1 perf-stat(1) 1 perf-stat(1)
2 ============ 2 ============
3 3
4 NAME 4 NAME
5 ---- 5 ----
6 perf-stat - Run a command and gather performan 6 perf-stat - Run a command and gather performance counter statistics
7 7
8 SYNOPSIS 8 SYNOPSIS
9 -------- 9 --------
10 [verse] 10 [verse]
11 'perf stat' [-e <EVENT> | --event=EVENT] [-a] 11 'perf stat' [-e <EVENT> | --event=EVENT] [-a] <command>
12 'perf stat' [-e <EVENT> | --event=EVENT] [-a] !! 12 'perf stat' [-e <EVENT> | --event=EVENT] [-a] -- <command> [<options>]
13 'perf stat' [-e <EVENT> | --event=EVENT] [-a] !! 13 'perf stat' [-e <EVENT> | --event=EVENT] [-a] record [-o file] -- <command> [<options>]
14 'perf stat' report [-i file] 14 'perf stat' report [-i file]
15 15
16 DESCRIPTION 16 DESCRIPTION
17 ----------- 17 -----------
18 This command runs a command and gathers perfor 18 This command runs a command and gathers performance counter statistics
19 from it. 19 from it.
20 20
21 21
22 OPTIONS 22 OPTIONS
23 ------- 23 -------
24 <command>...:: 24 <command>...::
25 Any command you can specify in a shell 25 Any command you can specify in a shell.
26 26
27 record:: 27 record::
28 See STAT RECORD. 28 See STAT RECORD.
29 29
30 report:: 30 report::
31 See STAT REPORT. 31 See STAT REPORT.
32 32
33 -e:: 33 -e::
34 --event=:: 34 --event=::
35 Select the PMU event. Selection can be 35 Select the PMU event. Selection can be:
36 36
37 - a symbolic event name (use 'perf lis 37 - a symbolic event name (use 'perf list' to list all events)
38 38
39 - a raw PMU event in the form of rN wh !! 39 - a raw PMU event (eventsel+umask) in the form of rNNN where NNN is a
40 that represents the raw register enc !! 40 hexadecimal event descriptor.
41 event control registers as described <<
42 /sys/bus/event_source/devices/cpu/fo <<
43 <<
44 - a symbolic or raw PMU event followed <<
45 and a list of event modifiers, e.g., <<
46 linkperf:perf-list[1] man page for d <<
47 41
48 - a symbolically formed event like 'pm 42 - a symbolically formed event like 'pmu/param1=0x3,param2/' where
49 param1 and param2 are defined as for 43 param1 and param2 are defined as formats for the PMU in
50 /sys/bus/event_source/devices/<pmu>/ 44 /sys/bus/event_source/devices/<pmu>/format/*
51 45
52 'percore' is a event qualifier that 46 'percore' is a event qualifier that sums up the event counts for both
53 hardware threads in a core. For exam 47 hardware threads in a core. For example:
54 perf stat -A -a -e cpu/event,percore 48 perf stat -A -a -e cpu/event,percore=1/,otherevent ...
55 49
56 - a symbolically formed event like 'pm 50 - a symbolically formed event like 'pmu/config=M,config1=N,config2=K/'
57 where M, N, K are numbers (in decima 51 where M, N, K are numbers (in decimal, hex, octal format).
58 Acceptable values for each of 'confi 52 Acceptable values for each of 'config', 'config1' and 'config2'
59 parameters are defined by correspond 53 parameters are defined by corresponding entries in
60 /sys/bus/event_source/devices/<pmu>/ 54 /sys/bus/event_source/devices/<pmu>/format/*
61 55
62 Note that the last two syntaxes suppor 56 Note that the last two syntaxes support prefix and glob matching in
63 the PMU name to simplify creation of e 57 the PMU name to simplify creation of events across multiple instances
64 of the same type of PMU in large syste 58 of the same type of PMU in large systems (e.g. memory controller PMUs).
65 Multiple PMU instances are typical for 59 Multiple PMU instances are typical for uncore PMUs, so the prefix
66 'uncore_' is also ignored when perform 60 'uncore_' is also ignored when performing this match.
