AMD recently had announced support for GPU-accelerated machine learning (ML) training workflows for Windows 10 users, thus enabling users with AMD hardware that varies from software engineers to students to access the machine learning training workflows and enhance their skills on the same PCs used for everyday tasks.
According AMD community site, this was done due to an increased in demand to learn the fundamentals of machine learning by both software engineers and students. Due to its limited workflows and tools, an operating system like Linux was needed and this results in needing a separate hardware from their main machine.
Microsoft engineers come up with their own solution by adding GPU-accelerated compute support to Windows 10 operating system starting from Build 2004. Generally known as Windows Subsystem for Linux (WSL), it basically allows the use of Linux command-line applications side-by-side with native Windows 10 application. According to Microsoft further, this update helps to lower down the restriction from sharpening skills on machine learning and get used with both environments, and this includes with hardware powered by AMD.
According to the community site further, the key components of the update is the use of the Microsoft DirectML API. It works in the enabled GPU-accelerated ML inference on any Direct 12 compatible GPUs, this includes AMD Radeon and Radeon Pro GPUs. Microsoft is working on to integrating DirectML into the most used machine learning tools, libraries and frameworks, thus ensuring the DirectML works perfectly within WSL environment.
Next, Microsoft premiere the feature to members of the Windows Insider Program. This includes, build of TensorFlow with a DirectML backend, students and beginners can opt their ML learning journey on Windows 10 along with the TensorFlow tutorial models or with the examples of library released by Microsoft.
AMD on the other hand, released a preview Radeon Software Adrenalin 2020 Edition driver that supports DirectX 12 compatible GPU-acceleration within the WSL. The driver is fully compatible with AMD Radeon graphics, Radeon Pro graphics, Ryzen and Ryzen PRO desktop processors with Radeon Vega graphics, as well as Ryzen and Ryzen PRO mobile processors with Radeon Vega graphics. AMD also prepared a list of compatible AMD products for public view.
Microsoft has released a video on GPU-accelerated ML training by using an AMD Ryzen Microsoft Surface Edition processor in the Microsoft Surface laptop and the video can be viewed here. If you are interested in trying out on your AMD hardware, you will be needing to download their preview driver and follow the instruction outline in Microsoft’s article along with their guidance in their documentation.