1 perf-diff(1)
2 ============
3
4 NAME
5 ----
6 perf-diff - Read perf.data files and display t
7
8 SYNOPSIS
9 --------
10 [verse]
11 'perf diff' [baseline file] [data file1] [[dat
12
13 DESCRIPTION
14 -----------
15 This command displays the performance differen
16 files captured via perf record.
17
18 If no parameters are passed it will assume per
19
20 The differential profile is displayed only for
21 specified perf.data files.
22
23 If no parameters are passed the samples will b
24 As the perf.data files could come from differe
25 could vary. So perf diff is based on the compa
26 symbols name.
27
28 OPTIONS
29 -------
30 -D::
31 --dump-raw-trace::
32 Dump raw trace in ASCII.
33
34 --kallsyms=<file>::
35 kallsyms pathname
36
37 -m::
38 --modules::
39 Load module symbols. WARNING: use only
40
41 -d::
42 --dsos=::
43 Only consider symbols in these dsos. C
44 file://filename entries. This option
45 of the Baseline/Delta column. See --p
46
47 -C::
48 --comms=::
49 Only consider symbols in these comms.
50 file://filename entries. This option
51 of the Baseline/Delta column. See --p
52
53 -S::
54 --symbols=::
55 Only consider these symbols. CSV that
56 file://filename entries. This option
57 of the Baseline/Delta column. See --p
58
59 -s::
60 --sort=::
61 Sort by key(s): pid, comm, dso, symbol
62 Please see description of --sort in th
63
64 -t::
65 --field-separator=::
66
67 Use a special separator character and
68 all occurrences of this separator in s
69 with a '.' character, that thus it's t
70
71 -v::
72 --verbose::
73 Be verbose, for instance, show the raw
74 diff.
75
76 -q::
77 --quiet::
78 Do not show any warnings or messages.
79
80 -f::
81 --force::
82 Don't do ownership validation.
83
84 --symfs=<directory>::
85 Look for files with symbols relative t
86
87 -b::
88 --baseline-only::
89 Show only items with match in baseline
90
91 -c::
92 --compute::
93 Differential computation selection - d
94 delta-abs (default is delta-abs). Def
95 diff.compute config option. See COMPA
96 more info.
97
98 --cycles-hist::
99 Report a histogram and the standard de
100 It can help us to judge if the reporte
101 not. This option should be used with '
102
103 -p::
104 --period::
105 Show period values for both compared h
106
107 -F::
108 --formula::
109 Show formula for given computation.
110
111 -o::
112 --order::
113 Specify compute sorting column number.
114 overhead and 1 (default) means sorting
115 (data from the first file other base ba
116 can be used only if enough data files a
117 The default value can be set using the
118
119 --percentage::
120 Determine how to display the overhead
121 Filters can be applied by --comms, --d
122
123 "relative" means it's relative to filt
124 sum of shown entries will be always 10
125 the original value before and after th
126
127 --time::
128 Analyze samples within given time wind
129 percent with multiple time ranges. Tim
130 or 'a%-b%,c%-%d,...'.
131
132 For example:
133
134 Select the second 10% time slice to di
135
136 perf diff --time 10%/2
137
138 Select from 0% to 10% time slice to di
139
140 perf diff --time 0%-10%
141
142 Select the first and the second 10% ti
143
144 perf diff --time 10%/1,10%/2
145
146 Select from 0% to 10% and 30% to 40% s
147
148 perf diff --time 0%-10%,30%-40%
149
150 It also supports analyzing samples wit
151 <start>,<stop>. Times have the format
152 is not given (i.e. time string is ',x.
153 the beginning of the file. If stop tim
154 string is 'x.y,') then analysis goes t
155 Multiple ranges can be separated by sp
156 to be quoted e.g. --time "1234.567,123
157 Time string is'a1.b1,c1.d1:a2.b2,c2.d2
158 for different perf.data files.
