About the talk: 

This webinar will cover high-performance communication software for GPU supercomputers. NCCL and NVSHMEM will be explained, including their historical context from MPI and SHMEM. Functionality and performance will be demonstrated through an example from linear algebra. Real-world results from both scientific and commercial AI use cases will also be presented.

About the speaker: 

Jeff Hammond is a Distinguished Engineer at NVIDIA in the data centre software organisation, focused on GPU communications. He has extensive experience with the design and use of parallel programming models and scientific applications. Jeff’s most notable achievements include the MPI-5 Application Binary Interface standard, development of the MPI-3 one-sided communication software ecosystem, and contributions to the NWChem quantum chemistry project. He received a PhD in Chemistry from the University of Chicago in 2009.

About the moderator:

Cláudia Brito is currently an Assistant Researcher at INESC TEC (and part of the DSR HASLab group) and an Invited Assistant Professor at the University of Minho. Her current research focus is on AI for Systems and Systems for AI. In detail, her research interests have been shifting towards the following topics: AI for systems optimisations and tuning, Storage optimisations for AI workloads, Energy consumption optimisation for data-intensive workloads, Application-specific AI for healthcare.