Windows now natively supports PyTorch through DirectML

Similar to DirectX which offers low-level APIs for Graphics, DirectML offers low-level API for machine learning in Windows. Instead of wasting time and effort to target different CPU, GPU and NPU hardware options, DirectML allows developers to build once and make it available for all supported Windows PCs. It also integrated with frameworks such as the ONNX Runtime, PyTorch and WebNN.

At Build 2024, Microsoft today announced that Windows now natively supports PyTorch through DirectML. With this native PyTorch support, thousands of Hugging Face models can work on Windows without any changes. Also, Microsoft is working with Nvidia to make these development workflows supported in over 100M RTX AI GPUs. PyTorch support on GPUs is available from today and NPU support will be coming soon.

Microsoft is also enabling the power of local AI for web applications through WebNN Developer Preview. WebNN is powered by DirectML and ONNX Runtime Web. WebNN allows developers to enable AI experiences at near native performance for tasks such as generative AI, image processing, natural language processing, computer vision and more. For now, WebNN Developer Preview supports
GPUs. NPU support is coming soon.

DirectML Availability:

• DirectML is now generally available across all Windows GPUs.
• DirectML support on Intel Core Ultra processors with Intel AI Boost is available as a Developer Preview with general availability coming soon
• DirectML support on Qualcomm Hexagon NPU in the Snapdragon X Elite SoC is coming soon.