Microsoft DirectML Powers Up: Unleashing AI on New NPU-Fueled PCs

In response to the integration of more artificial intelligence computing into the Windows operating system and the upcoming trend of AI PCs, Microsoft has announced an update to its DirectML technology next year to include support for Neural Processing Units (NPUs). This move is aimed at meeting the operational requirements of devices like Intel’s recently announced Core Ultra series processors, which feature NPU designs.

DirectML, a component of the DirectX series of technology APIs, is primarily designed for machine learning computations and can integrate with other machine learning frameworks and tools. This integration enables systems to accelerate machine learning tasks using GPUs and other computational components.

DirectML technology

Currently, DirectML is compatible with the cross-platform inference engine ONNX Runtime and the Olive tool, which is optimized for machine learning models in the Windows environment. The addition of support for NPU components means that it can leverage Intel Core Ultra series processors for more efficient machine learning computations.

Microsoft plans to release the NPU-compatible version of DirectML and the new ONNX Runtime in early 2024. With the launch of the Intel Core Ultra processors, the market will soon see an influx of AI PC products offering higher performance and better battery life. These products will be complemented by a comprehensive software and hardware ecosystem and advanced wireless connectivity technologies, enhancing the overall user experience of AI PCs.