67 61
68 62
69 -i:: 63 -i::
70 --no-inherit:: 64 --no-inherit::
71 child tasks do not inherit counters 65 child tasks do not inherit counters
72 -p:: 66 -p::
73 --pid=<pid>:: 67 --pid=<pid>::
74 stat events on existing process id (co 68 stat events on existing process id (comma separated list)
75 69
76 -t:: 70 -t::
77 --tid=<tid>:: 71 --tid=<tid>::
78 stat events on existing thread id (com 72 stat events on existing thread id (comma separated list)
79 73
80 -b:: <<
81 --bpf-prog:: <<
82 stat events on existing bpf program id <<
83 requiring root rights. bpftool-prog co <<
84 id all bpf programs in the system. For <<
85 <<
86 # bpftool prog | head -n 1 <<
87 17247: tracepoint name sys_enter tag 192d5 <<
88 <<
89 # perf stat -e cycles,instructions --bpf-pro <<
90 <<
91 Performance counter stats for 'BPF program( <<
92 <<
93 85,967 cycles <<
94 28,982 instructions <<
95 <<
96 1.102235068 seconds time elapsed <<
97 <<
98 --bpf-counters:: <<
99 Use BPF programs to aggregate readings <<
100 allows multiple perf-stat sessions tha <<
101 instructions, etc.) to share hardware <<
102 To use BPF programs on common events b <<
103 "perf config stat.bpf-counter-events=< <<
104 <<
105 --bpf-attr-map:: <<
106 With option "--bpf-counters", differen <<
107 information about shared BPF programs <<
108 Use "--bpf-attr-map" to specify the pa <<
109 The default path is /sys/fs/bpf/perf_a <<
110 <<
111 ifdef::HAVE_LIBPFM[] <<
112 --pfm-events events:: <<
113 Select a PMU event using libpfm4 syntax (see h <<
114 including support for event filters. For examp <<
115 inst_retired:any_p:u:c=1:i'. More than one eve <<
116 option using the comma separator. Hardware eve <<
117 events cannot be mixed together. The latter mu <<
118 option. The -e option and this one can be mixe <<
119 can be grouped using the {} notation. <<
120 endif::HAVE_LIBPFM[] <<
121 74
122 -a:: 75 -a::
123 --all-cpus:: 76 --all-cpus::
124 system-wide collection from all CPUs ( 77 system-wide collection from all CPUs (default if no target is specified)
125 78
126 --no-scale:: 79 --no-scale::
127 Don't scale/normalize counter values 80 Don't scale/normalize counter values
128 81
129 -d:: 82 -d::
130 --detailed:: 83 --detailed::
131 print more detailed statistics, can be 84 print more detailed statistics, can be specified up to 3 times
132 85
133 -d: detailed events, L1 an 86 -d: detailed events, L1 and LLC data cache
134 -d -d: more detailed events, dTLB 87 -d -d: more detailed events, dTLB and iTLB events
135 -d -d -d: very detailed events, addin 88 -d -d -d: very detailed events, adding prefetch events
136 89
137 -r:: 90 -r::
138 --repeat=<n>:: 91 --repeat=<n>::
139 repeat command and print average + std 92 repeat command and print average + stddev (max: 100). 0 means forever.
140 93
141 -B:: 94 -B::
142 --big-num:: 95 --big-num::
143 print large numbers with thousands' se !! 96 print large numbers with thousands' separators according to locale
144 Enabled by default. Use "--no-big-num" <<
145 Default setting can be changed with "p <<
146 97
147 -C:: 98 -C::
148 --cpu=:: 99 --cpu=::
149 Count only on the list of CPUs provided. Multi 100 Count only on the list of CPUs provided. Multiple CPUs can be provided as a
150 comma-separated list with no space: 0,1. Range 101 comma-separated list with no space: 0,1. Ranges of CPUs are specified with -: 0-2.
151 In per-thread mode, this option is ignored. Th 102 In per-thread mode, this option is ignored. The -a option is still necessary
152 to activate system-wide monitoring. Default is 103 to activate system-wide monitoring. Default is to count on all CPUs.
153 104
154 -A:: 105 -A::
155 --no-aggr:: 106 --no-aggr::
156 Do not aggregate counts across all monitored C 107 Do not aggregate counts across all monitored CPUs.
157 108
158 -n:: 109 -n::
159 --null:: 110 --null::
160 null run - Don't start any counters. !! 111 null run - don't start any counters
161 <<
162 This can be useful to measure just elapsed wal <<
163 raw overhead of perf stat itself, without runn <<
164 112
165 -v:: 113 -v::
166 --verbose:: 114 --verbose::
167 be more verbose (show counter open err 115 be more verbose (show counter open errors, etc)
168 116
169 -x SEP:: 117 -x SEP::
170 --field-separator SEP:: 118 --field-separator SEP::
171 print counts using a CSV-style output to make 119 print counts using a CSV-style output to make it easy to import directly into
172 spreadsheets. Columns are separated by the str 120 spreadsheets. Columns are separated by the string specified in SEP.
