1perf-stat(1)
2============
3
4NAME
5----
6perf-stat - Run a command and gather performance counter statistics
7
8SYNOPSIS
9--------
10[verse]
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]
15
16DESCRIPTION
17-----------
18This command runs a command and gathers performance counter statistics
19from it.
20
21
22OPTIONS
23-------
24<command>...::
25	Any command you can specify in a shell.
26
27record::
28	See STAT RECORD.
29
30report::
31	See STAT REPORT.
32
33-e::
34--event=::
35	Select the PMU event. Selection can be:
36
37	- a symbolic event name (use 'perf list' to list all events)
38
39	- a raw PMU event (eventsel+umask) in the form of rNNN where NNN is a
40	  hexadecimal event descriptor.
41
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_source/devices/<pmu>/format/*
45
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_source/devices/<pmu>/format/*
51
52	Note that the last two syntaxes support prefix and glob matching in
53	the PMU name to simplify creation of events accross multiple instances
54	of the same type of PMU in large systems (e.g. memory controller PMUs).
55	Multiple PMU instances are typical for uncore PMUs, so the prefix
56	'uncore_' is also ignored when performing this match.
57
58
59-i::
60--no-inherit::
61        child tasks do not inherit counters
62-p::
63--pid=<pid>::
64        stat events on existing process id (comma separated list)
65
66-t::
67--tid=<tid>::
68        stat events on existing thread id (comma separated list)
69
70
71-a::
72--all-cpus::
73        system-wide collection from all CPUs (default if no target is specified)
74
75-c::
76--scale::
77	scale/normalize counter values
78
79-d::
80--detailed::
81	print more detailed statistics, can be specified up to 3 times
82
83	   -d:          detailed events, L1 and LLC data cache
84        -d -d:     more detailed events, dTLB and iTLB events
85     -d -d -d:     very detailed events, adding prefetch events
86
87-r::
88--repeat=<n>::
89	repeat command and print average + stddev (max: 100). 0 means forever.
90
91-B::
92--big-num::
93        print large numbers with thousands' separators according to locale
94
95-C::
96--cpu=::
97Count only on the list of CPUs provided. Multiple CPUs can be provided as a
98comma-separated list with no space: 0,1. Ranges of CPUs are specified with -: 0-2.
99In per-thread mode, this option is ignored. The -a option is still necessary
100to activate system-wide monitoring. Default is to count on all CPUs.
101
102-A::
103--no-aggr::
104Do not aggregate counts across all monitored CPUs.
105
106-n::
107--null::
108        null run - don't start any counters
109
110-v::
111--verbose::
112        be more verbose (show counter open errors, etc)
113
114-x SEP::
115--field-separator SEP::
116print counts using a CSV-style output to make it easy to import directly into
117spreadsheets. Columns are separated by the string specified in SEP.
118
119--table:: Display time for each run (-r option), in a table format, e.g.:
120
121  $ perf stat --null -r 5 --table perf bench sched pipe
122
123   Performance counter stats for 'perf bench sched pipe' (5 runs):
124
125             # Table of individual measurements:
126             5.189 (-0.293) #
127             5.189 (-0.294) #
128             5.186 (-0.296) #
129             5.663 (+0.181) ##
130             6.186 (+0.703) ####
131
132             # Final result:
133             5.483 +- 0.198 seconds time elapsed  ( +-  3.62% )
134
135-G name::
136--cgroup name::
137monitor only in the container (cgroup) called "name". This option is available only
138in per-cpu mode. The cgroup filesystem must be mounted. All threads belonging to
139container "name" are monitored when they run on the monitored CPUs. Multiple cgroups
140can be provided. Each cgroup is applied to the corresponding event, i.e., first cgroup
141to first event, second cgroup to second event and so on. It is possible to provide
142an empty cgroup (monitor all the time) using, e.g., -G foo,,bar. Cgroups must have
143corresponding events, i.e., they always refer to events defined earlier on the command
144line. If the user wants to track multiple events for a specific cgroup, the user can
145use '-e e1 -e e2 -G foo,foo' or just use '-e e1 -e e2 -G foo'.
