Virtualization Technology News and Information
Microsoft Azure Previews Scalable Virtual Machines Powered by Nvidia H100 GPUs to Accelerate Generative AI


Microsoft Azure is a cloud computing platform that provides a wide range of services that help businesses meet their IT needs. And a powerful addition to this platform are its scalable virtual machines (VMs) using Nvidia H100 GPUs.

The H100 GPU is part of Nvidia's line of data center GPUs, and it is designed to accelerate workloads related to high-performance computing (HPC), artificial intelligence (AI), and machine learning (ML). With its high memory bandwidth and floating-point performance, the H100 GPU can handle complex computing tasks that traditional CPUs cannot.

By integrating the H100 GPU into Azure VMs, Microsoft is enabling its customers to take advantage of this powerful computing resource on demand. Whether it is for running complex simulations, training machine learning models, or analyzing large datasets, Azure customers can scale up or down their GPU-accelerated VMs as needed to meet their workload requirements.

One of the main benefits of using Azure VMs with H100 GPUs is scalability. Azure customers can provision VMs with a single H100 GPU or scale up to multiple GPUs as their workload demands increase. This flexibility allows businesses to optimize their computing resources and only pay for what they need.

Another advantage of using Azure VMs with H100 GPUs is the ease of deployment. Azure customers can spin up a GPU-accelerated VM in minutes, and they can use pre-configured templates to deploy popular AI and ML frameworks such as TensorFlow, PyTorch, and MXNet. This makes it easy for data scientists and developers to get started with GPU-accelerated computing without having to worry about hardware infrastructure.

In addition to scalability and ease of deployment, Azure VMs with H100 GPUs offer robust security features. Azure provides a range of security measures to protect data and workloads, including network security groups, virtual private networks (VPNs), and role-based access control (RBAC). These security features ensure that customers can run their GPU-accelerated workloads with confidence, knowing that their data is protected.

This week, Microsoft introduced the ND H100 v5 VM which enables on-demand in sizes ranging from eight to thousands of NVIDIA H100 GPUs interconnected by NVIDIA Quantum-2 InfiniBand networking. Microsoft said customers will see a significantly faster performance for AI models over their last generation ND A100 v4 VMs with innovative technologies like:

  • 8x NVIDIA H100 Tensor Core GPUs interconnected via next-gen NVSwitch and NVLink 4.0
  • 400 Gb/s NVIDIA Quantum-2 CX7 InfiniBand per GPU with 3.2Tb/s per VM in a non-blocking fat-tree network
  • NVSwitch and NVLink 4.0 with 3.6TB/s bisectional bandwidth between 8 local GPUs within each VM
  • 4th Gen Intel Xeon Scalable processors
  • PCIE Gen5 host to GPU interconnect with 64GB/s bandwidth per GPU
  • 16 Channels of 4800MHz DDR5 DIMMs 
And while Azure hasn't made any specific claims as to what that performance increase would look like, NVIDIA has advertised H100 as running up to 30x faster than the preceding Ampere architecture that is currently offered with the ND A100 v4 VMs.
"NVIDIA and Microsoft Azure have collaborated through multiple generations of products to bring leading AI innovations to enterprises around the world," said Ian Buck, Vice President of hyperscale and high-performance computing at NVIDIA. "The NDv5 H100 virtual machines will help power a new era of generative AI applications and services."
Microsoft Azure has opened up the preview of the ND H100 v5 VM service and you can sign up to request access here.
Published Tuesday, March 14, 2023 3:04 PM by David Marshall
Filed under: ,
There are no comments for this post.
To post a comment, you must be a registered user. Registration is free and easy! Sign up now!
<March 2023>