Show TOC

Performing Runtime AnalysisLocate this document in the navigation structure

Use

You can evaluate and display all the profiling data related to the runtime of the application. You can find out which methods took a long time or were called frequently, and for these reasons have affected performance.

Prerequisites

You have closed a profiling session in which you had activated the recording of runtime data. You are now in the evaluation overview for this session.

More information: Evaluating Profiling Data

Procedure

  1. In the navigation structure choose the node Start of the navigation path Runtime Analysis Next navigation step Methods (Sampling) End of the navigation path.

    On the right-hand side a table is displayed containing the methods that were run for time-based sampling during the profiling session within a VM.

    For each method the following information is shown:

    • Full class name (including the name of the package)

    • Method name including the parameter signature

    • Net time, that is, the percentage of times the method was found on top of the Java stack.

    • Gross time, that is, the percentage of times the method was found anywhere in the stack.

    • Number of callers, that is, the number of different method locations from which the method was called.

    • Number of callees, that is, the number of different method locations from which the method itself did the calling. This number is 0 if the method was found exclusively on the top on the stack. In this case, the net time is inevitably the same as the gross time.

    By default the table is sorted in descending order by the net time, which means that you can quickly identify the methods that took a particularly long time.

  2. (Optional) In the header line select the number of the VM you want to see the results for.

  3. Select the table line of the method you want by double-clicking it.

    The following information about the method appears:

    • Selected Method: In the upper section of the screen, in addition to the table values of the method, you will find the relative time, that is, the percentage of occurrences at the top of the stack compared to occurrences further down in the stack. The relative time measures how much time was spent within the method itself. With non-recursive method calls: relative time = net time * 100 : gross time.

    • Caller: In the middle section of the screen a list of callers is displayed with the class name, method name, and the number of calls made to the method as a percentage.

    • Callees: In the lower section of the screen a list of the called methods is displayed, with the class name, method name, and the number of all calls made from the method as a percentage.

  4. (Optional) If you find a method here you want to see the details for (caller, callee, etc.), select the method by double-clicking on it. Details will appear and then you can select other methods if you want. In this way you can trace method calls that are called particularly frequently or run for a particularly long time.

More Information

More Information