Intel and AMD Combine AI Development on x86 Processors

Intel and AMD have joined forces to simplify AI development on PyTorch, making the oneDNN library the standard for x86 CPUs.

The giants of the semiconductor world, Intel and AMD, signed a historic collaboration to simplify artificial intelligence development processes in the x86 ecosystem. The companies, coming together under the umbrella of the x86 Ecosystem Advisory Group (EAG), announced that they have standardized the oneDNN library to ensure that artificial intelligence workloads developed on PyTorch run much more efficiently on x86 CPUs.

This strategic move aims to end the fragmented structure in the industry by eliminating the need for developers to create separate software paths for both processor architectures.

This collaboration, which puts aside the competition that has been going on for years in the technology world, offers great convenience, especially for software developers. Jeff McVeigh, Intel's Vice President of Data Center Strategic Programs, emphasized that developers should not have to follow different hardware-based software paths to get the best performance. According to McVeigh, this consistent approach delivered through oneDNN fully unlocks the potential of modern processors.

AMD Combines ZenDNN Optimizations Under oneDNN

On the AMD side, Robert Hormuth reiterated the company's vision of creating an open software ecosystem. AMD has incorporated its own ZenDNN optimizations into the Intel-led oneDNN library, creating a one-stop shop for PyTorch users. This move allows x86 platforms to deliver much more competitive performance in AI workloads.

Industry Stakeholders Are Satisfied with the New Standard

This development is of great importance not only for Intel and AMD, but also for ecosystem partners such as Red Hat. Stephen Watt, Red Hat CTO Office, stated that this standardization will make the developer experience seamless and facilitate consistent deployment of AI workloads across different x86 architectures. Standardization enables complex and time-consuming hardware-oriented configurations to be replaced by practical, applicable solutions.

A New Era Begins for Developers

The main result of the collaboration will be that developers will be able to focus on the software itself rather than complex hardware-specific solutions. With this step, Intel and AMD are removing barriers to innovation in the x86 world.

Organizations will now have much greater flexibility when scaling AI projects and will be free from traditional bottlenecks such as vendor lock-in. This common platform will also allow future AI computing plugins to be integrated into the PyTorch framework more quickly.