HP Caliper User Guide Release 5.5 (5900-2351, August 2012)

The report shows two levels of information:
• Exact counts of CPU metrics summed across the entire run of an application
• Sampled IPs that are associated with particular locations in the measured application.
When compared with the fprof measurement, the cycles measurement provides the following
two additional pieces of information when invoked with the -r all option:
• Cycles Per Bundle: The average number of cycles elapsed to retire the bundle. If there are no
stalls, it should take exactly one cycle to retire a bundle. If the Cycle Per Bundle information
is more than 1, this means that many additional cycles of stall were seen on that bundle.
• Split issues: In general, when an instruction does not issue at the same time as the instruction
immediately before it, the instruction execution is said to have a split issue. The split issue can
occur when there are explicit stop bits in a bundle or there are insufficient machine resources
of the type required to execute the instruction. In the cycles report, the Cycles Per Bundle
information is prefixed with an asterisk (*) if instructions from that bundle are split issued.
The report shows measured data by thread, load module, function, source statement, and instruction
bundle.
Command-line options allow you to control the amount of data reported, how the data is sorted,
and the number of statements and instructions reported for each sampled program location.
Example Command Line for Text Report
$ caliper cycles -ra -o reports/sample.txt ./wordplay thequickbrownfox
Example Command Line for CSV Report
$ caliper cycles --csv csvout ./wordplay thequickbrownfox
cycles Metrics Summed for Entire Run
This section describes the metrics summed over the entire run of your application under HP Caliper.
Metrics for Integrity Servers Intel Itanium 9300 processors Systems and older
BACK_END_BUBBLE.FE Full pipe bubbles in main pipe due to frontend. This is the
number of cycles lost (stall cycles) due to instruction cache,
ITLB, and branch execution stalls.
BE_EXE_BUBBLE.ALL Full pipe bubbles in main pipe due to execution unit stalls.
This is the number of cycles lost (stall cycles) due to stalls
caused by the execution unit.
BE_EXE_BUBBLE.FRALL Full Pipe Bubbles in Main Pipe due to FR/FR or FR/load
dependency stalls. This is the number of cycles lost (stall
cycles) due to FR/FR or FR/load dependency.
BE_EXE_BUBBLE.GRALL Full pipe bubbles in main pipe due to general
register/general register or general register/load
dependency stalls. This is the number of cycles lost (stall
cycles) due to general register/general register or general
register/load dependency.
BE_EXE_BUBBLE.GRGR Full Pipe Bubbles in Main Pipe due to GR/GR dependency
stalls. This is the number of cycles lost (stall cycles) due to
GR/GR dependency stalls.
BE_FLUSH_BUBBLE.ALL Full Pipe Bubbles in Main Pipe due to pipeline flushes. This
is the number of cycles lost (stall cycles) due to branch
misprediction or exception/interruption flush.
BE_L1D_FPU_BUBBLE.L1D Full Pipe Bubbles in Main Pipe due to L1D cache. This is the
number of cycles lost (stall cycles) due to L1D cache and
L1/L2 DTLB.
cycles Measurement Report Description 187