MAX and Mojo: Extending PyTorch Across AI Hardware Ecosystems

youtu.be3 min read

At the PyTorch Conference, Modular announced the full open-sourcing of its Max accelerator stack alongside Mojo, aiming to provide a unified software layer for diverse AI hardware similar to what LLVM achieved for CPUs. The modular stack powers operations across chips from six different hardware vendors and is driving a new industry alliance to catalyze growth and collaboration among hardware makers and AI developers. Open article →

More on this topic