173 121
174 --table:: Display time for each run (-r option 122 --table:: Display time for each run (-r option), in a table format, e.g.:
175 123
176 $ perf stat --null -r 5 --table perf bench s 124 $ perf stat --null -r 5 --table perf bench sched pipe
177 125
178 Performance counter stats for 'perf bench s 126 Performance counter stats for 'perf bench sched pipe' (5 runs):
179 127
180 # Table of individual measurement 128 # Table of individual measurements:
181 5.189 (-0.293) # 129 5.189 (-0.293) #
182 5.189 (-0.294) # 130 5.189 (-0.294) #
183 5.186 (-0.296) # 131 5.186 (-0.296) #
184 5.663 (+0.181) ## 132 5.663 (+0.181) ##
185 6.186 (+0.703) #### 133 6.186 (+0.703) ####
186 134
187 # Final result: 135 # Final result:
188 5.483 +- 0.198 seconds time elaps 136 5.483 +- 0.198 seconds time elapsed ( +- 3.62% )
189 137
190 -G name:: 138 -G name::
191 --cgroup name:: 139 --cgroup name::
192 monitor only in the container (cgroup) called 140 monitor only in the container (cgroup) called "name". This option is available only
193 in per-cpu mode. The cgroup filesystem must be 141 in per-cpu mode. The cgroup filesystem must be mounted. All threads belonging to
194 container "name" are monitored when they run o 142 container "name" are monitored when they run on the monitored CPUs. Multiple cgroups
195 can be provided. Each cgroup is applied to the 143 can be provided. Each cgroup is applied to the corresponding event, i.e., first cgroup
196 to first event, second cgroup to second event 144 to first event, second cgroup to second event and so on. It is possible to provide
197 an empty cgroup (monitor all the time) using, 145 an empty cgroup (monitor all the time) using, e.g., -G foo,,bar. Cgroups must have
198 corresponding events, i.e., they always refer 146 corresponding events, i.e., they always refer to events defined earlier on the command
199 line. If the user wants to track multiple even 147 line. If the user wants to track multiple events for a specific cgroup, the user can
200 use '-e e1 -e e2 -G foo,foo' or just use '-e e 148 use '-e e1 -e e2 -G foo,foo' or just use '-e e1 -e e2 -G foo'.
201 149
202 If wanting to monitor, say, 'cycles' for a cgr 150 If wanting to monitor, say, 'cycles' for a cgroup and also for system wide, this
203 command line can be used: 'perf stat -e cycles 151 command line can be used: 'perf stat -e cycles -G cgroup_name -a -e cycles'.
204 152
205 --for-each-cgroup name:: <<
206 Expand event list for each cgroup in "name" (a <<
207 by comma). It also support regex patterns to <<
208 effect that repeating -e option and -G option <<
209 cannot be used with -G/--cgroup option. <<
210 <<
211 -o file:: 153 -o file::
212 --output file:: 154 --output file::
213 Print the output into the designated file. 155 Print the output into the designated file.
214 156
215 --append:: 157 --append::
216 Append to the output file designated with the 158 Append to the output file designated with the -o option. Ignored if -o is not specified.
