status dashboard

Jessica Alspaugh | 2 min read

Next Post


OCI File Storage service is now VMware certified

Vinay Rao | 4 min read