1 perf-diff(1) 1 perf-diff(1)
2 ============ 2 ============
3 3
4 NAME 4 NAME
5 ---- 5 ----
6 perf-diff - Read perf.data files and display t 6 perf-diff - Read perf.data files and display the differential profile
7 7
8 SYNOPSIS 8 SYNOPSIS
9 -------- 9 --------
10 [verse] 10 [verse]
11 'perf diff' [baseline file] [data file1] [[dat 11 'perf diff' [baseline file] [data file1] [[data file2] ... ]
12 12
13 DESCRIPTION 13 DESCRIPTION
14 ----------- 14 -----------
15 This command displays the performance differen 15 This command displays the performance difference amongst two or more perf.data
16 files captured via perf record. 16 files captured via perf record.
17 17
18 If no parameters are passed it will assume per 18 If no parameters are passed it will assume perf.data.old and perf.data.
19 19
20 The differential profile is displayed only for 20 The differential profile is displayed only for events matching both
21 specified perf.data files. 21 specified perf.data files.
22 22
23 If no parameters are passed the samples will b 23 If no parameters are passed the samples will be sorted by dso and symbol.
24 As the perf.data files could come from differe 24 As the perf.data files could come from different binaries, the symbols addresses
25 could vary. So perf diff is based on the compa 25 could vary. So perf diff is based on the comparison of the files and
26 symbols name. 26 symbols name.
27 27
28 OPTIONS 28 OPTIONS
29 ------- 29 -------
30 -D:: 30 -D::
31 --dump-raw-trace:: 31 --dump-raw-trace::
32 Dump raw trace in ASCII. 32 Dump raw trace in ASCII.
33 33
34 --kallsyms=<file>:: 34 --kallsyms=<file>::
35 kallsyms pathname 35 kallsyms pathname
36 36
37 -m:: 37 -m::
38 --modules:: 38 --modules::
39 Load module symbols. WARNING: use only 39 Load module symbols. WARNING: use only with -k and LIVE kernel
40 40
41 -d:: 41 -d::
42 --dsos=:: 42 --dsos=::
43 Only consider symbols in these dsos. C 43 Only consider symbols in these dsos. CSV that understands
44 file://filename entries. This option 44 file://filename entries. This option will affect the percentage
45 of the Baseline/Delta column. See --p 45 of the Baseline/Delta column. See --percentage for more info.
46 46
47 -C:: 47 -C::
48 --comms=:: 48 --comms=::
49 Only consider symbols in these comms. 49 Only consider symbols in these comms. CSV that understands
50 file://filename entries. This option 50 file://filename entries. This option will affect the percentage
51 of the Baseline/Delta column. See --p 51 of the Baseline/Delta column. See --percentage for more info.
52 52
53 -S:: 53 -S::
54 --symbols=:: 54 --symbols=::
55 Only consider these symbols. CSV that 55 Only consider these symbols. CSV that understands
56 file://filename entries. This option 56 file://filename entries. This option will affect the percentage
57 of the Baseline/Delta column. See --p 57 of the Baseline/Delta column. See --percentage for more info.
58 58
59 -s:: 59 -s::
60 --sort=:: 60 --sort=::
61 Sort by key(s): pid, comm, dso, symbol 61 Sort by key(s): pid, comm, dso, symbol, cpu, parent, srcline.
62 Please see description of --sort in th 62 Please see description of --sort in the perf-report man page.
63 63
64 -t:: 64 -t::
65 --field-separator=:: 65 --field-separator=::
66 66
67 Use a special separator character and 67 Use a special separator character and don't pad with spaces, replacing
68 all occurrences of this separator in s 68 all occurrences of this separator in symbol names (and other output)
69 with a '.' character, that thus it's t 69 with a '.' character, that thus it's the only non valid separator.
70 70
71 -v:: 71 -v::
72 --verbose:: 72 --verbose::
73 Be verbose, for instance, show the raw 73 Be verbose, for instance, show the raw counts in addition to the
74 diff. 74 diff.
