HP Caliper 5.3 User Guide (5900-1558, February 2011)

L1 instruction prefetch miss
percentage
Percentage of L1 instruction prefetches that are misses.
L1 instruction demand miss
percentage
Percentage of L1 instruction demand fetches that are misses.
L2 instruction demand misses Number of L2 instruction demand fetch misses.
L2 instruction prefetch misses Number of L2 instruction prefetch misses.
L2 instruction cache miss percentage Percentage of L2 instruction demand fetches and prefetches
that are misses.
L1 instruction cache misses per 1000
instructions retired
Number of instructions retired per L1 instruction cache miss.
L1 instruction prefetch misses per
1000 instructions retired
Number of instructions retired per L1 instruction prefetch
miss.
L1 instruction demand misses per
1000 instructions retired
Number of instructions retired per L1 instruction demand
fetch miss.
L2 instruction cache misses per 1000
instructions retired
Number of instructions retired per L2 instruction cache miss.
L2 instruction prefetch misses per
1000 instructions retired
Number of instructions retired per L2 instruction prefetch
miss.
L2 instruction demand misses per
1000 instructions retired
Number of instructions retired per L2 instruction demand
fetch miss.
icache Measurement Report Metrics
Table 27 (page 206) shows the information found in icache reports.
In this table, “program object” refers to any of the following:
• Thread
• Load module
• Function
• Statement
• Cache line
• Instruction
Table 27 Information in icache Measurement Reports
DescriptionColumn
Percent of the total for <metric> attributable to a given program object. The <metric> is the same
as the <metric> HP Caliper uses for sorting, except when the sort metric is avg-latency or
address, in which case sampled misses is used.
% Total <metric>
A running sum of the percent of total for <metric> accounted for by the given program object
and those listed above it.
Cumulat % of Total
Total number of sampled instruction cache misses attributed to the given program object.Sampled Icache
Misses
Number of cycles expended on instruction cache misses summed across samples for the given
program object.
Icache Latency
Cycles
Average number of cycles expended on instruction cache misses across samples for the given
program object.
Avg. Icache Laten.
Cycles
Kernel Thread ID suffixed with the the name of the routine that the thread will execute once it is
created.
Kernel Thread
Identification
Number
206 Descriptions of Measurement Reports