GPU acceleration, AI and visualization technology are heading all the way from the Microsoft Azure cloud, to the data center and to the edge. Businesses in industries such as retail, manufacturing and public safety are already bringing Azure’s capabilities to the edge with Azure Stack Edge, which Microsoft is making generally available today.
At Microsoft Ignite, RXR Realty, the third largest real estate owner in New York City, is showcasing how it has been working with key partners and used Microsoft Azure to create and deploy an intelligent, secure, hyperscalable solution in just a few months. The solution, which is designed to help tenants stay safe after buildings reopened for business, uses Azure Stack Edge Pro, powered by NVIDIA T4 Tensor Core GPUs and Azure AI’s spatial analysis, a new computer vision capability within Azure Cognitive Services.
And JRCS Company, of Japan, uses live video analytics on Azure Stack Edge with NVIDIA T4 GPUs on maritime vessels to detect objects and automatically provide collision alerts for ship captains. (Watch in closed caption, an interview with JRCS in English subtitles and Japanese.)
NVIDIA’s collaboration with Microsoft is expanding with Azure Stack Hub, a platform that extends Azure, bringing the agility and innovation of cloud computing to customers’ on-premises environments. Customers can build, deploy and run hybrid and edge computing apps consistently across their IT ecosystems, with flexibility for diverse workloads.
Azure Stack Hub supports Quadro Virtual Workstations powered by NVIDIA T4 Tensor Core GPUs, which enable professional visualization workloads such as computer-aided design, rendering, content creation and simulation for remote work environments.
Esri ArcGIS, industry-leading software used for mapping and spatial analytics, runs on Azure Stack Hub with the NVIDIA T4 GPU and Quadro Virtual Workstation. It uses a deep learning model for inferencing to automate the detection of damaged California homes from the Woolsey fires, and visualize the results with Quadro Virtual Workstation all in real time on any device, anywhere.
Inferencing and machine learning workloads can also be run on Azure Stack Hub with either NVIDIA T4 or V100 GPUs. One example is Addenbrooke’s Hospital’s recent collaboration with Microsoft Azure Stack Hub to demonstrate an inferencing pipeline that accelerates a key part of cancer treatment for patients.
The robust inference pipeline code, based on Microsoft’s Project InnerEye components, will enable the hospital to run containerized models on virtual machines, automating the selection and inference of the correct AI model for each CT scan.
“The collaboration between Project InnerEye and Addenbrooke’s Hospital exemplifies what can be achieved when NVIDIA’s technology is applied to solve real-world challenges in medical imaging,” said Mona Flores, global head of Medical AI at NVIDIA.
Solving the issue of latency, intermittent connectivity and security regulation, Azure Stack can help enterprises improve:
- Edge solutions: Azure Stack processes data locally, which eliminates latency and connectivity requirements.
- Cloud applications and regulations: Develop and deploy applications in Azure and deploy on-prem with Azure Stack Hub to meet any regulatory requirements.
- Cloud applications on-premises: Flexibility in using IaaS and PaaS services, containers to modernize existing apps or build new ones from the ground up in Azure Stack Hub, without compromising consistencies and the DevOps processes.
- GPU acceleration: Azure Stack Hub and Azure Stack Edge bring acceleration to professional Learn more about Azure Stack Hub for professional visualization, Azure Machine Learning and Azure Cognitive Services at Microsoft Ignite, Sept. 22-24: visualization and machine learning workloads to where the data is collected and generated.
- Software deployments: NVIDIA NGC catalog for GPU-optimized AI software and pre-trained models simplifies and accelerates end-to-end workloads.
Customers can train in the cloud with Azure Machine Learning and deploy on the edge for inference. Our recent work on accelerating AI for COVID-19 on Microsoft Azure Machine Learning using Clara Imaging from NVIDIA NGC shows the power of combined technology.
- Session OD201 | What’s New with the Azure Stack Portfolio | Natalie Mackevicius
- Session DB106 | What’s New in Azure Cognitive Services | Seth Juarez, Cory Clarke
- Session OD204 | New to Cognitive Services: Spatial Analysis and Metrics Advisor | Adina Trufinescu, Qun Ying
- Session DB120 | Fast-Track Development of Production-Ready ML Models with Azure Machine Learning | Aniththa Umamahesan, Lu Zhang, Vijai Kannan
NVIDIA and Microsoft are working with many startups, research institutes and hospitals around the world to find ways to tackle global healthcare challenges. ImmunityBio, Whiteboard Coordinator, University of California Riverside and IRCCS San Raffaele Hospital in Italy are just a few of our innovative partners.
We’ve also announced our latest, most advanced GPU, the NVIDIA A100, is now available for beta on Azure. Based on the NVIDIA Ampere architecture, NVIDIA A100 delivers the greatest generational leap ever, boosting AI performance by up to 20x over its predecessor. It unifies AI training and inference and will further propel developers, researchers and engineers to do their life’s work faster.
Learn more about NVIDIA and Microsoft’s collaborations.