75 75
76 -q:: 76 -q::
77 --quiet:: 77 --quiet::
78 Do not show any warnings or messages. !! 78 Do not show any message. (Suppress -v)
79 79
80 -f:: 80 -f::
81 --force:: 81 --force::
82 Don't do ownership validation. 82 Don't do ownership validation.
83 83
84 --symfs=<directory>:: 84 --symfs=<directory>::
85 Look for files with symbols relative t 85 Look for files with symbols relative to this directory.
86 86
87 -b:: 87 -b::
88 --baseline-only:: 88 --baseline-only::
89 Show only items with match in baseline 89 Show only items with match in baseline.
90 90
91 -c:: 91 -c::
92 --compute:: 92 --compute::
93 Differential computation selection - d 93 Differential computation selection - delta, ratio, wdiff, cycles,
94 delta-abs (default is delta-abs). Def 94 delta-abs (default is delta-abs). Default can be changed using
95 diff.compute config option. See COMPA 95 diff.compute config option. See COMPARISON METHODS section for
96 more info. 96 more info.
97 97
98 --cycles-hist:: 98 --cycles-hist::
99 Report a histogram and the standard de 99 Report a histogram and the standard deviation for cycles data.
100 It can help us to judge if the reporte 100 It can help us to judge if the reported cycles data is noisy or
101 not. This option should be used with ' 101 not. This option should be used with '-c cycles'.
102 102
103 -p:: 103 -p::
104 --period:: 104 --period::
105 Show period values for both compared h 105 Show period values for both compared hist entries.
106 106
107 -F:: 107 -F::
108 --formula:: 108 --formula::
109 Show formula for given computation. 109 Show formula for given computation.
110 110
111 -o:: 111 -o::
112 --order:: 112 --order::
113 Specify compute sorting column number. 113 Specify compute sorting column number. 0 means sorting by baseline
114 overhead and 1 (default) means sorting 114 overhead and 1 (default) means sorting by computed value of column 1
115 (data from the first file other base ba 115 (data from the first file other base baseline). Values more than 1
116 can be used only if enough data files a 116 can be used only if enough data files are provided.
117 The default value can be set using the 117 The default value can be set using the diff.order config option.
118 118
119 --percentage:: 119 --percentage::
120 Determine how to display the overhead 120 Determine how to display the overhead percentage of filtered entries.
121 Filters can be applied by --comms, --d 121 Filters can be applied by --comms, --dsos and/or --symbols options.
122 122
123 "relative" means it's relative to filt 123 "relative" means it's relative to filtered entries only so that the
124 sum of shown entries will be always 10 124 sum of shown entries will be always 100%. "absolute" means it retains
125 the original value before and after th 125 the original value before and after the filter is applied.
126 126
127 --time:: 127 --time::
128 Analyze samples within given time wind 128 Analyze samples within given time window. It supports time
129 percent with multiple time ranges. Tim 129 percent with multiple time ranges. Time string is 'a%/n,b%/m,...'
130 or 'a%-b%,c%-%d,...'. 130 or 'a%-b%,c%-%d,...'.
131 131
132 For example: 132 For example:
133 133
134 Select the second 10% time slice to di 134 Select the second 10% time slice to diff:
135 135
136 perf diff --time 10%/2 136 perf diff --time 10%/2
137 137
138 Select from 0% to 10% time slice to di 138 Select from 0% to 10% time slice to diff:
139 139
140 perf diff --time 0%-10% 140 perf diff --time 0%-10%
141 141
142 Select the first and the second 10% ti 142 Select the first and the second 10% time slices to diff:
143 143
144 perf diff --time 10%/1,10%/2 144 perf diff --time 10%/1,10%/2
145 145
146 Select from 0% to 10% and 30% to 40% s 146 Select from 0% to 10% and 30% to 40% slices to diff:
147 147
148 perf diff --time 0%-10%,30%-40% 148 perf diff --time 0%-10%,30%-40%
149 149
150 It also supports analyzing samples wit 150 It also supports analyzing samples within a given time window
151 <start>,<stop>. Times have the format 151 <start>,<stop>. Times have the format seconds.nanoseconds. If 'start'
152 is not given (i.e. time string is ',x. 152 is not given (i.e. time string is ',x.y') then analysis starts at
153 the beginning of the file. If stop tim 153 the beginning of the file. If stop time is not given (i.e. time
154 string is 'x.y,') then analysis goes t 154 string is 'x.y,') then analysis goes to the end of the file.
155 Multiple ranges can be separated by sp 155 Multiple ranges can be separated by spaces, which requires the argument
156 to be quoted e.g. --time "1234.567,123 156 to be quoted e.g. --time "1234.567,1234.789 1235,"
157 Time string is'a1.b1,c1.d1:a2.b2,c2.d2 157 Time string is'a1.b1,c1.d1:a2.b2,c2.d2'. Use ':' to separate timestamps
158 for different perf.data files. 158 for different perf.data files.
