Related Topics:
Related Topics:| Choosing the Data to Collect |
Hardware counter overflow profiling records the call stack of each LWP at the time the hardware counter of the CPU on which the LWP is running overflows. The data recorded includes a timestamp and the IDs of the CPU, the thread, and the LWP. Typically, counters keep track of such things as instruction-cache misses, data-cache misses, cycles, floating-point operations, or instructions issued or executed.
Hardware counter overflow profiling can be done only on UltraSPARC III systems running the SolarisTM 8 Operating Environment (SPARCTM Platform Edition) and on Intel systems (Pentium II or III) running the Solaris 8 Operating Environment (Intel Platform Edition).
The Sampling Collector can record data for two hardware counters in an experiment. These counters must count data on different registers. To record data for more than two counters or for counters which count on the same register you must run more than one experiment.
The counters available depend on the specific CPU microprocessor and operating environment. You can display the list of available counters by clicking Show in the Sampling Collector Window or by typing the collect command with no arguments.
The Sampling Collector, the dbx collector command (see the collector(1) man page) or the collect command (see the collect(1) man page) allows you to select the types of counter whose overflow is being monitored, and to set overflow values for them.
By default, the Function List Display in the The Performance Analyzer shows the exclusive and inclusive hardware counter overflow profiling count, or the corresponding time if the count is in cycles.
Note - Hardware counter overflow profiling and clock-based profiling are mutually exclusive. If you enable one, the other is automatically disabled, with a warning.