Choose the right size for your workload with NVads A10 v5 virtual machines, now generally available

Visualization workloads entail a wide range of use cases: from computer-aided design (CAD), to virtual desktops, to high-end simulations. Traditionally, when running these graphics-heavy visualization workloads in the cloud, customers have been limited to purchasing virtual machines (VMs) with full GPUs, which increased costs and limited flexibility. So, in 2019, we introduced the first GPU-partitioned (GPU-P) virtual machine offering in the cloud. And today, your options just got wider. Introducing the general availability of NVads A10 v5 GPU accelerated virtual machines, now available in US South Central, US West2, US West3, Europe West, and Europe North regions. Azure is the first public cloud to offer GPU partitioning (GPU-P) on NVIDIA GPUs.

NVads A10 v5 virtual machines feature NVIDIA A10 Tensor Core GPUs, up to 72 AMD EPYC™ 74F3 vCPUs with clock frequencies up to 4.0 GHz, 880 GB of RAM, 256 MB of L3 cache, and simultaneous multithreading (SMT).

Pay-as-you-go, one-year and three-year Azure Reserved Instances, and Spot virtual machines pricing for Windows and Linux deployments are now available.

Flexible and affordable NVIDIA GPU-powered workstations in the cloud

Many enterprises today use NVIDIA vGPU technology on-premises to create virtual GPUs that can be shared across multiple virtual machines. We are always innovating to provide cloud infrastructure that makes it easy for customers to migrate to the cloud. By working with NVIDIA, we have implemented SR-IOV-based GPU partitioning that provides customers cost-effective options, similar to the vGPU profiles configured on-premises to pick the right-sized GPU-powered virtual machine for the workload. The SR-IOV-based GPU partitioning provides a strong, hardware-backed security boundary with predictable performance for each virtual machine.

With support for NVIDIA vGPU, customers can select from virtual machines with one-sixth of an A10 GPU and scale all the way up to two full A10 GPU configurations. This offers cost-effective entry-level and low-intensity GPU workloads on NVIDIA GPUs, while still giving customers the option to scale up to powerful full-GPU and multi-GPU processing power. Each GPU partition in the NVads A10 v5 series virtual machines includes the full NVIDIA RTX(GRID) license and customers can either deploy a single virtual workstation per user or offer multiple sessions using the Windows Enterprise multi-session operating system. Our customers love the integrated license validation feature as it simplifies the user experience by eliminating the need to deploy dedicated license server infrastructure and provides customers with a unified pricing model.

"The NVIDIA A10 GPU-accelerated instances in Azure with support for GPU partitioning are transformational for customers seeking cost-effective cloud options for graphics- and compute-intensive workloads. Now, enterprises can access powerful RTX Virtual Workstation instances accelerated by NVIDIA Ampere architecture-based A10 GPUs—sized to meet the performance requirements of creative and technical professionals working across industries such as manufacturing, architecture, and media and entertainment."— Anne Hecht, Senior Director, Product Marketing, NVIDIA.

NVIDIA RTX Virtual Workstations include the latest enhancements in AI, ray tracing, and simulation to enable incredible 3D designs, photorealistic simulations, and stunning visual effects—at faster speeds than ever.

Pick the right-sized GPU virtual machine for any workload

The NVads A10 v5 virtual machine series is designed to offer the right choice for any workload and provide the optimum configurations for both single-user and multi-session environments. The flexible GPU-partitioned virtual machine sizes enable a wide variety of graphics, video, and AI workloads—some of which weren’t previously possible. These include virtual production and visual effects, engineering design and simulation, game development and streaming, virtual desktops/workstations, and many more.

“In the world of CAD design, cost performance and flexibility are of prime importance for our users. Microsoft has completed extensive testing with Siemens NX and we found significant benefits in performance for multiple user scenarios. With GPU partitioning, Microsoft Azure can now enable multiple users to use Siemens NX and efficiently utilize GPU resources offering customers great performance at a reasonable hardware price point.”—George Rendell, Vice President Product Management, Siemens NX.

High performance for GPU-accelerated graphics applications

The NVIDIA A10 Tensor core GPUs in the NVads A10 v5 virtual machines offer great performance for graphics applications. The AMD EPYC™ 74F3 vCPUs with clock frequencies up to 4.0 GHz offer impressive performance for single-threaded applications.

Next steps

For more information on topics covered here, see the following documentation:

Source: Azure Blog Feed

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.