Polyhedral Analyses using MLIR's Affine dialect - Vinayaka Bandishti

Abstract: Polyhedral analyses provide a succint way to reason about optimizations on affine nests. Since many of the ML/DL, image processing, scientific computing workloads are largely comprised of affine nests, understanding these analyses is highly important. In this tutorial-style beginner-friendly session, some basic concepts of polyhedral analyses using MLIR's Affine dialect will be introduced. Quick demo-like introductions to MLIR's affine dialect, some of its ops and structures will be covered as well. About the Speaker: Vinayaka Bandishti is a Staff Software Engineer at Polymage Labs working on advanced compiler infrastructure to accelerate AI/ML computations. Before joining Polymage Labs, He worked as a Senior Software Engineer at Veveo. He completed his Bachelor of Engineering in Computer Science from Sri Jayachamarajendra College of Engineering and Master of Science from IISc Bangalore LinkedIn:   / vinayaka-bandishti-0825265a   Phabricator: https://reviews.llvm.org/p/vinayaka-p... Slides: https://drive.google.com/file/d/1LKLn...