Analyzed Graph Primitives

The following structures provide execution statistics about STM entities such as Runnables, Hyperepochs, Epochs, Resources, and Clients.

RunInformationFull

RunInformationFull is a container for information obtained during execution of an STM entity (e.g., a runnable, a hyperepoch). It contains all start and end times of the entity in addition to its aggregate run statistics.

It has the following fields:

Field	Type	Description
start_times	numpy.ndarray	A numpy ndarray containing the times at which the associated STM entity began execution. One array entry exists for each frame.
end_times	numpy.ndarray	A numpy ndarray containing the times at which the associated STM entity ended execution. One array entry exists for each frame.
duration	numpy.ndarray	A numpy ndarray containing the per-frame durations of the associated STM entity. This attribute is equivalent to “end_times - start_times”. One array entry exists for each frame. A value of 0 (zero) indicates that the given frame was skipped.
overruns	numpy.ndarray	A numpy ndarray containing the indices into start_times, end_times, and duration in which a frame overrun occurred.
skipped_frames	numpy.ndarray	A numpy ndarray containing the indices into start_times, end_times, and duration in which a frame skip occurred.
missing_frames	numpy.ndarray	A numpy ndarray containing the indices into start_times, end_times, and duration in which one or more markers were missing. In this case, we consider the entire frame skipped.
duration_stats	StatisticsWithOverruns	Aggregate statistics describing run durations.
overrun_stats	Statistics	Duration distribution of the runnables which have overrun. This is different from the overrun info within the duration_stats variable above. This shows the duration distribution for only the frames which have overrun, while the prior includes statistics for both overrunning and non-overrunning frames.
start_time_jitter_stats	Statistics	A Statistics object describing the jitter in the start times of an STM entity.
can_intentionally_skip	bool	A boolean indicating whether this run info can have some runs intentionally skipped; because of round robin execution, etc.

Statistics

A container for the statistics describing the data obtained during a runtime instance of an STM entity (for example a runnable or a hyperepoch).

It has the following fields:

Field	Type	Description
num_frames_executed	int	The number of frames that were executed during the run (frame skips not counted).
mean_duration	float	The mean of the frame durations.
std_duration	float	The standard deviation of the frame durations.
percentile_99_duration	float	The ninety-ninth percentile of frame durations.
max_duration	int	The maximum frame duration.
min_duration	int	The minimum frame duration.

StatisticsWithOverruns

A container for the statistics describing the data obtained during a runtime instance of an STM entity (for example a runnable or a hyperepoch). Expands upon Statistics by including overrun-related statistics.

It has the following fields, in addition to the fields described under Statistics:

Field	Type	Description
num_overruns	int	The number of overruns which occurred.
overrun_fraction	float	The fraction of frames which overrun as a float between 0 and 1.
mean_overprovision	float	The mean in nanoseconds of the discrepancy between scheduled duration and actual duration (scheduled time / how long the runnable actually took). Note that this value can be negative as overruns occur.
mean_overprovision_frac	float	The mean as a fraction of the discrepancy between scheduled duration and actual duration (mean_overprovision / scheduled duration). Note that this value can be negative as overruns occur.

CriticalPathInfo

CriticalPathInfo is a container for the critical path information obtained from the STM Schedule and runtime log for a particular hyperepoch.

It has the following fields:

Field	Type	Description
path_metrics	list(CriticalRunnableMetrics)	A list of metrics per runnable in the critical path(s).
scheduled_path_length	int	Length of the scheduled critical path in ns.
mean_exec_path_length	float	Mean of observed critical path length in ns (includes overhead).
overhead_fraction	float	Fraction of the critical path contributed by overhead (0 <= overhead_fraction <= 1 ).
resource_utilization	dict(str→float)	A Dictionary containing the resource type and the fraction of critical path operating on that resource type.

CriticalRunnableMetrics

A container for the following runnable metrics:

Field	Type	Description
runnable_id	str	A string indicating the runnable ID.
start_time	int	The scheduled start time of the runnable in ns.
wcet	int	The WCET specified in the YAML input to the STM compiler.
mean_exec_time	int	The mean of measured execution times in ns for this runnable.
mean_exec_time_with_overhead	int	The mean of measured execution times for this runnable, including STM overhead for scheduling and logging in ns.
mean_overhead	int	The mean of STM overhead seen for this runnable, in ns.
resource_type	str	The type of hardware resource used by this runnable.