159
160 For example, we get the timestamp info
161
162 perf script -i perf.data.old
163 mgen 13940 [000] 3946.361400: ...
164
165 perf script -i perf.data
166 mgen 13940 [000] 3971.150589 ...
167
168 perf diff --time 3946.361400,:3971.1
169
170 It analyzes the perf.data.old from the
171 the end of perf.data.old and analyzes
172 timestamp 3971.150589 to the end of pe
173
174 --cpu:: Only diff samples for the list of CPUs
175 be provided as a comma-separated list
176 CPUs are specified with -: 0-2. Defaul
177 CPUs.
178
179 --pid=::
180 Only diff samples for given process ID
181
182 --tid=::
183 Only diff samples for given thread ID
184
185 --stream::
186 Enable hot streams comparison. Stream
187 aggregated by the branch records from
188
189 COMPARISON
190 ----------
191 The comparison is governed by the baseline fil
192 file is iterated for samples. All other perf.d
193 the command line are searched for the baseline
194 is found, specified computation is made and re
195
196 All samples from non-baseline perf.data files,
197 baseline entry, are displayed with empty space
198 and possible computation results (delta) in th
199
200 Example files samples:
201 - file A with samples f1, f2, f3, f4, f6
202 - file B with samples f2, f4, f5
203 - file C with samples f1, f2, f5
204
205 Example output:
206 x - computation takes place for pair
207 b - baseline sample percentage
208
209 - perf diff A B C
210
211 baseline/A compute/B compute/C samples
212 ---------------------------------------
213 b x f1
214 b x x f2
215 b f3
216 b x f4
217 b f6
218 x x f5
219
220 - perf diff B A C
221
222 baseline/B compute/A compute/C samples
223 ---------------------------------------
224 b x x f2
225 b x f4
226 b x f5
227 x x f1
228 x f3
229 x f6
230
231 - perf diff C B A
232
233 baseline/C compute/B compute/A samples
234 ---------------------------------------
235 b x f1
236 b x x f2
237 b x f5
238 x f3
239 x x f4
240 x f6
241
242 COMPARISON METHODS
243 ------------------
244 delta
245 ~~~~~
246 If specified the 'Delta' column is displayed w
247
248 d = A->period_percent - B->period_percent
249
250 with:
251 - A/B being matching hist entry from data/ba
252 (or perf.data/perf.data.old) respectively.
253
254 - period_percent being the % of the hist ent
255 single data file
256
257 - with filtering by -C, -d and/or -S, period
258 relative to how entries are filtered. Use
259 prevent such fluctuation.
260
261 delta-abs
262 ~~~~~~~~~
263 Same as 'delta` method, but sort the result wi
264
265 ratio
266 ~~~~~
267 If specified the 'Ratio' column is displayed w
268
269 r = A->period / B->period
270
271 with:
272 - A/B being matching hist entry from data/ba
273 (or perf.data/perf.data.old) respectively.
274
275 - period being the hist entry period value
276
277 wdiff:WEIGHT-B,WEIGHT-A
278 ~~~~~~~~~~~~~~~~~~~~~~~
279 If specified the 'Weighted diff' column is dis
280
281 d = B->period * WEIGHT-A - A->period * WEIG
282
283 - A/B being matching hist entry from data/ba
284 (or perf.data/perf.data.old) respectively.
285
286 - period being the hist entry period value
287
288 - WEIGHT-A/WEIGHT-B being user supplied weig
289 behind ':' separator like '-c wdiff:1,2'.
290 - WEIGHT-A being the weight of the data fi
291 - WEIGHT-B being the weight of the baselin
292
293 cycles
294 ~~~~~~
295 If specified the '[Program Block Range] Cycles
296 It displays the cycles difference of same prog
297 two perf.data. The program basic block is the
298
299 '[Program Block Range]' indicates the range of
300 Source line is reported if it can be found oth
301 instead.
302
303 SEE ALSO
304 --------
305 linkperf:perf-record[1], linkperf:perf-report[
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.