India, Oct. 26 -- A new collaboration between two longstanding processor rivals aims to reduce software fragmentation in PyTorch, potentially making enterprise AI workloads easier to deploy across x86 infrastructure.

Running AI workloads across different x86 processors can introduce unnecessary complexity for developers. Although modern machine-learning frameworks provide common programming interfaces, achieving efficient performance often requires hardware-specific optimisation libraries. A new collaboration between AMD and Intel aims to address this challenge by bringing their PyTorch CPU optimisations together within oneDNN, an open-source deep learning performance library.

Announced on October 20, 2026, the initiative follows a reco...