- Advertisement -Newspaper WordPress Theme
HardwareMicrosoft Introduces Preliminary Support for Intel Neural Processors in DirectML API

Microsoft Introduces Preliminary Support for Intel Neural Processors in DirectML API

Enhancing AI Acceleration with Intel Core Ultra Chips

In a strategic move, Microsoft has unveiled a preliminary version of the DirectML API, incorporating support for Intel’s Neural Processing Units (NPU), specifically tailored for Intel Core Ultra processors. This development aligns with Intel’s release of Core Ultra chips featuring embedded NPUs, designed to hardware-accelerate artificial intelligence applications.

DirectML 1.13.1 and ONNX Runtime 1.17: Powering NPU Acceleration

The preliminary release, comprising DirectML version 1.13.1 and ONNX Runtime version 1.17, is set to facilitate NPU acceleration on PCs equipped with Intel Core Ultra processors. Microsoft’s collaboration extends beyond Intel, with a strategic partnership formed with Samsung to enhance the performance of AI applications leveraging DirectML.

Intel and Samsung Collaboration: A Boost for AI Applications

Intel reports that Samsung has leveraged Microsoft DirectML and Intel AI Boost technology in its recently launched Samsung Galaxy Book 4 laptop featuring Intel Core Ultra chips. Samsung successfully transferred face and object classification functions within its Windows Gallery app to the Intel Core Ultra processor with its NPU, utilizing the Microsoft DirectML API. According to Intel, this resulted in accelerated project development.

Limitations and Future Improvements

As this release is in its preliminary stage, Microsoft has outlined certain limitations for PCs with Intel Core Ultra processors. Currently, NPU support in DirectML is compatible with only a subset of machine learning models. Some models may not function at all, exhibit significant latency, or demonstrate lower accuracy. However, Microsoft assures that compatibility and performance will improve over time for a broader range of models.

Community Engagement: Feedback and Development

In a bid for community involvement, developers are encouraged to provide feedback and report issues on the DirectML GitHub page. This collaborative approach aims to refine and optimize the compatibility of NPU support in DirectML, ensuring a seamless and efficient experience for developers leveraging Intel Core Ultra processors in AI application development.

- Advertisement -Newspaper WordPress Theme

Latest article

- Advertisement -Newspaper WordPress Theme

More article