146
147If wanting to monitor, say, 'cycles' for a cgroup and also for system wide, this
148command line can be used: 'perf stat -e cycles -G cgroup_name -a -e cycles'.
149
150-o file::
151--output file::
152Print the output into the designated file.
153
154--append::
155Append to the output file designated with the -o option. Ignored if -o is not specified.
156
157--log-fd::
158
159Log output to fd, instead of stderr.  Complementary to --output, and mutually exclusive
160with it.  --append may be used here.  Examples:
161     3>results  perf stat --log-fd 3          -- $cmd
162     3>>results perf stat --log-fd 3 --append -- $cmd
163
164--pre::
165--post::
166	Pre and post measurement hooks, e.g.:
167
168perf stat --repeat 10 --null --sync --pre 'make -s O=defconfig-build/clean' -- make -s -j64 O=defconfig-build/ bzImage
169
170-I msecs::
171--interval-print msecs::
172Print count deltas every N milliseconds (minimum: 1ms)
173The overhead percentage could be high in some cases, for instance with small, sub 100ms intervals.  Use with caution.
174	example: 'perf stat -I 1000 -e cycles -a sleep 5'
175
176--interval-count times::
177Print count deltas for fixed number of times.
178This option should be used together with "-I" option.
179	example: 'perf stat -I 1000 --interval-count 2 -e cycles -a'
180
181--interval-clear::
182Clear the screen before next interval.
183
184--timeout msecs::
185Stop the 'perf stat' session and print count deltas after N milliseconds (minimum: 10 ms).
186This option is not supported with the "-I" option.
187	example: 'perf stat --time 2000 -e cycles -a'
188
189--metric-only::
190Only print computed metrics. Print them in a single line.
191Don't show any raw values. Not supported with --per-thread.
192
193--per-socket::
194Aggregate counts per processor socket for system-wide mode measurements.  This
195is a useful mode to detect imbalance between sockets.  To enable this mode,
196use --per-socket in addition to -a. (system-wide).  The output includes the
197socket number and the number of online processors on that socket. This is
198useful to gauge the amount of aggregation.
199
200--per-core::
201Aggregate counts per physical processor for system-wide mode measurements.  This
202is a useful mode to detect imbalance between physical cores.  To enable this mode,
203use --per-core in addition to -a. (system-wide).  The output includes the
204core number and the number of online logical processors on that physical processor.
205
206--per-thread::
207Aggregate counts per monitored threads, when monitoring threads (-t option)
208or processes (-p option).
209
210-D msecs::
211--delay msecs::
212After starting the program, wait msecs before measuring. This is useful to
213filter out the startup phase of the program, which is often very different.
214
215-T::
216--transaction::
217
218Print statistics of transactional execution if supported.
219
220STAT RECORD
221-----------
222Stores stat data into perf data file.
223
224-o file::
225--output file::
226Output file name.
227
228STAT REPORT
229-----------
230Reads and reports stat data from perf data file.
231
232-i file::
233--input file::
234Input file name.
235
236--per-socket::
237Aggregate counts per processor socket for system-wide mode measurements.
238
239--per-core::
240Aggregate counts per physical processor for system-wide mode measurements.
241
242-M::
243--metrics::
244Print metrics or metricgroups specified in a comma separated list.
245For a group all metrics from the group are added.
246The events from the metrics are automatically measured.
247See perf list output for the possble metrics and metricgroups.
248
249-A::
250--no-aggr::
251Do not aggregate counts across all monitored CPUs.
252
253--topdown::
254Print top down level 1 metrics if supported by the CPU. This allows to
255determine bottle necks in the CPU pipeline for CPU bound workloads,
256by breaking the cycles consumed down into frontend bound, backend bound,
257bad speculation and retiring.