159 159
160 For example, we get the timestamp info 160 For example, we get the timestamp information from 'perf script'.
161 161
162 perf script -i perf.data.old 162 perf script -i perf.data.old
163 mgen 13940 [000] 3946.361400: ... 163 mgen 13940 [000] 3946.361400: ...
164 164
165 perf script -i perf.data 165 perf script -i perf.data
166 mgen 13940 [000] 3971.150589 ... 166 mgen 13940 [000] 3971.150589 ...
167 167
168 perf diff --time 3946.361400,:3971.1 168 perf diff --time 3946.361400,:3971.150589,
169 169
170 It analyzes the perf.data.old from the 170 It analyzes the perf.data.old from the timestamp 3946.361400 to
171 the end of perf.data.old and analyzes 171 the end of perf.data.old and analyzes the perf.data from the
172 timestamp 3971.150589 to the end of pe 172 timestamp 3971.150589 to the end of perf.data.
173 173
174 --cpu:: Only diff samples for the list of CPUs 174 --cpu:: Only diff samples for the list of CPUs provided. Multiple CPUs can
175 be provided as a comma-separated list 175 be provided as a comma-separated list with no space: 0,1. Ranges of
176 CPUs are specified with -: 0-2. Defaul 176 CPUs are specified with -: 0-2. Default is to report samples on all
177 CPUs. 177 CPUs.
178 178
179 --pid=:: 179 --pid=::
180 Only diff samples for given process ID 180 Only diff samples for given process ID (comma separated list).
181 181
182 --tid=:: 182 --tid=::
183 Only diff samples for given thread ID 183 Only diff samples for given thread ID (comma separated list).
184 184
185 --stream:: 185 --stream::
186 Enable hot streams comparison. Stream 186 Enable hot streams comparison. Stream can be a callchain which is
187 aggregated by the branch records from 187 aggregated by the branch records from samples.
188 188
189 COMPARISON 189 COMPARISON
190 ---------- 190 ----------
191 The comparison is governed by the baseline fil 191 The comparison is governed by the baseline file. The baseline perf.data
192 file is iterated for samples. All other perf.d 192 file is iterated for samples. All other perf.data files specified on
193 the command line are searched for the baseline 193 the command line are searched for the baseline sample pair. If the pair
194 is found, specified computation is made and re 194 is found, specified computation is made and result is displayed.
195 195
196 All samples from non-baseline perf.data files, 196 All samples from non-baseline perf.data files, that do not match any
197 baseline entry, are displayed with empty space 197 baseline entry, are displayed with empty space within baseline column
198 and possible computation results (delta) in th 198 and possible computation results (delta) in their related column.
199 199
200 Example files samples: 200 Example files samples:
201 - file A with samples f1, f2, f3, f4, f6 201 - file A with samples f1, f2, f3, f4, f6
202 - file B with samples f2, f4, f5 202 - file B with samples f2, f4, f5
203 - file C with samples f1, f2, f5 203 - file C with samples f1, f2, f5
204 204
205 Example output: 205 Example output:
206 x - computation takes place for pair 206 x - computation takes place for pair
207 b - baseline sample percentage 207 b - baseline sample percentage
208 208
209 - perf diff A B C 209 - perf diff A B C
210 210
211 baseline/A compute/B compute/C samples 211 baseline/A compute/B compute/C samples
212 --------------------------------------- 212 ---------------------------------------
213 b x f1 213 b x f1
214 b x x f2 214 b x x f2
215 b f3 215 b f3
216 b x f4 216 b x f4
217 b f6 217 b f6
218 x x f5 218 x x f5
219 219
220 - perf diff B A C 220 - perf diff B A C
221 221
222 baseline/B compute/A compute/C samples 222 baseline/B compute/A compute/C samples
223 --------------------------------------- 223 ---------------------------------------
224 b x x f2 224 b x x f2
225 b x f4 225 b x f4
226 b x f5 226 b x f5
227 x x f1 227 x x f1
228 x f3 228 x f3
229 x f6 229 x f6
230 230
231 - perf diff C B A 231 - perf diff C B A
232 232
233 baseline/C compute/B compute/A samples 233 baseline/C compute/B compute/A samples
234 --------------------------------------- 234 ---------------------------------------
235 b x f1 235 b x f1
236 b x x f2 236 b x x f2
237 b x f5 237 b x f5
238 x f3 238 x f3
239 x x f4 239 x x f4
240 x f6 240 x f6
241 241
242 COMPARISON METHODS 242 COMPARISON METHODS
243 ------------------ 243 ------------------
244 delta 244 delta
245 ~~~~~ 245 ~~~~~
246 If specified the 'Delta' column is displayed w 246 If specified the 'Delta' column is displayed with value 'd' computed as:
247 247
248 d = A->period_percent - B->period_percent 248 d = A->period_percent - B->period_percent
249 249
250 with: 250 with:
251 - A/B being matching hist entry from data/ba 251 - A/B being matching hist entry from data/baseline file specified
252 (or perf.data/perf.data.old) respectively. 252 (or perf.data/perf.data.old) respectively.
