Microsoft announced that next year it will add NPU support to DirectX machine learning framework DirectML, which adapts to AI chips such as Intel Core Ultra.

CTOnews.com, December 18, Microsoft recently announced that it will add NPU support to the DirectML machine learning framework early next year to adapt to Intel Core Ultra and other built-in NPU chips.

▲ source Microsoft official press release (the same below) Microsoft launched the DirectML framework belonging to the DirectX family in 2021, which focuses on "machine learning", can directly access GPU for in-depth learning, and can provide help for a series of AI scenarios such as "picture noise reduction", "game pre-rendering", "ray tracing" and so on.

Microsoft said that after DirectML supports NPU, developers can improve the performance of the model by running the AI model through API such as cross-platform reasoning engine "ONNX Runtime" and using the NPU built into the chip.

CTOnews.com noted that Microsoft also said that DirectML will "directly support" the core Ultra chip early next year, and the NPU built into the related chip will become "the first NPU on the Windows platform to support DirectML."

▲ Intel previously released the Core Ultra series processors. Besides Intel Core Ultra, the current new generation of chips from Qualcomm, AMD and other manufacturers also have built-in NPU, allowing AI computing with relatively low power consumption, thus releasing CPU and GPU computing resources.

Related reading: "Microsoft announces the public release of DirectML machine learning API: can use A card to accelerate the realization of light pursuit, DLSS and other effects"