Greetings everyone. Here is our first attempt at collecting performance data for gedit, for a single test scenario. We have used quantify and the forte tools for obtaining the results. Please go through the attached html report and give us your valuable comments, so that we can improve towards providing more efficient and useful performance metrics. We have put forth our first round of findings in the report. So, please let us know if we are sailing in the right direction. The report details only a small portion of the voluminous data. Zipped archive links to the complete performance data are provided in the report, in case you require additional metrics. Based on your inputs, we will send across reports for more test scenarios and components. Regards, Muktha and AnandTitle:
|
(version 1.0) 1. Component Summary
2. Brief Description of
Tests Done to Collect Performance Data
3. Data Test : Gedit_01 As per the quantify data collected, the top five functions which consumed the maximum percentage of the total execution time, have been considered for the analysis. The forte leaks data showed a total of 12 MB leak when a 4 MB file was loaded.
3.1.1 Forte Collect Data
3.1.2 Quantify Data
[1] Due to huge sizes of the er_print data for leaks, the data in the attached gunzipped files have been limited to the top 500 leaks only. The complete leaks information is however available and can be provided, if required. 4. Appendix 4.1 Forte data Data is collected using the 'collect' and 'er_print' forte profiling tools. Data can be collected using the 'collect'command. All data can be converted into performance metrics computed against functions, callers and callees of any functions, and against source and disassembly of the target program. The 'er_print' command is used to get an ASCII output of the performance metrics obtained from the 'collect' command. Three types of function-level metrics can be computed using the performance analysis tools: Exclusive metrics: are calculated from events which occur inside the function itself: they exclude metrics coming from calls to other functions. Inclusive metrics: are collected from events which occur inside the function and any functions it calls: they include metrics coming from calls to other functions Attributed
metrics:are the parts of the inclusive metric of the selected
function that are due to calls from a caller or calls to a callee.
Exclusive user CPU time is the amount of time spent in the function itself, in seconds Inclusive
user CPU time is the amount of time spent in the function itself and any
functions it calls, in seconds
Callers-callees data: Prints the callers-callees panel for each of the functions, in the order in which they are sorted. The selected (middle) function is marked with an asterisk. The callers of the selected function are shown above. The functions which the selected function calls are listed below. Leaks data: The ?Leaks? report shows what
memory has been leaked while the program was running.
4.2 Quantify data Functions list: Displays the list of function calls sorted by percentage of '.root' Exports file: Is the ASCII version of the quantify data. Elaborates on all the callers of a particular function, the number of times the function was called, propagated time, the function descendant time and many more metrics. Annotated source: Presents a line by line performance data using the functions source code. | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||