Intel and AMD announced on October 20, 2026 that they would merge their PyTorch optimization work for x86 processors into a shared open-source library called oneDNN. AMD will integrate ZenDNN into the project and join the UXL Foundation, which governs oneDNN, as a steering member. The initiative aims to provide a single development path and accelerate access to capabilities such as AI Compute Extensions, or ACE. The two companies are focusing primarily on inference, edge computing and latency-sensitive tasks that complement GPU accelerators as Arm-based architectures gain ground. The technical integration of ZenDNN and the joint governance of oneDNN will determine whether the partnership delivers concrete benefits for developers.
Source: Read the original article