217 159
218 --log-fd:: 160 --log-fd::
219 161
220 Log output to fd, instead of stderr. Compleme 162 Log output to fd, instead of stderr. Complementary to --output, and mutually exclusive
221 with it. --append may be used here. Examples 163 with it. --append may be used here. Examples:
222 3>results perf stat --log-fd 3 !! 164 3>results perf stat --log-fd 3 -- $cmd
223 3>>results perf stat --log-fd 3 --append !! 165 3>>results perf stat --log-fd 3 --append -- $cmd
224 <<
225 --control=fifo:ctl-fifo[,ack-fifo]:: <<
226 --control=fd:ctl-fd[,ack-fd]:: <<
227 ctl-fifo / ack-fifo are opened and used as ctl <<
228 Listen on ctl-fd descriptor for command to con <<
229 'disable': disable events). Measurements can b <<
230 --delay=-1 option. Optionally send control com <<
231 to synchronize with the controlling process. E <<
232 disable events during measurements: <<
233 <<
234 #!/bin/bash <<
235 <<
236 ctl_dir=/tmp/ <<
237 <<
238 ctl_fifo=${ctl_dir}perf_ctl.fifo <<
239 test -p ${ctl_fifo} && unlink ${ctl_fifo} <<
240 mkfifo ${ctl_fifo} <<
241 exec {ctl_fd}<>${ctl_fifo} <<
242 <<
243 ctl_ack_fifo=${ctl_dir}perf_ctl_ack.fifo <<
244 test -p ${ctl_ack_fifo} && unlink ${ctl_ack_f <<
245 mkfifo ${ctl_ack_fifo} <<
246 exec {ctl_fd_ack}<>${ctl_ack_fifo} <<
247 <<
248 perf stat -D -1 -e cpu-cycles -a -I 1000 <<
249 --control fd:${ctl_fd},${ctl_fd_ack <<
250 \-- sleep 30 & <<
251 perf_pid=$! <<
252 <<
253 sleep 5 && echo 'enable' >&${ctl_fd} && read <<
254 sleep 10 && echo 'disable' >&${ctl_fd} && rea <<
255 <<
256 exec {ctl_fd_ack}>&- <<
257 unlink ${ctl_ack_fifo} <<
258 <<
259 exec {ctl_fd}>&- <<
260 unlink ${ctl_fifo} <<
261 <<
262 wait -n ${perf_pid} <<
263 exit $? <<
264 <<
265 166
266 --pre:: 167 --pre::
267 --post:: 168 --post::
268 Pre and post measurement hooks, e.g.: 169 Pre and post measurement hooks, e.g.:
269 170
270 perf stat --repeat 10 --null --sync --pre 'mak !! 171 perf stat --repeat 10 --null --sync --pre 'make -s O=defconfig-build/clean' -- make -s -j64 O=defconfig-build/ bzImage
271 172
272 -I msecs:: 173 -I msecs::
273 --interval-print msecs:: 174 --interval-print msecs::
274 Print count deltas every N milliseconds (minim 175 Print count deltas every N milliseconds (minimum: 1ms)
275 The overhead percentage could be high in some 176 The overhead percentage could be high in some cases, for instance with small, sub 100ms intervals. Use with caution.
276 example: 'perf stat -I 1000 -e cycles 177 example: 'perf stat -I 1000 -e cycles -a sleep 5'
277 178
278 If the metric exists, it is calculated by the <<
279 <<
280 --interval-count times:: 179 --interval-count times::
281 Print count deltas for fixed number of times. 180 Print count deltas for fixed number of times.
282 This option should be used together with "-I" 181 This option should be used together with "-I" option.
283 example: 'perf stat -I 1000 --interval 182 example: 'perf stat -I 1000 --interval-count 2 -e cycles -a'
284 183
285 --interval-clear:: 184 --interval-clear::
286 Clear the screen before next interval. 185 Clear the screen before next interval.
287 186
288 --timeout msecs:: 187 --timeout msecs::
289 Stop the 'perf stat' session and print count d 188 Stop the 'perf stat' session and print count deltas after N milliseconds (minimum: 10 ms).
290 This option is not supported with the "-I" opt 189 This option is not supported with the "-I" option.
291 example: 'perf stat --time 2000 -e cyc 190 example: 'perf stat --time 2000 -e cycles -a'
292 191
293 --metric-only:: 192 --metric-only::
294 Only print computed metrics. Print them in a s 193 Only print computed metrics. Print them in a single line.
295 Don't show any raw values. Not supported with 194 Don't show any raw values. Not supported with --per-thread.
296 195
297 --per-socket:: 196 --per-socket::
298 Aggregate counts per processor socket for syst 197 Aggregate counts per processor socket for system-wide mode measurements. This
299 is a useful mode to detect imbalance between s 198 is a useful mode to detect imbalance between sockets. To enable this mode,
300 use --per-socket in addition to -a. (system-wi 199 use --per-socket in addition to -a. (system-wide). The output includes the
301 socket number and the number of online process 200 socket number and the number of online processors on that socket. This is
302 useful to gauge the amount of aggregation. 201 useful to gauge the amount of aggregation.
303 202
304 --per-die:: 203 --per-die::
305 Aggregate counts per processor die for system- 204 Aggregate counts per processor die for system-wide mode measurements. This
306 is a useful mode to detect imbalance between d 205 is a useful mode to detect imbalance between dies. To enable this mode,
307 use --per-die in addition to -a. (system-wide) 206 use --per-die in addition to -a. (system-wide). The output includes the
308 die number and the number of online processors 207 die number and the number of online processors on that die. This is
309 useful to gauge the amount of aggregation. 208 useful to gauge the amount of aggregation.
