微软近日宣布，将于明年初为旗下的机器学习框架DirectML添加NPU支持。此举旨在充分利用搭载NPU的芯片，如英特尔酷睿Ultra等，以提升AI模型在DirectML上的运行效能。
在添加NPU支持后，开发者可通过跨平台推理引擎“ONNX Runtime”等API，利用芯片内置的NPU运行AI模型。这将极大地提高模型在实际应用中的性能表现，进一步降低AI技术在实际应用中的门槛。
据悉，DirectML是微软DirectX家族的一员，致力于为游戏和多媒体应用提供高性能的图形和计算能力。此次引入NPU支持，将使DirectML在处理AI任务时具备更高的灵活性和效率。
英文翻译：
News title: Microsoft to add NPU support to DirectX
Keywords: DirectML, NPU, AI models
News content:
Microsoft recently announced that it will add NPU support to its machine learning framework DirectML in early 2023. This move aims to fully leverage chips equipped with NPUs, such as Intel Core Ultra, to enhance the performance of AI models on DirectML.
After adding NPU support, developers can run AI models through cross-platform inference engine “ONNX Runtime” and other APIs by utilizing the built-in NPU on the chip. This will greatly improve the performance of the models in practical applications, further reducing the threshold for AI technology in real-world use.
It is reported that DirectML is a member of the Microsoft DirectX family, dedicated to providing high-performance graphics and computing capabilities for games and multimedia applications. By introducing NPU support, DirectML will enhance its flexibility and efficiency in handling AI tasks.
【来源】https://www.ithome.com/0/740/014.htm
Views: 1