Edmond Chow - Georgia Institute of Technology The standard parallel iterative methods for solving linear and nonlinear systems of equations are all synchronous, requiring a global synchronization between all processes at each iteration either explicitly or implicitly. Extreme scale, massively parallel computations may become bottlenecked at these synchronization points, especially in the presence of load imbalance or heterogeneous hardware. This talk will discuss parallel iterative methods that operate asynchronously, meaning that the processes never synchronize with each other. Topics include the convergence of asynchronous iterative methods and asynchronous versions of optimized Schwarz and multigrid methods.