Scalable Synchronization Mechanisms for Manycore Operating Systems

This lecture by the instructor from Georgia Tech discusses the challenges of handling the increasing data growth and the regressions in operating systems. It covers the importance of synchronization mechanisms in many-core machines, the issues with current lock designs, and the need for scalable solutions. The lecture presents the instructor's contributions in designing scalable blocking locks, relaxed TLB coherence, and avoiding performance regressions. It also explores the impact of the instructor's research on evolving OS services for heterogeneous hardware and applications, emphasizing the critical role of lock algorithms in achieving high throughput and minimal memory footprint.