310 209
311 --per-cluster:: <<
312 Aggregate counts per processor cluster for sys <<
313 is a useful mode to detect imbalance between c <<
314 use --per-cluster in addition to -a. (system-w <<
315 cluster number and the number of online proces <<
316 useful to gauge the amount of aggregation. The <<
317 related CPUs can be gotten from /sys/devices/s <<
318 <<
319 --per-cache:: <<
320 Aggregate counts per cache instance for system <<
321 default, the aggregation happens for the cache <<
322 in the system. To specify a particular level, <<
323 alongside the option in the format [Ll][1-9][0 <<
324 Using option "--per-cache=l3" or "--per-cache= <<
325 information at the boundary of the level 3 cac <<
326 <<
327 --per-core:: 210 --per-core::
328 Aggregate counts per physical processor for sy 211 Aggregate counts per physical processor for system-wide mode measurements. This
329 is a useful mode to detect imbalance between p 212 is a useful mode to detect imbalance between physical cores. To enable this mode,
330 use --per-core in addition to -a. (system-wide 213 use --per-core in addition to -a. (system-wide). The output includes the
331 core number and the number of online logical p 214 core number and the number of online logical processors on that physical processor.
332 215
333 --per-thread:: 216 --per-thread::
334 Aggregate counts per monitored threads, when m 217 Aggregate counts per monitored threads, when monitoring threads (-t option)
335 or processes (-p option). 218 or processes (-p option).
336 219
337 --per-node:: 220 --per-node::
338 Aggregate counts per NUMA nodes for system-wid 221 Aggregate counts per NUMA nodes for system-wide mode measurements. This
339 is a useful mode to detect imbalance between N 222 is a useful mode to detect imbalance between NUMA nodes. To enable this
340 mode, use --per-node in addition to -a. (syste 223 mode, use --per-node in addition to -a. (system-wide).
341 224
342 -D msecs:: 225 -D msecs::
343 --delay msecs:: 226 --delay msecs::
344 After starting the program, wait msecs before !! 227 After starting the program, wait msecs before measuring. This is useful to
345 disabled). This is useful to filter out the st !! 228 filter out the startup phase of the program, which is often very different.
346 which is often very different. <<
347 229
348 -T:: 230 -T::
349 --transaction:: 231 --transaction::
350 232
351 Print statistics of transactional execution if 233 Print statistics of transactional execution if supported.
352 234
353 --metric-no-group:: <<
354 By default, events to compute a metric are pla <<
355 group tries to enforce scheduling all or none <<
356 --metric-no-group option places events outside <<
357 increase the chance of the event being schedul <<
358 accuracy. However, as events may not be schedu <<
359 for metrics like instructions per cycle can be <<
360 may no longer be being measured at the same ti <<
361 <<
362 --metric-no-merge:: <<
363 By default metric events in different weak gro <<
364 group contains all the events needed by anothe <<
365 group will be eliminated reducing event multip <<
366 that certain groups of metrics sum to 100%. A <<
367 group is that the group may require multiplexi <<
368 small group that need not have multiplexing is <<
369 forbids the event merging logic from sharing e <<
370 may be used to increase accuracy in this case. <<
371 <<
372 --metric-no-threshold:: <<
373 Metric thresholds may increase the number of e <<
374 compute whether a metric has exceeded its thre <<
375 may not be desirable, for example, as the even <<
376 multiplexing. This option disables the adding <<
377 events for a metric. However, if there are suf <<
378 compute the threshold then the threshold is st <<
379 color the metric's computed value. <<
380 <<
381 --quiet:: <<
382 Don't print output, warnings or messages. This <<
383 record below to only write data to the perf.da <<
384 <<
385 STAT RECORD 235 STAT RECORD
386 ----------- 236 -----------
387 Stores stat data into perf data file. 237 Stores stat data into perf data file.
388 238
389 -o file:: 239 -o file::
390 --output file:: 240 --output file::
391 Output file name. 241 Output file name.
392 242
393 STAT REPORT 243 STAT REPORT
394 ----------- 244 -----------
395 Reads and reports stat data from perf data fil 245 Reads and reports stat data from perf data file.
396 246
397 -i file:: 247 -i file::
398 --input file:: 248 --input file::
399 Input file name. 249 Input file name.
400 250
401 --per-socket:: 251 --per-socket::
402 Aggregate counts per processor socket for syst 252 Aggregate counts per processor socket for system-wide mode measurements.
403 253
404 --per-die:: 254 --per-die::
405 Aggregate counts per processor die for system- 255 Aggregate counts per processor die for system-wide mode measurements.
