Notice: This Wiki is now read only and edits are no longer possible. Please see: https://gitlab.eclipse.org/eclipsefdn/helpdesk/-/wikis/Wiki-shutdown-plan for the plan.
Difference between revisions of "Profiling with TPTP - plug-in development"
(→Known issues) |
(→Known issues) |
||
Line 49: | Line 49: | ||
------------- | ------------- | ||
− | ←[[TPTP|TPTP wiki page]] | + | ←[[TPTP|Back to TPTP wiki page]] |
Revision as of 15:56, 10 October 2008
Contents
Overview
- This page shows how to quick start a profiling session on a plug-in project. TPTP must be installed on an Eclipse workbench, for more information on installation, please read Install TPTP with Update Manager.
Profile a plug-in project
- 1. Select a plug-in project and select Profile As > Eclipse Application.
- 2. In the profile configuration dialog, under Monitor tab, select Java Profiling agent and select Edit Options.
- 3. In Filter page, add a new filter.
- 4. Input a new filter that includes your plug-in packages. eg.,
- 5. Click Finish to apply filter.
- 6. Click OK to start profile session.
- 7. Select Yes to switch to profiling perspective upon request.
- 8. Profiling resources (process, agent, and analysis type) are created in Profiling Monitor view.
- 9. Interact with your Run-time workbench for profiling agent to collect profiling data.
- 10.In the development workbench, double-click on an analysis type to open a profiling view to show profiling data collected.
- 11.In profiling table view, select any column header to sort the table.
Execution Analysis : Identify a bottleneck
TPTP provides tools and views to help identify execution bottleneck problem, please take a look at Execution Analysis for more information.
Memory Analysis : Identify a memory leak
TPTP provides tools and views to help identify potential memory problem in an application, please take a look at Memory Analysis for more information.
Thread Analysis : Identify threading problem
TPTP provides tools and views to analysis threading problem in an application, please take a look at Thread Analysis for more information.
Performance Overhead
The Performance overhead of a profile session mainly depends on the profile configuration which includes configuration of profile filter and profile options. Please see Profiling configuration for detail about how to control the amount of profiling data hence the overhead in a session.
Hints and Tips
See TPTP FAQ page for detail on hints and tips for a profiling session.
Known issues
Refer to TPTP FAQ page for known issues and workarounds.