Moore's law, the observation that the number of circuit elements on top-end chips is increasing exponentially, has different consequences for different kinds of chips. On microprocessors, growth has been used to increase circuit and computational concurrency. DRAM memory parts, however, have gotten much bigger without much change in externally visible concurrency or speed. These diverging trends are leading to an "impedance mismatch" between processors and memory that is making McKee and Wulf's "memory wall" worse rather than better. In this talk we will address aspects of this problem in several ways. First, we will present the results of empirical measurements across a wide range of recent systems. Second, we will discuss the implications for applications, compilers, and system software. Third, we will present an overview of current project's at the Renaissance Computing Institute that are addressing aspects of the problem. Fourth, we will finish by speculating on future architectural trends.