Virtualization and the Cost of Indirection | Richard M. Karp Distinguished Lecture

Some types of virtualization, such as virtual memory, are implemented by providing a layer of indirection between what the program sees and what the system implements. This layer of indirection is typically ignored in theoretical analysis but has a real (and, in some cases, increasing) impact on system performance. In this Richard M. Karp Distinguished Lecture in the Algorithmic Foundations for Emerging Computing Technologies program, Martín Farach-Colton covers a variety of cases where the cost of indirection becomes significant, including new architectures, such as for hardware accelerators and shared memory.