HP Caliper Advisor Rule Writer Guide

Table 3-1 Analysis Object Data Members (continued)
DescriptionTypeData
Contains the operating system type of the reference dataset. One of the
following: HPUX_OS or LINUX_OS.
integeros_type
Contains the operating system version of the reference dataset. One of the
following: HPUX_11_23_OS or LINUX_2_6_OS.
integeros_version
Contains the full absolute path of physical objects or an empty string (“”) for
execution objects.
stringpath
Contains the PMU type of the reference dataset. One of the following:
ITANIUM2_PMU or MONTECITO_PMU.
integerpmu_type
Contains the type of the object. One of the following: SYSTEM_OBJECT,
APPLICATION_OBJECT, PROCESS_OBJECT, THREAD_OBJECT,
KERNEL_OBJECT, EXECUTABLE_OBJECT, IMAGE_OBJECT, or
FUNCTION_OBJECT.
Note that the current version only supports EXECUTABLE_OBJECT.
integertype
Accessor Functions
Accessor functions are the means by which rules retrieve performance data for the
object being analyzed. Each accessor function retrieves different types of performance
data, and accessor functions are implemented internally in vastly different ways.
However, they are more alike than different in how they are used.
Every accessor function has as an argument a metric list, a set of metrics to be retrieved.
Three types of accessor functions are available in this release of HP Caliper:
• Global metrics accessor function
A global metrics accessor function is used to retrieve global metrics such as total
PMU event counts and statistics:
get_metrics(metric_list)
get_metrics_nth(metric_list, n)
• Histogram accessor function
A histogram accessor function is used to retrieve histogram (flat profile)
performance metric:
get_histogram(histogram_type, metric_list)
get_histogram_nth(histogram_type, metric_list, n)
• Run information accessor function
A run information accessor function is used to retrieve miscellaneous information
about a performance measurement run:
get_run_info(metric_list)
32 How the HP Caliper Advisor Works