253 253
254 - period_percent being the % of the hist ent 254 - period_percent being the % of the hist entry period value within
255 single data file 255 single data file
256 256
257 - with filtering by -C, -d and/or -S, period 257 - with filtering by -C, -d and/or -S, period_percent might be changed
258 relative to how entries are filtered. Use 258 relative to how entries are filtered. Use --percentage=absolute to
259 prevent such fluctuation. 259 prevent such fluctuation.
260 260
261 delta-abs 261 delta-abs
262 ~~~~~~~~~ 262 ~~~~~~~~~
263 Same as 'delta` method, but sort the result wi 263 Same as 'delta` method, but sort the result with the absolute values.
264 264
265 ratio 265 ratio
266 ~~~~~ 266 ~~~~~
267 If specified the 'Ratio' column is displayed w 267 If specified the 'Ratio' column is displayed with value 'r' computed as:
268 268
269 r = A->period / B->period 269 r = A->period / B->period
270 270
271 with: 271 with:
272 - A/B being matching hist entry from data/ba 272 - A/B being matching hist entry from data/baseline file specified
273 (or perf.data/perf.data.old) respectively. 273 (or perf.data/perf.data.old) respectively.
274 274
275 - period being the hist entry period value 275 - period being the hist entry period value
276 276
277 wdiff:WEIGHT-B,WEIGHT-A 277 wdiff:WEIGHT-B,WEIGHT-A
278 ~~~~~~~~~~~~~~~~~~~~~~~ 278 ~~~~~~~~~~~~~~~~~~~~~~~
279 If specified the 'Weighted diff' column is dis 279 If specified the 'Weighted diff' column is displayed with value 'd' computed as:
280 280
281 d = B->period * WEIGHT-A - A->period * WEIG 281 d = B->period * WEIGHT-A - A->period * WEIGHT-B
282 282
283 - A/B being matching hist entry from data/ba 283 - A/B being matching hist entry from data/baseline file specified
284 (or perf.data/perf.data.old) respectively. 284 (or perf.data/perf.data.old) respectively.
285 285
286 - period being the hist entry period value 286 - period being the hist entry period value
287 287
288 - WEIGHT-A/WEIGHT-B being user supplied weig 288 - WEIGHT-A/WEIGHT-B being user supplied weights in the the '-c' option
289 behind ':' separator like '-c wdiff:1,2'. 289 behind ':' separator like '-c wdiff:1,2'.
290 - WEIGHT-A being the weight of the data fi 290 - WEIGHT-A being the weight of the data file
291 - WEIGHT-B being the weight of the baselin 291 - WEIGHT-B being the weight of the baseline data file
292 292
293 cycles 293 cycles
294 ~~~~~~ 294 ~~~~~~
295 If specified the '[Program Block Range] Cycles 295 If specified the '[Program Block Range] Cycles Diff' column is displayed.
296 It displays the cycles difference of same prog 296 It displays the cycles difference of same program basic block amongst
297 two perf.data. The program basic block is the 297 two perf.data. The program basic block is the code between two branches.
298 298
299 '[Program Block Range]' indicates the range of 299 '[Program Block Range]' indicates the range of a program basic block.
300 Source line is reported if it can be found oth 300 Source line is reported if it can be found otherwise uses symbol+offset
301 instead. 301 instead.
302 302
303 SEE ALSO 303 SEE ALSO
304 -------- 304 --------
305 linkperf:perf-record[1], linkperf:perf-report[ 305 linkperf:perf-record[1], linkperf:perf-report[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.