Browsing the time distribution
When a benchmark is sampled [S] or traced [T], Browse Distribution in the Results menu is enabled. Selecting this item builds an execution tree from the selected benchmarks in the benchmark list and opens a Method Distribution browser.
Note:
The Browse Distribution menu items are grayed when no benchmarks are selected or when run [R] benchmarks are selected (run R] benchmarks contain no data from the execution stack).
Browsing the time distribution answers the following questions:
• What kind of work does the code do?
• Which operations are expensive?
Browsing the distribution of time by application, class, and method provides an overview of the code, and design decisions often become apparent. For example, the code may make extensive use of
Dictionaries when
Arrays or instance variables would work equally well. Using this information, you know that
Dictionary operations are a problem but does not know where in the code they are used. This question is answered in
Browsing the method execution.
In the Method Distribution browser, selecting applications, classes, and methods updates the bar graph to show a sorted view of the time spent in all occurrences of a method in the execution tree. When checked, the Show SubApplications button causes the applications list to show subapplications. Selecting applications, classes, and methods from the appropriate lists filters out other items and updates the bar graph. This allows you to focus on the expensive operations. A status line gives summary information about the execution tree.
The
Time menu filters raw, execution, scavenge, and global garbage collection time for the bar graph. For a description of this menu, see
Filtering raw time.
Last modified date: 01/29/2015