Please use this identifier to cite or link to this item:
https://www.um.edu.mt/library/oar/handle/123456789/35096
Title: | On the efficiency of extracting static information from memory dumps |
Authors: | Frendo, Dylan |
Keywords: | Computer software -- Testing Java (Computer program language) Memory management (Computer science) |
Issue Date: | 2018 |
Citation: | Frendo, D. (2018). On the efficiency of extracting static information from memory dumps (Bachelor's dissertation). |
Abstract: | Profiling is a form of dynamic analysis on a program, used to investigate its behavior. It is useful to measure different types of information regarding the application, like the heap, CPU utilization, and time consumption of method. The gathered data can then be used to help identify bottlenecks, extract system events for runtime verification or improve static analysis coverage. In this project, the heap and CPU profiler Hprof is examined. Hprof forms part of the HeapDL toolchain. This toolchain takes whole-heap snapshots, which are provided by Hprof, during program execution. One of the toolchain's drawbacks is the high overhead incurred when obtaining the heap snapshots by Hprof. Thus, the main sources of over- heads in Hprof are examined, and designs to overcome them are proposed. With the new implementation, calls to the JVM to get the stack traces are avoided by building a call trie locally in Hprof. The trie consists of all the method invocations that were used throughout the application's execution. This results in the profiler knowing the entire stack traces of all threads when an object is allocated. When tested on real-life applications, the new implementation manages to decrease the overall overheads of Hprof up to 50% on certain applications. The efficiency of the new implementation depends on the object allocation to method call/return ratio in the profiled application. When this ratio is greater than 1:10, the new implementation is more efficient. The implementation is inaccurate by 0.001% when the output is compared with the original Hprof. |
Description: | B.SC.(HONS)COMP.SCI. |
URI: | https://www.um.edu.mt/library/oar//handle/123456789/35096 |
Appears in Collections: | Dissertations - FacICT - 2018 Dissertations - FacICTCS - 2018 |
Files in This Item:
File | Description | Size | Format | |
---|---|---|---|---|
18BCS006.pdf Restricted Access | 1.25 MB | Adobe PDF | View/Open Request a copy |
Items in OAR@UM are protected by copyright, with all rights reserved, unless otherwise indicated.