At Oracle OpenWorld, NVIDIA and Oracle Bringing AI to the Enterprise

by Ikroop Dhillon

Oracle is pushing the envelope on what NVIDIA GPUs can do in the cloud.

Find out how, next week at Oracle OpenWorld and Code One in San Francisco, where NVIDIA and Oracle will showcase their growing collaboration to bring AI and GPU-accelerated applications to the enterprise.

Integrating CUDA-X libraries into GraalVM applications, enhancing conversational AI with Oracle Digital Assistant, and accelerating data science pipelines through the Oracle Cloud Infrastructure Data Science service are a few examples of how enterprise customers and developers worldwide will benefit from GPU-accelerated computing.

The companies first teamed up by bringing bare metal GPUs to the public cloud through Oracle Cloud Infrastructure, fueling innovation across a broad range of industries. Engineers, developers, data scientists and researchers are using these instances to power visualization, AI/machine learning, big data, database and HPC workloads.

“We’re working with NVIDIA to bring GPU-accelerated computing to market on our cloud platform and address the demands of our largest customers for stability and performance,” said Karan Batta, director of product management for Oracle Cloud Infrastructure, at Oracle. “Bare metal computing combined with the cloud and NVIDIA GPUs gives enterprise customers unprecedented power and flexibility.”

Attendees of Oracle OpenWorld and CodeOne 2019 should check out the following highlights of our collaboration.

Bringing NVIDIA GPU Acceleration to GraalVM 

Learn how we’re integrating the rich ecosystem of CUDA-X libraries for developers into applications created with GraalVM languages. By creating a polyglot binding for GPUs, we’re enabling GPU kernels to be directly launched from GraalVM languages such as R, JavaScript, Scala and other JVM-based languages.

This new capability will let customers more easily unlock their data and integrate machine learning and deep learning into applications. For example, financial customers could increase the speed and accuracy of risk predictions and ecommerce software applications will be able to deliver faster insights to customers.

Simplifying NVIDIA GPU Access: A Polyglot Binding for GPUs with GraalVM [DEV6097] | Wednesday, Sept. 18, 6-6:45 p.m. | Moscone South – Room 314

Enabling Conversational AI

Conversational AI models must be able to replicate human-like interaction and provide near real-time responses. Yet these models are massive and highly complex as they must understand context, complex search patterns and provide intelligent responses in real time. Learn how Oracle Digital Assistant is using NVIDIA GPUs to accelerate conversational AI and deliver efficient and responsive speech recognition, training and inference.

Oracle Speaks: Conversational AI for Enterprise Apps with a Chatbot Assistant [GEN5607] | Tuesday, Sept. 17, 1:45-2:30 p.m. | Moscone South – Room 209

Real-Time Streaming Prognositics in Oracle Cloud

Low latencies and high throughputs are critical to achieving real-time streaming prognostics, particularly in cloud environments. Learn how NVIDIA GPU acceleration of MSET2, Oracle’s advanced machine learning pattern recognition innovation, yields substantial reductions in latency and enhancements to throughput rates in the Oracle Cloud.

This work will enable customers in dense-sensor IoT industries such as utilities, smart manufacturing, oil and gas, and commercial aviation to harness vast amounts of data from plants and physical assets to gain valuable prognostic insights.

NVIDIA GPU Acceleration of Autonomous Anomaly Detection for Dense-Sensor IoT Prognostics [DEV6645] | Monday, Sept. 16, 12:30-1:15 p.m. | Moscone South – Room 206

Accelerating Machine Learning on OCI Data Science

The Oracle Cloud Infrastructure Data Science service enables data scientists to efficiently collaborate and build, train and deploy machine learning models on Oracle Cloud.  Soon, with NVIDIA GPU shapes on the service, along with RAPIDS, a suite of software libraries that enable end-to-end execution of data science pipelines, enterprise data scientists will be able to iterate on models that are orders of magnitude faster and dramatically accelerate model deployment.

Deep Dive on Oracle Data Science Cloud [PRO5225] | Tuesday, Sept. 17, 11:15 a.m.-12 p.m. | Moscone South – Room 212

Find out more about how developers, data scientists and customers can take advantage of these new capabilities. Register for Oracle OpenWorld and Code One and get started with GPUs in the Oracle Cloud today.

For analysis on this topic, read the blog from Futurum Research’s Daniel Newman.