Microsoft Windows 10 Hardware-Accelerated GPU Scheduling Get NVIDIA And AMD Support

Microsoft Windows 10 had received an important feature that modern-day GPUs can utilize and benefit the GPU runtimes. The Windows 10 May 2020, 20H1 v2004 Cumulative Update contains the new Hardware-Accelerated GPU Scheduling feature, which remained ‘Off’ by default. However, with NVIDIA and now AMD lending its support to the feature, Windows 10 OS users with dedicated graphics cards or discrete GPUs should switch the feature ‘On’.

With the Windows 10 May 2020 Update, Microsoft introduced a new GPU Scheduler. However, the company has intentionally left the setting as an opt-in. In other words, the settings continue to remain and off by default via a toggle button in graphics settings. However, with NVIDIA and AMD now supporting the feature on their GPUs, it is perhaps the right time to turn on this “significant and fundamental change to the driver model”.

AMD Follows NVIDIA And Adds Support For GPU Scheduling In Radeon Software Adrenalin 2020 Edition 20.5.1 Beta driver:

AMD has officially added support for GPU Scheduling in its Radeon Software Adrenalin 2020 Edition 20.5.1 Beta driver. It is important to note that the software, as well as the feature, is still experimental. Hence Microsoft has kept the feature off by default.

AMD Radeon RX 5600 and Radeon RX 5700 series graphics cards have the ability to take over the duties of scheduling GPU usage. In other words, the new feature has transferred the responsibility of scheduling GPU usage and runtimes from a software platform, directly onto the compatible or supporting GPU installed in the computer.

NVIDIA recently announced its support for Windows graphics scheduling. However, Microsoft’s new DirectX Ultimate graphics API getting support GeForce RTX GPU put the announcement behind. Microsoft has cautioned that the GPU Scheduling feature will need to undergo a few more rounds of testing before it is switched on by default.

Windows 10 May 2020 Update Installers Can Experience Reduced Latency Caused By Buffering Between The CPU And GPU?

Microsoft is confident that enabling GPU Scheduling on supporting graphics cards should significantly reduce overhead for GPU scheduling. Simply put, users can experience improved GPU responsiveness. Moreover, this should allow additional innovation in GPU workload management in the future.

Hardware-accelerated GPU scheduling is available in Windows 10 version 2004. The feature was enabled by the Windows Display Driver Model (WDDM) v2.7 driver in this version of Windows 10. As the setting is off by default, users must opt-in to it in Settings -> System -> Display -> Graphics Settings. It is important to note that not all PCs running the latest Windows 10 20H1 or v2004 Cumulative update will have the setting. The core interface of the setting will only appear if the GPU and GPU driver supports the GPU scheduler.

Hardware GPU Scheduling was introduced way back in Windows Vista. It is a software component that allocates workload from multiple sources onto a GPU. Alternatively, all applications that needed GPU-acceleration would send as much traffic as they could to the GPU driver. GPU scheduling is similar to the OS thread scheduler as it selectively or sequentially allocates workloads, and essentially, doesn’t overburden GPU driver with simultaneous dumps of tasks.

Some of the newer generations of GPUs by NVIDIA, AMD, as well as Intel, have a dedicated hardware component in-built to perform scheduling. With the introduction of the setting, Windows offloads GPU scheduling duties onto the hardware component. Essentially, shifting from software to hardware-based GPU scheduling should free up some CPU resources and potentially reduce latencies at various stages of the graphics rendering pipeline.

It is interesting to note that Microsoft intends to evolve more in the direction of reducing latencies by offloading the CPU in future versions of Windows and DirectX. This is completely opposite to the company’s original direction in which it favored host-signal processing instead of hardware-accelerated signal processing.


Alap Naik Desai

A B.Tech Plastics (UDCT) and a Windows enthusiast. Optimizing the OS, exploring software, searching and deploying solutions to strange and weird issues is Alap's main interest.