Abstract: A method for monitoring memory performance of a program. A frequently occurring event is detected and a stack associated with the program is identified in response to a detection of the periodically occurring event, which may be a page fault. The stack is examined to identify each routine (and specific invocation point, or offset) that is currently executing in association with the program. Each routine (and offset) is represented as a node in a tree structure. Classification of the page fault is performed. Page faults classified as data faults can be further sub-classified to provide additional information.