We present the Memory Trace Visualizer (MTV), a tool that provides interactive visualization and analysis of the sequence of memory operations performed by a program as it runs. As improvements in processor performance continue to outpace improvements in memory performance, tools to understand memory access patterns are increasingly important for optimizing data intensive programs such as those found in scientific computing. Using visual representations of abstract data structures, a simulated cache, and of the effects of memory operations, animating memory operations, MTV can expose memory performance bottlenecks and guide programmers toward memory system optimization opportunities. Visualization of detailed memory traces operations provides a powerful and intuitive way to expose patterns and discover bottlenecks, and is an important addition to existing statistical performance measurements.
Back to research page.