Features and Improvements
Intel® Extension for TensorFlow* extends the official TensorFlow capabilities, allowing TensorFlow workloads to run on Intel® Data Center GPU Max Series. This release includes the following features and improvements:
-
Toolkit Support: Supports Intel® oneAPI Base Toolkit (version 2025.0.1) and Intel® Deep Learning Essentials (version 2025.0.1)
-
Updated Support: Intel® Extension for TensorFlow* has been upgraded to support oneDNN v3.7
Bug Fixes
- Fixes onednn build issue and relocates the xbyak header library paths
- C++ linkage specification have been added to sycl headers, removing C compilation action for sycl includes
- Security fix for a known CVE in c-ares; made available with an update to the OpenXLA patch
What's Changed
- OpenXLA patch is updated with a security fix to c-ares library
- Links to Ahead-of-Time (AOT) Compilation documentation have been updated