Ignorez la navigation

Microsoft shares what's next in machine learning at NVIDIA GTC

Publié le 14 septembre, 2022

Sr. Product Marketing Manager, Azure Marketing

Placeholder

Finding scalable solutions for today’s global challenges requires forward-thinking, transformative tools. As environmental, economic, and public health concerns mount, Microsoft Azure is addressing these challenges head on with high-performance computing (HPC), AI, and machine learning. The behind-the-scenes power for everything from MRI scans to energy management and financial services, these technologies are equipping customers and developers with innovative solutions that break through the boundaries of what’s possible in data and compute, paving the way for growth opportunities that span industries and applications around the world.

Microsoft Azure is committed to unlocking these new opportunities for our customers, providing the broadest range of NVIDIA GPUs at the edge, on-premises, in the cloud, and for hybrid environments.

At NVIDIA GTC we will demonstrate this commitment by showing how Azure’s advanced HPC capabilities, and AI/machine learning in the cloud are driving transformation and making an impact together with NVIDIA’s latest technology.

Microsoft Azure’s collaboration with NVIDIA was developed with our customers in mind and focused on opening new doors to innovation with graphics processing unit (GPU) acceleration in the cloud.

Learn more by registering today for NVIDIA GTC, a free, online event running September 19 to 22, 2022.

Placeholder

Get a chance to win an NVIDIA Jetson Nano or swag box

In both of our sessions you have a chance to win a SWAG box complete with a HPC t-shirt and mug or a Jetson Nano. Attend these sessions and don’t forget to look for the special link to enter!

Microsoft Sessions at NVIDIA GTC

The new SDK and CLI in Azure Machine Learning.
Bala Venkataraman, Principal Program Manager, Microsoft.

Video on demand

Azure Machine Learning is committed to simplifying the adoption of its platform for training and production. In 2022, we announced the general availability of Azure Machine Learning CLI v2 and the preview of Azure Machine Learning Python SDK v2. Both launches demonstrate our continued focus on making workflows easier and managing their entire lifecycle starting from training single jobs to pipelines and model deployments. In this session, learn about the key improvements in usability and productivity, and the new features that come with the command-line interpreter (CLI) and software development kit (SDK) v2 of Azure Machine Learning.

Operationalize large model training on Azure Machine Learning using multi-node NVIDIA A100 GPUs.
Sharmeelee Bijlani, Program Manager Azure Machine Learning, Microsoft; Razvan Tanase, Principal Engineering Manager Azure Machine Learning, Microsoft.

Wednesday, September 21, 10:00 to 10:50 AM PDT (1:00 to 1:50 PM EDT, 7:00 to 7:50 AM CEST)

In recent years, deep learning models have grown exponentially in size, demonstrating an acute need for customers to train and fine-tune them using large-scale data infrastructure, advanced GPUs, and an immense amount of memory. Fortunately, developers can now use simple training pipelines on Azure Machine Learning to train large models running on the latest multi-node NVIDIA GPUs. This session will describe these software innovations to customers through Azure Machine Learning (including a fully optimized PyTorch environment) that offers great performance and an easy-to-use interface for large-scale training. We’ll also highlight the power of Azure Machine Learning through experiments using 1,024 A100 Tensor Core GPUs to scale the training of a two-trillion parameter model with a streamlined user experience at 1,000 plus GPU scale.

Watch Party #1: Operationalize large-model training on Azure Machine Learning using multi-node NVIDIA A100 GPUs.
Mary Howell, NVIDIA.

Wednesday, Sep 21st, 3:00 - 3:30 PM PDT
In this GTC Watch Party, we will be replaying our Operationalize Large-Model Training on Azure Machine Learning using Multi-Node NVIDIA A100 GPUs session. Participants will be joined by experts from across Microsoft and NVIDIA who bring fresh insights and experiences to the table, taking the session to a whole new level of understanding. Interaction is core to our GTC Watch Parties, and we encourage you to join the discussion with any comments or questions. 

Watch Party #2: Operationalize large-model training on Azure Machine Learning using multi-node NVIDIA A100 GPUs.
Gabrielle Davelaar, AI Technical Specialist, Microsoft; Maxim Salnikov, Senior Azure GTM Manager, Microsoft; Henk Boelman, Senior Cloud Advocate–AI and Machine Learning, Microsoft; Alexander Young, Technical Marketing Engineer, NVIDIA; Ulrich Knechtel, Microsoft Partner Manager (EMEA), NVIDIA.

Thursday, September 22, 2:00 to 3:30 PM CEST (5:00 to 6:30 AM PDT, 8:00 to 9:30 AM EDT)

In this GTC Watch Party, we will be replaying our Operationalize Large-Model Training on Azure Machine Learning using Multi-Node NVIDIA A100 GPUs session. Participants will be joined by experts from across Microsoft and NVIDIA who bring fresh insights and experiences to the table, taking the session to a whole new level of understanding. Interaction is core to our GTC Watch Parties, and we encourage you to join the discussion with any comments or questions.