258
259Frontend bound means that the CPU cannot fetch and decode instructions fast
260enough. Backend bound means that computation or memory access is the bottle
261neck. Bad Speculation means that the CPU wasted cycles due to branch
262mispredictions and similar issues. Retiring means that the CPU computed without
263an apparently bottleneck. The bottleneck is only the real bottleneck
264if the workload is actually bound by the CPU and not by something else.
265
266For best results it is usually a good idea to use it with interval
267mode like -I 1000, as the bottleneck of workloads can change often.
268
269The top down metrics are collected per core instead of per
270CPU thread. Per core mode is automatically enabled
271and -a (global monitoring) is needed, requiring root rights or
272perf.perf_event_paranoid=-1.
273
274Topdown uses the full Performance Monitoring Unit, and needs
275disabling of the NMI watchdog (as root):
276echo 0 > /proc/sys/kernel/nmi_watchdog
277for best results. Otherwise the bottlenecks may be inconsistent
278on workload with changing phases.
279
280This enables --metric-only, unless overriden with --no-metric-only.
281
282To interpret the results it is usually needed to know on which
283CPUs the workload runs on. If needed the CPUs can be forced using
284taskset.
285
286--no-merge::
287Do not merge results from same PMUs.
288
289When multiple events are created from a single event specification,
290stat will, by default, aggregate the event counts and show the result
291in a single row. This option disables that behavior and shows
292the individual events and counts.
293
294Multiple events are created from a single event specification when:
2951. Prefix or glob matching is used for the PMU name.
2962. Aliases, which are listed immediately after the Kernel PMU events
297   by perf list, are used.
298
299--smi-cost::
300Measure SMI cost if msr/aperf/ and msr/smi/ events are supported.
301
302During the measurement, the /sys/device/cpu/freeze_on_smi will be set to
303freeze core counters on SMI.
304The aperf counter will not be effected by the setting.
305The cost of SMI can be measured by (aperf - unhalted core cycles).
306
307In practice, the percentages of SMI cycles is very useful for performance
308oriented analysis. --metric_only will be applied by default.
309The output is SMI cycles%, equals to (aperf - unhalted core cycles) / aperf
310
311Users who wants to get the actual value can apply --no-metric-only.
312
313EXAMPLES
314--------
315
316$ perf stat -- make
317
318   Performance counter stats for 'make':
319
320        83723.452481      task-clock:u (msec)       #    1.004 CPUs utilized
321                   0      context-switches:u        #    0.000 K/sec
322                   0      cpu-migrations:u          #    0.000 K/sec
323           3,228,188      page-faults:u             #    0.039 M/sec
324     229,570,665,834      cycles:u                  #    2.742 GHz
325     313,163,853,778      instructions:u            #    1.36  insn per cycle
326      69,704,684,856      branches:u                #  832.559 M/sec
327       2,078,861,393      branch-misses:u           #    2.98% of all branches
328
329        83.409183620 seconds time elapsed
330
331        74.684747000 seconds user
332         8.739217000 seconds sys
333
334TIMINGS
335-------
336As displayed in the example above we can display 3 types of timings.
337We always display the time the counters were enabled/alive:
338
339        83.409183620 seconds time elapsed
340
341For workload sessions we also display time the workloads spent in
342user/system lands:
343
344        74.684747000 seconds user
345         8.739217000 seconds sys
346
347Those times are the very same as displayed by the 'time' tool.
348
349CSV FORMAT
350----------
351
352With -x, perf stat is able to output a not-quite-CSV format output
353Commas in the output are not put into "". To make it easy to parse
354it is recommended to use a different character like -x \;
355
356The fields are in this order:
357
358	- optional usec time stamp in fractions of second (with -I xxx)
359	- optional CPU, core, or socket identifier
360	- optional number of logical CPUs aggregated
361	- counter value
362	- unit of the counter value or empty
363	- event name
364	- run time of counter
365	- percentage of measurement time the counter was running
366	- optional variance if multiple values are collected with -r
367	- optional metric value
368	- optional unit of metric
369
370Additional metrics may be printed with all earlier fields being empty.
371
372SEE ALSO
373--------
374linkperf:perf-top[1], linkperf:perf-list[1]
375