Discrete Device Assignment (DDA)ĭiscrete Device Assignment (DDA), also known as GPU pass-through, allows you to dedicate one or more physical GPUs to a virtual machine. For more information, see GPU Acceleration in Windows containers. In addition to VM workloads, Windows Server also supports GPU acceleration of containerized workloads within Windows Containers. The following graphics virtualization technologies are available to Hyper-V VMs in Windows Server: However, if your workload runs directly on physical Windows Server hosts, then you have no need for graphics virtualization your apps and services already have access to the GPU capabilities and APIs natively supported in Windows Server. If your workload is virtualized with Hyper-V, then you'll need to employ graphics virtualization in order to provide GPU acceleration from the physical GPU to your virtualized apps or services. GPU virtualization technologies enable GPU acceleration in a virtualized environment, typically within virtual machines. Many HPC and machine learning frameworks have an option to enable GPU acceleration consider whether this might benefit your specific workload. Alternatively, they may offer better cost-effectiveness than a CPU-only architecture at a comparable performance level. HPC and ML workloads: For highly data-parallel computational workloads, such as high-performance compute and machine learning model training or inference, GPUs can dramatically shorten time to result, time to inference, and training time.
For this kind of workload, a single GPU-enabled VM may be able to match the throughput of many CPU-only VMs. Remote rendering, encoding, and visualization workloads: These graphics-oriented workloads tend to rely heavily on a GPU's specialized capabilities, such as efficient 3D rendering and frame encoding/decoding, in order to achieve cost-effectiveness and throughput goals.Most customers consider GPUs a necessity for a reasonable user experience with these kinds of apps. Some types of apps, such as CAD/CAM apps, simulation apps, games, and rendering/visualization apps, rely heavily on 3D rendering to deliver smooth and responsive interactivity. App and desktop remoting (VDI/DaaS) workloads: If you're building an app or desktop remoting service with Windows Server, consider the catalogue of apps you expect your users to run.Here's what you should consider before choosing GPU acceleration: When to use GPU accelerationĭepending on your workload, you may want to consider GPU acceleration. This article introduces the graphics virtualization capabilities available in Windows Server. Plan for GPU acceleration in Windows ServerĪpplies to: Windows Server 2022, Windows Server 2016, Microsoft Hyper-V Server 2016, Windows Server 2019, Microsoft Hyper-V Server 2019
0 kommentar(er)
