6 perf-stat - Run a command and gather performance counter statistics
11 'perf stat' [-e <EVENT> | --event=EVENT] [-a] <command>
12 'perf stat' [-e <EVENT> | --event=EVENT] [-a] -- <command> [<options>]
13 'perf stat' [-e <EVENT> | --event=EVENT] [-a] record [-o file] -- <command> [<options>]
14 'perf stat' report [-i file]
18 This command runs a command and gathers performance counter statistics
25 Any command you can specify in a shell.
35 Select the PMU event. Selection can be:
37 - a symbolic event name (use 'perf list' to list all events)
39 - a raw PMU event (eventsel+umask) in the form of rNNN where NNN is a
40 hexadecimal event descriptor.
42 - a symbolically formed event like 'pmu/param1=0x3,param2/' where
43 param1 and param2 are defined as formats for the PMU in
44 /sys/bus/event_sources/devices/<pmu>/format/*
46 - a symbolically formed event like 'pmu/config=M,config1=N,config2=K/'
47 where M, N, K are numbers (in decimal, hex, octal format).
48 Acceptable values for each of 'config', 'config1' and 'config2'
49 parameters are defined by corresponding entries in
50 /sys/bus/event_sources/devices/<pmu>/format/*
54 child tasks do not inherit counters
57 stat events on existing process id (comma separated list)
61 stat events on existing thread id (comma separated list)
66 system-wide collection from all CPUs
70 scale/normalize counter values
74 repeat command and print average + stddev (max: 100). 0 means forever.
78 print large numbers with thousands' separators according to locale
82 Count only on the list of CPUs provided. Multiple CPUs can be provided as a
83 comma-separated list with no space: 0,1. Ranges of CPUs are specified with -: 0-2.
84 In per-thread mode, this option is ignored. The -a option is still necessary
85 to activate system-wide monitoring. Default is to count on all CPUs.
89 Do not aggregate counts across all monitored CPUs in system-wide mode (-a).
90 This option is only valid in system-wide mode.
94 null run - don't start any counters
98 be more verbose (show counter open errors, etc)
101 --field-separator SEP::
102 print counts using a CSV-style output to make it easy to import directly into
103 spreadsheets. Columns are separated by the string specified in SEP.
107 monitor only in the container (cgroup) called "name". This option is available only
108 in per-cpu mode. The cgroup filesystem must be mounted. All threads belonging to
109 container "name" are monitored when they run on the monitored CPUs. Multiple cgroups
110 can be provided. Each cgroup is applied to the corresponding event, i.e., first cgroup
111 to first event, second cgroup to second event and so on. It is possible to provide
112 an empty cgroup (monitor all the time) using, e.g., -G foo,,bar. Cgroups must have
113 corresponding events, i.e., they always refer to events defined earlier on the command
118 Print the output into the designated file.
121 Append to the output file designated with the -o option. Ignored if -o is not specified.
125 Log output to fd, instead of stderr. Complementary to --output, and mutually exclusive
126 with it. --append may be used here. Examples:
127 3>results perf stat --log-fd 3 -- $cmd
128 3>>results perf stat --log-fd 3 --append -- $cmd
132 Pre and post measurement hooks, e.g.:
134 perf stat --repeat 10 --null --sync --pre 'make -s O=defconfig-build/clean' -- make -s -j64 O=defconfig-build/ bzImage
137 --interval-print msecs::
138 Print count deltas every N milliseconds (minimum: 10ms)
139 The overhead percentage could be high in some cases, for instance with small, sub 100ms intervals. Use with caution.
140 example: 'perf stat -I 1000 -e cycles -a sleep 5'
143 Aggregate counts per processor socket for system-wide mode measurements. This
144 is a useful mode to detect imbalance between sockets. To enable this mode,
145 use --per-socket in addition to -a. (system-wide). The output includes the
146 socket number and the number of online processors on that socket. This is
147 useful to gauge the amount of aggregation.
150 Aggregate counts per physical processor for system-wide mode measurements. This
151 is a useful mode to detect imbalance between physical cores. To enable this mode,
152 use --per-core in addition to -a. (system-wide). The output includes the
153 core number and the number of online logical processors on that physical processor.
156 Aggregate counts per monitored threads, when monitoring threads (-t option)
157 or processes (-p option).
161 After starting the program, wait msecs before measuring. This is useful to
162 filter out the startup phase of the program, which is often very different.
167 Print statistics of transactional execution if supported.
171 Stores stat data into perf data file.
179 Reads and reports stat data from perf data file.
186 Aggregate counts per processor socket for system-wide mode measurements.
189 Aggregate counts per physical processor for system-wide mode measurements.
193 Do not aggregate counts across all monitored CPUs.
199 $ perf stat -- make -j
201 Performance counter stats for 'make -j':
203 8117.370256 task clock ticks # 11.281 CPU utilization factor
204 678 context switches # 0.000 M/sec
205 133 CPU migrations # 0.000 M/sec
206 235724 pagefaults # 0.029 M/sec
207 24821162526 CPU cycles # 3057.784 M/sec
208 18687303457 instructions # 2302.138 M/sec
209 172158895 cache references # 21.209 M/sec
210 27075259 cache misses # 3.335 M/sec
212 Wall-clock time elapsed: 719.554352 msecs
216 linkperf:perf-top[1], linkperf:perf-list[1]