406 256
407 --per-cluster:: <<
408 Aggregate counts perf processor cluster for sy <<
409 <<
410 --per-cache:: <<
411 Aggregate counts per cache instance for system <<
412 default, the aggregation happens for the cache <<
413 in the system. To specify a particular level, <<
414 alongside the option in the format [Ll][1-9][0 <<
415 option "--per-cache=l3" or "--per-cache=L3" wi <<
416 information at the boundary of the level 3 cac <<
417 <<
418 --per-core:: 257 --per-core::
419 Aggregate counts per physical processor for sy 258 Aggregate counts per physical processor for system-wide mode measurements.
420 259
421 -M:: 260 -M::
422 --metrics:: 261 --metrics::
423 Print metrics or metricgroups specified in a c 262 Print metrics or metricgroups specified in a comma separated list.
424 For a group all metrics from the group are add 263 For a group all metrics from the group are added.
425 The events from the metrics are automatically 264 The events from the metrics are automatically measured.
426 See perf list output for the possible metrics !! 265 See perf list output for the possble metrics and metricgroups.
427 <<
428 When threshold information is availabl <<
429 color red is used to signify a metric <<
430 while green shows it hasn't. The defau <<
431 no threshold information was available <<
432 couldn't be computed. <<
433 266
434 -A:: 267 -A::
435 --no-aggr:: 268 --no-aggr::
436 --no-merge:: !! 269 Do not aggregate counts across all monitored CPUs.
437 Do not aggregate/merge counts across monitored <<
438 <<
439 When multiple events are created from a single <<
440 stat will, by default, aggregate the event cou <<
441 in a single row. This option disables that beh <<
442 individual events and counts. <<
443 <<
444 Multiple events are created from a single even <<
445 <<
446 1. PID monitoring isn't requested and the syst <<
447 CPU. For example, a system with 8 SMT threa <<
448 opened on each thread and aggregation is pe <<
449 <<
450 2. Prefix or glob wildcard matching is used fo <<
451 example, multiple memory controller PMUs ma <<
452 suffix of _0, _1, etc. By default the event <<
453 combined if the PMU is specified without th <<
454 uncore_imc rather than uncore_imc_0. <<
455 <<
456 3. Aliases, which are listed immediately after <<
457 by perf list, are used. <<
458 <<
459 --hybrid-merge:: <<
460 Merge core event counts from all core PMUs. In <<
461 systems by default each core PMU will report i <<
462 separately. This option forces core PMU counts <<
463 a behavior closer to having a single CPU type <<
464 270
465 --topdown:: 271 --topdown::
466 Print top-down metrics supported by the CPU. T !! 272 Print top down level 1 metrics if supported by the CPU. This allows to
467 bottle necks in the CPU pipeline for CPU bound !! 273 determine bottle necks in the CPU pipeline for CPU bound workloads,
468 the cycles consumed down into frontend bound, !! 274 by breaking the cycles consumed down into frontend bound, backend bound,
469 speculation and retiring. !! 275 bad speculation and retiring.
470 276
471 Frontend bound means that the CPU cannot fetch 277 Frontend bound means that the CPU cannot fetch and decode instructions fast
472 enough. Backend bound means that computation o 278 enough. Backend bound means that computation or memory access is the bottle
473 neck. Bad Speculation means that the CPU waste 279 neck. Bad Speculation means that the CPU wasted cycles due to branch
474 mispredictions and similar issues. Retiring me 280 mispredictions and similar issues. Retiring means that the CPU computed without
475 an apparently bottleneck. The bottleneck is on 281 an apparently bottleneck. The bottleneck is only the real bottleneck
476 if the workload is actually bound by the CPU a 282 if the workload is actually bound by the CPU and not by something else.
477 283
478 For best results it is usually a good idea to 284 For best results it is usually a good idea to use it with interval
479 mode like -I 1000, as the bottleneck of worklo 285 mode like -I 1000, as the bottleneck of workloads can change often.
480 286
481 This enables --metric-only, unless overridden <<
482 <<
483 The following restrictions only apply to older <<
484 on newer CPUs (IceLake and later) TopDown can <<
485 <<
486 The top down metrics are collected per core in 287 The top down metrics are collected per core instead of per
487 CPU thread. Per core mode is automatically ena 288 CPU thread. Per core mode is automatically enabled
488 and -a (global monitoring) is needed, requirin 289 and -a (global monitoring) is needed, requiring root rights or
489 perf.perf_event_paranoid=-1. 290 perf.perf_event_paranoid=-1.
