The Call Tree view displays the performance data as a hierarchical tree.

 

Read on, or click a part of the image below to jump straight to the help for that area.

 

 

 

As with many of the tab views, the display is split into two resizable panes.

 

•the left side shows an ordered hierarchical view of the functions called in the program's execution

 

The tree's popup menu provides options to filter data or to examine it in more detail.
 

•The right side shows the source code view for any function selected on the left

 

 

The call tree view

 

Performance data is shown in an ordered tree for every function that has been hooked and called.

 

Functions are initially ordered by Total Time - the time a function and its child functions contribute to the total run time.

 

Child functions that get called will appear as child nodes in the tree; just expand a node to dig deeper.

 

Selecting a function shows its source code in the right hand pane.

 

Double clicking any item will display the function on the Relations tab.

 

As this is a call tree, functions can appear in multiple locations if called by different parent functions.

 

Unlike the Call Graph, function timings shown here are for the time spent in the relevant part of the call tree, not a combined total for everywhere that a function may be used.

 

 

Window orientation

 

To adapt to your screen layout, the horizontal or vertical orientation of the call tree and source code panes can be toggled with the orientation button.

 

               

 

 

Updating the display

 

•Refresh updates the display with the called functions and performance data - as does the button on the Tools menu and toolbar

 

A refresh automatically expands the first nodes in the tree and highlights the most time consuming low-level function.

 

As there's no auto update here, you'll need to use this Refresh button to update the display when you wish.

 

•Collapse / Expand All hide or show every node in the tree

 

•Display... shows the Call Tree Display Settings dialog

 

 

 

 

Call tree colours

 

The tree is coloured using the customisable Hotspot Colours settings that range from 100% down to 0%.

 

 

•Colours choose the way the colour scheme is applied in the tree

 

•Parent % colouring is based on function time relative to that of its parent's function time
 

In the example below:

 

qsCompFunc is in the 90-100% colour band as it contributed 97.44% of its parent's function time

 
doCombSort is in the 90-100% colour band as it contributed 98.44% of its parent's function time

 

 

•Total % colouring is based on function time relative to the total execution time

 

Now in the example:

 

qsCompFunc is in the 50-60% colour band as it contributed 34.51% to the total run time

 
doCombSort is in the 30-40% colour band as it contributed 57.47% to the total run time

 

 

•No Colour black and white display only

 

 

Sorting the data

 

The tree is initially ordered by most time consuming function first (Total Time)

 

•Sort choose one of the criteria from the list Refresh the display

 

   The options are:

 

•Total Time the time all calls to a function and it's child functions contribute to the total run time
 

•Average Time the time an average call to a function and it's child functions contribute to the total run time
 

•Call Count the number of times a function appears in the sampled collection data

 

  When in sampling mode, Call Count will be the only available option here

 

•Ascending / Descending change the ordering direction of the data

 

Note that while your application is executing, the sorted data is live. Sorting may not complete correctly as the data may change during the sort. Only when your program has finished executing is the sorted data guaranteed accurate.

 

 

Managing the data being displayed

 

You can't filter out functions by name but there are two ways to change what's displayed in the view:

 

Change the amount of detail shown for each function:

 

•Address check to show the address in memory for the function

 

•Filename check to show the file location in which the function was found

 

•Modulename shows the function's module

 

 

Reduce the scope of the tree by hiding less significant functions:

 

•Threshold set a percentage threshold below which contributing functions are hidden

 

Higher numbers exclude more nodes.

 

You'll need to do a manual Refresh to update the display.

 
By default the threshold applies to the top level functions only.

 

The threshold percentage is based on each callstack's contribution to the total runtime or total number of function calls.

 

The default All option includes all nodes.

 

•Apply Threshold to Children check to hide any function (as opposed to just top level nodes) in the tree that lies below the threshold contribution

 

 

Finding text

 

To find text in the tree, use the Find Dialog where you can search for functions, files and modules.

 

The source code view has its own Find and Goto dialogs.

 

 

Call tree menu options

 

The following popup menu is available over the tree to add filters, examine relations or edit code.

 

Menu actions apply to the function in the tree at the menu-click location.

 

 

 

Menu options: Previous and next call tree nodes

 

Functions may appear in the tree more than once if called by different parent functions.

 

•Go to Previous Call Tree Node for this Function jump to the previous instance of this function in the tree

 

+has the same action

 

•Go to Next Call Tree Node for this Function jump to the next instance of this function in the tree

 

+has the same action

 

If there are no previous or next instances, nothing will happen.

