Maxime Hugues

Performance optimization of distributed training jobs is complex due to the multiple levels of parallelism and the need to drill down in the accelerator, CPU, and network metrics at nanosecond granularity throughout the stack, from the communication libraries driving collective network operations to the ML framework.

AWS Batch can run various types of workloads, such as machine learning and training LLMs, processing logs for compliance, and traditional high performance computing (HPC) applications like computational fluid dynamics. Join this workshop to learn how to run HPC and other workloads using AWS Batch.

Join this workshop to gain a solid skills foundation to run common high performance computing (HPC) workloads using cloud technologies. Start with a primer on modern cloud foundations and how they map to common HPC concepts.

In this workshop, gain an understanding of the recently added AWS Batch support for managing and scheduling large-scale data analysis using Amazon EKS. Join this hands-on workshop and learn AWS Batch fundamental concepts and how AWS Batch works together with your other AWS services running within Amazon EKS.

Elastic Fabric Adapter (EFA), built on the Scalable Reliable Datagram (SRD) protocol, is the foundation for scaling message passing interface (MPI) and NVIDIA Collective Communications Library–based high performance computing (HPC) and machine learning (ML) codes at AWS. This builders’ session addresses application scaling challenges and is led by an AWS EFA team principal engineer.