490 291
491 Topdown uses the full Performance Monitoring U 292 Topdown uses the full Performance Monitoring Unit, and needs
492 disabling of the NMI watchdog (as root): 293 disabling of the NMI watchdog (as root):
493 echo 0 > /proc/sys/kernel/nmi_watchdog 294 echo 0 > /proc/sys/kernel/nmi_watchdog
494 for best results. Otherwise the bottlenecks ma 295 for best results. Otherwise the bottlenecks may be inconsistent
495 on workload with changing phases. 296 on workload with changing phases.
496 297
>> 298 This enables --metric-only, unless overridden with --no-metric-only.
>> 299
497 To interpret the results it is usually needed 300 To interpret the results it is usually needed to know on which
498 CPUs the workload runs on. If needed the CPUs 301 CPUs the workload runs on. If needed the CPUs can be forced using
499 taskset. 302 taskset.
500 303
501 --record-tpebs:: !! 304 --no-merge::
502 Enable automatic sampling on Intel TPEBS retir !! 305 Do not merge results from same PMUs.
503 modifier). Without this option, perf would not !! 306
504 at runtime. Currently, a zero value is assigne !! 307 When multiple events are created from a single event specification,
505 this option is not set. The TPEBS hardware fea !! 308 stat will, by default, aggregate the event counts and show the result
506 Rapids microarchitecture. This option only exi !! 309 in a single row. This option disables that behavior and shows
507 Intel platforms with TPEBS feature. !! 310 the individual events and counts.
508 <<
509 --td-level:: <<
510 Print the top-down statistics that equal the i <<
511 users to print the interested top-down metrics <<
512 level 1 top-down metrics. <<
513 <<
514 As the higher levels gather more metrics and u <<
515 will be less accurate. By convention a metric <<
516 appending '_group' to it and this will increas <<
517 gathering all metrics for a level. For example <<
518 highlight 'tma_frontend_bound'. This metric ma <<
519 'tma_frontend_bound_group' with <<
520 'perf stat -M tma_frontend_bound_group...'. <<
521 311
522 Error out if the input is higher than the supp !! 312 Multiple events are created from a single event specification when:
>> 313 1. Prefix or glob matching is used for the PMU name.
>> 314 2. Aliases, which are listed immediately after the Kernel PMU events
>> 315 by perf list, are used.
523 316
524 --smi-cost:: 317 --smi-cost::
525 Measure SMI cost if msr/aperf/ and msr/smi/ ev 318 Measure SMI cost if msr/aperf/ and msr/smi/ events are supported.
526 319
527 During the measurement, the /sys/device/cpu/fr 320 During the measurement, the /sys/device/cpu/freeze_on_smi will be set to
528 freeze core counters on SMI. 321 freeze core counters on SMI.
529 The aperf counter will not be effected by the 322 The aperf counter will not be effected by the setting.
530 The cost of SMI can be measured by (aperf - un 323 The cost of SMI can be measured by (aperf - unhalted core cycles).
531 324
532 In practice, the percentages of SMI cycles is 325 In practice, the percentages of SMI cycles is very useful for performance
533 oriented analysis. --metric_only will be appli 326 oriented analysis. --metric_only will be applied by default.
534 The output is SMI cycles%, equals to (aperf - 327 The output is SMI cycles%, equals to (aperf - unhalted core cycles) / aperf
535 328
536 Users who wants to get the actual value can ap 329 Users who wants to get the actual value can apply --no-metric-only.
537 330
538 --all-kernel:: 331 --all-kernel::
539 Configure all used events to run in kernel spa 332 Configure all used events to run in kernel space.
540 333
541 --all-user:: 334 --all-user::