 

 

Menu options: Call graph

 

The Call Graph view is similar to the Call Tree, but where a function would appear multiple times, it instead appears once as the main instance, and is then linked to from other locations.

 

The times shown in a call graph are accumulated total times rather than the time spent strictly under their parent functions.

 

•Go to Call Graph Node open the Call Graph view at the main entry for the selected function

 

 

 

Menu options: Function filter (Instrumentation)

 

While the display filter controls visibility of hooked data, instrumentation filters control which functions are hooked in the first place.

 

The function instrumentation filter sub-menu lets you add hook filters at different levels of granularity.

 

 

The affect of adding function filters here depends on the current filter settings:

 

•If the current filters are set to hook everything, adding new filters will switch to excluding newly selected hooks
 

•Otherwise, the current filter will be retained, i.e. hook or don't hook newly selected items  

 

 

The first three options add filters to the Class and Function Filter Settings:

 

•By Class adds a new filter, excluding the entire class from the results of subsequent sessions
 

•By Class and Method excludes only the selected function from new sessions
 

•By Function excludes all matching function names irrespective of their containing class or even if not in a class at all
 

 

The next two, Filename and Directory, are part of the Source Files Filter settings.

 

•By FileName adds a new filter, excluding all functions in the same file (as the selected item) from the results of subsequent sessions
 

•By Directory excludes functions in all files in the same directory as the selected function
 

 

Finally, the DLL level is controlled by the Hooked DLLs settings.

 

•By DLL excludes functions in all files belonging to the same executable or DLL as the selected function
 

 

Instrumentation filters become effective at the start of the next session. Adding a filter during a session will show the relevant rows in grey so that you can see which files would be filtered, but the performance data will continue to be included for the rest of the current session.

 

 

Menu options: Line timing filter (Instrumentation)

 

Line timing instrumentation filters control which lines are hooked for line timing and are independent of the function filters above.

 

The affect of adding line timing filters here depends on the current line timing filter settings:

 

•If the current filters are set to hook everything, adding new filters will switch to only including newly selected hooks

 

Note that this is the opposite of function filters.
 

•Otherwise, the current filter will be retained, i.e. hook or don't hook newly selected items  

 

The line timing instrumentation filter sub-menu lets you add hook filters at three different levels of granularity.

 

Each option add filters to the Line Timing Filter Settings:

 

•By Class adds a new filter, appending the entire class in the line timing results of subsequent sessions
 

•By Class and Method include only the selected function in the line timing of new sessions
 

•By Function includes all matching function names irrespective of their containing class or even if not in a class at all
 

Instrumentation filters become effective at the start of the next session. Adding a filter during a session will show the relevant rows in grey so that you can see which files would be filtered, but the performance data will continue to be included for the rest of the current session.

 

 

Menu option: Show in Analysis tab

 

The Analysis tab shows results of a query on the call tree based on the selected function and the criteria chosen from the menu:

 

•Show in Analysis tab choosing any item in the following sub-menu, to be switched to the Analysis tab

 

 

For example:

 

•Class the Analysis tab will show all points in the call tree that match the class of the selected function
 

•File name shows all points in the call tree that match the file name of the selected function

 

•Slower Functions (%) shows all functions that are slower than the selected function

 

 

Menu options: Show in Relations tabs

 

•Show in Relations tab switch to the Relations tab where you can examine functions that either call, or are called by, the selected function

 

Double clicking any item will also display it in the Relations tab.

 

 

Menu options: Remove entry

 

You can hide some entries from the Call Tree display.

 

Hiding entries is a very temporary action since a function and its children will only be hidden from view until the next Refresh.

 

•Remove entry hides the selected function in the tree

 

Only the selected node is hidden. Other calls to the same function in a different part of the call tree will remain.

 

 

Menu option: editing source code

 

•Edit Source Code... opens the default or preferred editor to edit the source code

 

 

Menu options: Expand and collapse

 

The last few menu options expand and collapse parts of the tree.

 

•Expand Next Hotspot finds the next most significant performance hotspot in the call tree and expands all the callstack nodes to make it visible

 

•Collapse / Expand Entry close or recursively open the selected function to its full extent

 

•Collapse / Expand All completely collapse or expand the entire tree

 

 

The file source code view

 

Clicking on a function in the call tree shows that function's source code file in the right hand pane.

 

The source code view is described in detail separately as its behaviour is the same for all views.