The APS View displays information about the adaptive partitioning scheduling (APS) on a target.
For more information about adaptive partitioning, see:
To configure APS for a connected target, you must use the
Target Navigator.
The
APS View shows APS-related statistics and settings in multiple panels within its top pane;
the subsections that follow describe these panels. The bottom pane displays
graphs
of the partition budgets and CPU usage.
APS System Information
This panel displays the following APS state fields:
- current number of partitions
- partition, process, and thread IDs where the last bankruptcy (exhaustion of critical budget) occurred
- average number of computational cycles executed per millisecond
- averaging windows, which are time intervals used by the scheduler to calculate CPU usage
- APS security flags in effect, which determine who can add partitions and modify their attributes
- relative and overall times that the system was idle for longer than the second and third averaging windows
(by default, these are 1 and 10 seconds)
Note:
When you add or modify partitions using the
SchedCtl() C library function or the
aps
utility, the changes are reflected in this and all other panels when the view is refreshed.
Partitions
In this panel, the
Partition Information table shows these details for each partition:
- partition name
- parent partition
- partition ID
- budget, which is the percentage of CPU time guaranteed for the partition
- critical budget, which is the time that critical threads can run over budget
On the right, the Partition Child Processes and Threads area lists the processes and threads
assigned to each partition. The list is interactive, allowing you to move processes and threads between partitions by
dragging and dropping. Moving one of these items might cause others to be moved as well.
Partition Statistics
This panel displays CPU usage statistics related to the averaging windows. The table shows the relative CPU times during
which threads in each partition ran for longer than the second and third windows. It also shows the total CPU times that
all or just critical thread runtimes exceeded those two windows.
By default, the second and third averaging windows are 1 and 10 seconds, but you can change them by setting a new first
window, through SchedCtl() or aps. The second window is set to 10 times this duration
and the third window to 100 times.
APS Bankruptcy
In this panel, the Bankruptcy Information table lists the IDs of each partition's notification
process and thread (which handle the event emitted when the partition exhausts its critical budget and thus,
becomes bankrupt), and the IDs of the process and thread running when the last bankruptcy occurred.
On the right, the Bankruptcy Flags area lists the system's bankruptcy settings, which determine how
it reacts to a bankruptcy. You can change the bankruptcy settings through SchedCtl() or
aps.
Partition and CPU usage graphs
The
APS View contains the following graphs in multiple tabs:
- Partition Budgets
- This piechart shows the partition budgets, which are the percentages of CPU time assigned to the partitions.
- CPU Usage
- This line chart shows recent CPU usage for each partition.
- Critical Time Usage
- This line chart shows critical time usage for each partition, which refers to the time (in milliseconds) that
critical threads within a partition ran over budget.