Oracle Cloud Infrastructure (OCI) continues to enable its customers to make breakthroughs in numerous industries, such as telecommunications, manufacturing, autonomous driving, and medical research. We offer the best price-performance for NVIDIA GPU-accelerated computing in the cloud, enabling enterprises to utilize the data that exists in your Oracle apps and databases and making it easy to import your data with free inbound data transfer.
This year at the NVIDIA GPU Technology Conference (GTC), we’re highlighting the services that allow customers to extract value out of their data, the products that enable data to be captured and processed at the edge, and the advancements teams are making on top of our Generation 2 Cloud.
Large workloads using our NVIDIA GPU shapes, such as our BM.GPU4.8, are being utilized for training AI models in telecommunications, HPC simulation, autonomous driving, and medical research. BM.GPU4.8 is stacked with eight NVIDIA A100 Tensor Core GPUs and interconnected by our high-speed, low-latency RDMA cluster networking.
MIT’s Kylie Ying and Zhijian Liu and Carnegie Mellon’s Nayana Suvarna have been working with OCI as they build out their autonomous driving stack to compete in the Indianapolis Autonomous Challenge (IAC). This competition tests the limits of autonomous driving by pushing the vehicles to perform a 20-lap race at speeds of up to 180 mph.
To achieve this feat, the team needed to collect massive amounts of sensor data, train deep neural networks on the latest generation of NVIDIA A100 Tensor Core GPUs on Oracle Cloud, and test those trained models on the track. Advancements in this field are crucial in developing higher-performing and safer vehicles for the public.
John Hopkins University’s Tanmay Nath has been working with OCI as he develops robust brain models, using convolutional neural networks (CNN) based on ResNet Architectures to understand the neurophysiological underpinnings of pain. This level of research uses massive amounts of neuroimaging data from functional MRI scans. It also uses activation maps extracted from those scans, which are used to train a 3D-CNN model across NVIDIA V100 Tensor Core GPUs and NVIDIA A100 Tensor Core GPUs on OCI.
With all of this data and deep learning training, Tanmay has been able to create a generalized model for predicting pain in patients. This research can help to change the way we understand and treat people more effectively who are dealing with pain.
Worldwide data is expected to hit 175 zettabytes by 2025. The need for organizations and enterprises to effectively extract value from their data becomes more important by the day. With Oracle’s Accelerated Data Science (ADS) SDK, you can build and evaluate higher-quality machine learning (ML) models.
Enabling customers to increase business flexibility by putting enterprise-trusted data to work quickly and support data-driven business objectives with easier deployment of ML models. The ADS SDK allows data scientists to speed up their workflows by automating common tasks, such as exploratory data analysis, feature engineering, algorithm selection, and tuning.
Oracle Roving Edge Infrastructure is a new cloud-integrated service that extends the power of the cloud beyond the data center, allowing organizations to run selected cloud capabilities in remote and austere environments. Oracle Roving Edge Infrastructure extends existing customers’ OCI tenancy by physically putting customer infrastructure and platform services where data is generated on the edge or beyond.
A single Oracle Roving Edge Device (RED) contains 40 Intel Xeon Gold CPU cores, 512 GB of RAM, 61-TB NVMe storage, and an NVIDIA T4 Tensor Core GPU. REDs conform to military-grade specifications for shock, vibration, and temperature and include ingress protection, operating and non-operating altitudes, and electromagnetic interference (EMI) shielding. The devices are built with tamper evident seals and tamper-resistant enclosures to support the hardware security requirements of FIPS 140-2 Level 2.
Having this extension of Oracle Cloud at the edge is great, but we’ve taken it a step further. We paired it with Oracle's MSET2 prognostic ML algorithm, implemented with NVIDIA T4 Tensor Core GPUs. This pairing attains unprecedented reductions in computational latencies and breakthrough throughput acceleration factors for large-scale ML streaming prognostics from dense-sensor fleets of assets in the following fields:
U.S. Department of Defense assets
Oil and gas
Prognostic cybersecurity for data center IT assets
Department of Defense supervisory control and data acquisition assets and networks
To learn more about the work being done on these topics, join Oracle at NVIDIA GTC 2021, where Oracle is hosting four virtual sessions. All sessions this year are free of charge. You only have to register.
SS33049: Oracle’s Guang Wang and Dan Itkis present “MSET2 streaming prognostics for IoT telemetry on Oracle Roving Edge Infrastructure” on April 15 at 9:00 a.m. PT.
SS33052: Oracle’s Lyudmil Pelov presents “Accelerate your Data Science workflow with Oracle Cloud Data Science Accelerated Data Science SDK.”
SS33053: MIT’s Kylie Ying and Zhijian Liu and Pitt Robotics Team Captain Nayana Suvarna present “Novel challenges in autonomous vehicle development for motorsports.”
SS33075: Oracle’s Rajib Ghosh and Johns Hopkins University’s Tanmay Nath present “High-dimensional mediation using deep learning: Understanding the stimulus-pain relationship in humans.”