Legion: Programming Heterogeneous, Distributed Parallel Machines

Explore the Legion programming model and runtime system in this 44-minute lecture by Alex Aiken from Stanford University. Delve into the challenges of parallel programming, focusing on the crucial aspects of data partitioning, placement, and movement in complex, hierarchical machines. Discover how Legion addresses these issues by providing a framework for describing abstract hierarchical organizations of both data and computation. Learn about the design principles and rationale behind Legion, and gain insights into recent developments in high-productivity libraries that leverage Legion's capabilities. Understand how this approach can help programmers tackle the increasing complexity of parallel programming in heterogeneous, distributed environments.