Parallel Stochastic Asynchronous Coordinate Descent: Tight Bounds on the Possible Parallelism

Abstract: Several works have shown linear speedup is achieved by an asynchronous parallel implementation of stochastic coordinate descent so long as there is not too much parallelism. More specifically, it is known that if all updates are of similar duration, then linear speedup is possible with up to $\Theta(\sqrt n L_{\max}/L_{\overline{\mathrm{res}}})$ processors, where $L_{\max}$ and $L_{\overline{\mathrm{res}}}$ are suitable Lipschitz parameters. This paper shows the bound is tight for almost all possible values of these parameters.