Microsoft is helping customers across industries step up, transforming AI and machine learning at the Edge

Nuance’s Dragon Ambient eXperience helps doctors document care faster with AI on Azure

Nuance developed an AI-based clinical solution that automatically turns doctor-patient conversations into accurate medical notes. Built with Azure and PyTorch, this solution saves doctors transcription time, reducing administrative burdens and helping them conduct more focused, higher-quality interactions with their patients.

Energy utility Elva builds a highly secure DevOps platform with Azure infrastructure and network security services

Elva looked to build a secure, cloud-first DevOps platform that could meet Norway’s data residency and compliance requirements, delivering automated services that would help develop network grid technology. Using Azure DDoS Protection, Azure Web Application Firewall, and Azure Kubernetes Service, Elva realized its goal, enhancing its in-house development and data integration capabilities. 

The Royal Bank of Canada creates personalized offers while protecting data privacy with Azure confidential computing

The Royal Bank of Canada (RBC) partnered with Microsoft to create a privacy-preserving multi-party data sharing platform built on Azure confidential computing. Called VCR, this solution enables RBC to personalize offerings and protect privacy at the same time, creating exceptional digital experiences that clients can trust.  

Placeholder

Recapping 2022 moments with Azure and NVIDIA technologies

Azure NC A100 v4-series

At Microsoft, our NC series virtual machines allow customers access to almost limitless AI hardware infrastructure so they can be productive quickly. Last summer, we leveled up, announcing the general availability of Azure NC A100 v4 series virtual machines. Powered by NVIDIA A100 80GB PCle Tensor Core GPUs and 3rd Gen AMD EPYC™ processors, these virtual machines help our customers gain insights faster, innovate with speed, do more with less, and are the most performant and cost-competitive NC series offering for a diverse set of workloads.

DeepSpeed on Azure

Azure Machine Learning uses large fleets of the latest NVIDIA GPUs powered by NVIDIA Quantum InfiniBand interconnects to tackle large-scale AI training and tuning. Last July, we announced a breakthrough in our software stack, using DeepSpeed and 1,024 NVIDIA A100 GPUs to scale the training of a two trillion parameter model with a streamlined user experience at 1,000 plus GPU scale. We are bringing these software innovations to you through Azure Machine Learning (including a fully optimized PyTorch environment) that offers great performance and an easy-to-use interface for large-scale training.

NVads A10 v5 virtual machines

Traditionally, graphics-heavy visualization workloads that run in the cloud require virtual machines with full GPUs that are both costly and inflexible. To combat this, we introduced the first GPU-partitioned (GPU-P) virtual machine offering in the cloud, and just last July, we announced the general availability of NVads A10 v5 GPU accelerated virtual machines. Azure is the first public cloud to offer GPU partitioning on NVIDIA GPUs, and our new NVads A10 v5 virtual machines are designed to offer the right choice for any workload and provide optimum configurations for both single-user and multi-session environments. Dig into our latest virtual machine innovation.

NVIDIA Jetson AGX Orin-powered edge AI devices now available

Microsoft is pleased to announce that the NVIDIA Jetson AGX Orin SoM is now powering Azure Certified edge devices from industry-leading device builders including AAEON, Advantech, and AVerMedia, along with the NVIDIA Jetson AGX Orin developer kit.

Developers and solution builders can now leverage powerful NVIDIA Jetson AGX Orin devkits and production modules with Microsoft Azure to create, deploy, and operate powerful AI solutions at the edge, accelerating product development and deployment at scale. The NVIDIA Orin Nano modules have set a new baseline for entry-level edge AI and robotics, building on the momentum behind the Jetson Orin platform worldwide. Stay tuned for new Jetson Orin NX and Orin Nano partner products launching to meet customer needs in AI solution development.

NVIDIA DLI training powered by Azure

We’re proud to host NVIDIA deep learning institute (DLI) training at NVIDIA GTC again this year, with instructor-led workshops around accelerated computing, accelerated data science, and deep learning. Hosted on Microsoft Azure, these sessions enable and empower you to leverage NVIDIA GPUs on the Microsoft Azure platform to solve the world’s most interesting and relevant problems. Register for a DLI workshop today.

Join us at NVIDIA GTC

In collaboration with NVIDIA, Microsoft delivers purpose-built AI, machine learning, and HPC solutions in the cloud to meet even the most demanding real-world applications at scale. Join us at NVIDIA GTC September 19 to 22, to see how every enterprise can leverage the power of GPUs at the edge, on-premises, in the cloud, and for hybrid solutions.

Learn more