Microsoft has announced that they have open-sourced their TensorFlow-DirectML extension of TensorFlow on Windows that is now on GitHub for general availability.
This extension broadens the reach of TensorFlow beyond its traditional Graphics Processing Unit (GPU) support. It enables high-performance training and inferencing of machine learning models on any Windows devices with a DirectX 12-capable GPU through DirectML.
TensorFlow is a widely used machine learning framework for developing, training, and distributing machine learning models. It can leverage both CPUs and GPUs. DirectML is a hardware-accelerated deep learning API on Windows. It is a low-level library built on top of Direct3D 12; the API is designed for high-performance, low-latency applications that require absolute control over resource allocation and work scheduling. Dedicated hardware such as the GPU is often used to accelerate machine learning workloads seeing they often involve tremendous amounts of computation far more in training models.
The first preview of TensorFlow-DirectML was released in June 2020. Open-sourcing this extension brings development and collaboration closer to the TensorFlow community.
You can check out the full announcement post on the Microsoft cloud blog found here. You can also learn more about TensorFlow from the official website here. To learn more about DirectML, check out the official repository on GitHub here. Otherwise, you can get the full TensorFlow-DirectML source code from the GitHub repository here.