Μελέτη αλγορίθμων αντικατάστασης σελίδας με πραγματικά δεδομένα
Study of page replacement algorithms with real data
View/ Open
Abstract
The paper presented is related to page replacement algorithms and, more specifically, the analysis of applications’ behaviour as to the page faults they generate. A ready tool was initially used, which generates program traces in memory, for the collection and analysis of real data.
The tool mentioned is the Intel PIN. By using this tool, we can execute several benchmarks and record the memory positions they use. As a result, it is possible to find the reference sequence in the form of pages. The following step was to implement another tool, using the C programming language, that finds which page faults take place, as well as their total number, according to the reference sequence, for different page replacement algorithms as well as page frames. Finally, the behaviour of each algorithm was analysed, according to its total number of page faults, after the execution of each benchmark separately.