| 摘要 |
An interactive, information logging and processing tool that provides information regarding a process's data structure utilization to reduce the working set of a process's dynamically allocated memory. The actual working set (AWS) determinant determines which portions of the dynamically-allocated pages, referred to as the process's virtual memory (VM) working set (VWS) are actually used. It then determines the actual working set of a dynamically allocated memory for a given benchmark. The basic approach of the AWS determinator is to observe which data structures cause page faults when the targeted process is severely thrashing. AWS determinator includes a data logger and a data analyzer. The data logger ensures that a consistent benchmark is achieved in order to obtain the most accurate results, increases the number and granularity of heap page faults for the target process to enable the page fault mechanism of the processor to count the number of times the associated data structure is accessed, and log all heap page faults and transactions. The data analyzer is an interactive, information processing tool that efficiently process the large amounts of data logged by the data logger during a benchmark run. The data analyzer also allows users to interactively explore the processed data, to gain insights into a process's heap AWS. The data analyzer correlates each block of heap memory to a particular C data structure. Then, after the benchmark is completed and the above information is logged and correlated, an information processing step is performed wherein an approximate determination of the targeted process's heap AWS is performed.
|