542 Configure all used events to run in user space 335 Configure all used events to run in user space.
543 336
544 --percore-show-thread:: <<
545 The event modifier "percore" has supported to <<
546 for all hardware threads in a core and show th <<
547 <<
548 This option with event modifier "percore" enab <<
549 counts for all hardware threads in a core but <<
550 hardware thread. This is essentially a replace <<
551 convenient for post processing. <<
552 <<
553 --summary:: <<
554 Print summary for interval mode (-I). <<
555 <<
556 --no-csv-summary:: <<
557 Don't print 'summary' at the first column for <<
558 This option must be used with -x and --summary <<
559 <<
560 This option can be enabled in perf config by s <<
561 'stat.no-csv-summary'. <<
562 <<
563 $ perf config stat.no-csv-summary=true <<
564 <<
565 --cputype:: <<
566 Only enable events on applying cpu with this t <<
567 (e.g. core or atom)" <<
568 <<
569 EXAMPLES 337 EXAMPLES
570 -------- 338 --------
571 339
572 $ perf stat \-- make !! 340 $ perf stat -- make
573 341
574 Performance counter stats for 'make': 342 Performance counter stats for 'make':
575 343
576 83723.452481 task-clock:u (msec) 344 83723.452481 task-clock:u (msec) # 1.004 CPUs utilized
577 0 context-switches:u 345 0 context-switches:u # 0.000 K/sec
578 0 cpu-migrations:u 346 0 cpu-migrations:u # 0.000 K/sec
579 3,228,188 page-faults:u 347 3,228,188 page-faults:u # 0.039 M/sec
580 229,570,665,834 cycles:u 348 229,570,665,834 cycles:u # 2.742 GHz
581 313,163,853,778 instructions:u 349 313,163,853,778 instructions:u # 1.36 insn per cycle
582 69,704,684,856 branches:u 350 69,704,684,856 branches:u # 832.559 M/sec
583 2,078,861,393 branch-misses:u 351 2,078,861,393 branch-misses:u # 2.98% of all branches
584 352
585 83.409183620 seconds time elapsed 353 83.409183620 seconds time elapsed
586 354
587 74.684747000 seconds user 355 74.684747000 seconds user
588 8.739217000 seconds sys 356 8.739217000 seconds sys
589 357
590 TIMINGS 358 TIMINGS
591 ------- 359 -------
592 As displayed in the example above we can displ 360 As displayed in the example above we can display 3 types of timings.
593 We always display the time the counters were e 361 We always display the time the counters were enabled/alive:
594 362
595 83.409183620 seconds time elapsed 363 83.409183620 seconds time elapsed
596 364
597 For workload sessions we also display time the 365 For workload sessions we also display time the workloads spent in
598 user/system lands: 366 user/system lands:
599 367
600 74.684747000 seconds user 368 74.684747000 seconds user
601 8.739217000 seconds sys 369 8.739217000 seconds sys
602 370
603 Those times are the very same as displayed by 371 Those times are the very same as displayed by the 'time' tool.
604 372
605 CSV FORMAT 373 CSV FORMAT
606 ---------- 374 ----------
607 375
608 With -x, perf stat is able to output a not-qui 376 With -x, perf stat is able to output a not-quite-CSV format output
609 Commas in the output are not put into "". To m 377 Commas in the output are not put into "". To make it easy to parse
610 it is recommended to use a different character 378 it is recommended to use a different character like -x \;
611 379
612 The fields are in this order: 380 The fields are in this order:
613 381
614 - optional usec time stamp in fraction 382 - optional usec time stamp in fractions of second (with -I xxx)
615 - optional CPU, core, or socket identi 383 - optional CPU, core, or socket identifier
616 - optional number of logical CPUs aggr 384 - optional number of logical CPUs aggregated
617 - counter value 385 - counter value
618 - unit of the counter value or empty 386 - unit of the counter value or empty
619 - event name 387 - event name
620 - run time of counter 388 - run time of counter
621 - percentage of measurement time the c 389 - percentage of measurement time the counter was running
622 - optional variance if multiple values 390 - optional variance if multiple values are collected with -r
623 - optional metric value 391 - optional metric value
624 - optional unit of metric 392 - optional unit of metric
625 393
626 Additional metrics may be printed with all ear 394 Additional metrics may be printed with all earlier fields being empty.
627 <<
628 include::intel-hybrid.txt[] <<
629 <<
630 JSON FORMAT <<
631 ----------- <<
632 <<
633 With -j, perf stat is able to print out a JSON <<
634 that can be used for parsing. <<
635 <<
636 - timestamp : optional usec time stamp in frac <<
637 - optional aggregate options: <<
638 - core : core identifier (with <<
639 - die : die identifier (with - <<
640 - socket : socket identifier ( <<
641 - node : node identifier (with <<
642 - thread : thread identifier ( <<
643 - counter-value : counter value <<
644 - unit : unit of the counter value or empty <<
645 - event : event name <<
646 - variance : optional variance if multiple val <<
647 - runtime : run time of counter <<
648 - metric-value : optional metric value <<
649 - metric-unit : optional unit of metric <<
650 395
651 SEE ALSO 396 SEE ALSO
652 -------- 397 --------
653 linkperf:perf-top[1], linkperf:perf-list[1] 398 linkperf:perf-top[1], linkperf:perf-list[1]
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.