Designing Networking Stacks for Machine Learning Frameworks

Learn about designing networking software and hardware for machine learning frameworks in this 41-minute technical talk from Enfabrica's Raghu Raja at the OpenFabrics Alliance. Gain foundational knowledge of Neural Networks and the ML framework ecosystem from a network architect's perspective, explore the challenges faced when developing networking solutions for rapidly evolving ML models and applications, and examine a detailed case study involving NCCL implementation. Discover practical insights through a prototype demonstration and understand key action items for the OpenFabrics Alliance community to address current networking challenges in machine learning environments.