1LLVM-PROFDATA(1) LLVM LLVM-PROFDATA(1)
2
3
4
6 llvm-profdata - Profile data tool
7
9 llvm-profdata command [args...]
10
12 The llvm-profdata tool is a small utility for working with profile data
13 files.
14
16 · merge
17
18 · show
19
20 · overlap
21
23 SYNOPSIS
24 llvm-profdata merge [options] [filename...]
25
26 DESCRIPTION
27 llvm-profdata merge takes several profile data files generated by PGO
28 instrumentation and merges them together into a single indexed profile
29 data file.
30
31 By default profile data is merged without modification. This means that
32 the relative importance of each input file is proportional to the num‐
33 ber of samples or counts it contains. In general, the input from a
34 longer training run will be interpreted as relatively more important
35 than a shorter run. Depending on the nature of the training runs it may
36 be useful to adjust the weight given to each input file by using the
37 -weighted-input option.
38
39 Profiles passed in via -weighted-input, -input-files, or via positional
40 arguments are processed once for each time they are seen.
41
42 OPTIONS
43 -help Print a summary of command line options.
44
45 -output=output, -o=output
46 Specify the output file name. Output cannot be - as the result‐
47 ing indexed profile data can't be written to standard output.
48
49 -weighted-input=weight,filename
50 Specify an input file name along with a weight. The profile
51 counts of the supplied filename will be scaled (multiplied) by
52 the supplied weight, where where weight is a decimal integer >=
53 1. Input files specified without using this option are assigned
54 a default weight of 1. Examples are shown below.
55
56 -input-files=path, -f=path
57 Specify a file which contains a list of files to merge. The
58 entries in this file are newline-separated. Lines starting with
59 '#' are skipped. Entries may be of the form <filename> or
60 <weight>,<filename>.
61
62 -remapping-file=path, -r=path
63 Specify a file which contains a remapping from symbol names in
64 the input profile to the symbol names that should be used in the
65 output profile. The file should consist of lines of the form
66 <input-symbol> <output-symbol>. Blank lines and lines starting
67 with # are skipped.
68
69 The llvm-cxxmap tool can be used to generate the symbol remap‐
70 ping file.
71
72 -instr (default)
73 Specify that the input profile is an instrumentation-based pro‐
74 file.
75
76 -sample
77 Specify that the input profile is a sample-based profile.
78
79 The format of the generated file can be generated in one of
80 three ways:
81
82 -binary (default)
83
84 Emit the profile using a binary encoding. For instrumenta‐
85 tion-based profile the output format is the indexed binary for‐
86 mat.
87
88 -text
89
90 Emit the profile in text mode. This option can also be used with
91 both sample-based and instrumentation-based profile. When this
92 option is used the profile will be dumped in the text format
93 that is parsable by the profile reader.
94
95 -gcc
96
97 Emit the profile using GCC's gcov format (Not yet supported).
98
99 -sparse[=true|false]
100 Do not emit function records with 0 execution count. Can only be
101 used in conjunction with -instr. Defaults to false, since it can
102 inhibit compiler optimization during PGO.
103
104 -num-threads=N, -j=N
105 Use N threads to perform profile merging. When N=0, llvm-prof‐
106 data auto-detects an appropriate number of threads to use. This
107 is the default.
108
109 EXAMPLES
110 Basic Usage
111 Merge three profiles:
112
113 llvm-profdata merge foo.profdata bar.profdata baz.profdata -output merged.profdata
114
115 Weighted Input
116 The input file foo.profdata is especially important, multiply its
117 counts by 10:
118
119 llvm-profdata merge -weighted-input=10,foo.profdata bar.profdata baz.profdata -output merged.profdata
120
121 Exactly equivalent to the previous invocation (explicit form; useful
122 for programmatic invocation):
123
124 llvm-profdata merge -weighted-input=10,foo.profdata -weighted-input=1,bar.profdata -weighted-input=1,baz.profdata -output merged.profdata
125
127 SYNOPSIS
128 llvm-profdata show [options] [filename]
129
130 DESCRIPTION
131 llvm-profdata show takes a profile data file and displays the informa‐
132 tion about the profile counters for this file and for any of the speci‐
133 fied function(s).
134
135 If filename is omitted or is -, then llvm-profdata show reads its input
136 from standard input.
137
138 OPTIONS
139 -all-functions
140 Print details for every function.
141
142 -counts
143 Print the counter values for the displayed functions.
144
145 -function=string
146 Print details for a function if the function's name contains the
147 given string.
148
149 -help Print a summary of command line options.
150
151 -output=output, -o=output
152 Specify the output file name. If output is - or it isn't speci‐
153 fied, then the output is sent to standard output.
154
155 -instr (default)
156 Specify that the input profile is an instrumentation-based pro‐
157 file.
158
159 -text Instruct the profile dumper to show profile counts in the text
160 format of the instrumentation-based profile data representation.
161 By default, the profile information is dumped in a more human
162 readable form (also in text) with annotations.
163
164 -topn=n
165 Instruct the profile dumper to show the top n functions with the
166 hottest basic blocks in the summary section. By default, the
167 topn functions are not dumped.
168
169 -sample
170 Specify that the input profile is a sample-based profile.
171
172 -memop-sizes
173 Show the profiled sizes of the memory intrinsic calls for shown
174 functions.
175
176 -value-cutoff=n
177 Show only those functions whose max count values are greater or
178 equal to n. By default, the value-cutoff is set to 0.
179
180 -list-below-cutoff
181 Only output names of functions whose max count value are below
182 the cutoff value.
183
184 -showcs
185 Only show context sensitive profile counts. The default is to
186 filter all context sensitive profile counts.
187
189 SYNOPSIS
190 llvm-profdata overlap [options] [base profile file] [test profile file]
191
192 DESCRIPTION
193 llvm-profdata overlap takes two profile data files and displays the
194 overlap of counter distribution between the whole files and between any
195 of the specified functions.
196
197 In this command, overlap is defined as follows: Suppose base profile
198 file has the following counts: {c1_1, c1_2, ..., c1_n, c1_u_1, c2_u_2,
199 ..., c2_u_s}, and test profile file has {c2_1, c2_2, ..., c2_n, c2_v_1,
200 c2_v_2, ..., c2_v_t}. Here c{1|2}_i (i = 1 .. n) are matched counters
201 and c1_u_i (i = 1 .. s) and c2_v_i (i = 1 .. v) are unmatched counters
202 (or counters only existing in) base profile file and test profile file,
203 respectively. Let sum_1 = c1_1 + c1_2 + ... + c1_n + c1_u_1 + c2_u_2
204 + ... + c2_u_s, and sum_2 = c2_1 + c2_2 + ... + c2_n + c2_v_1 + c2_v_2
205 + ... + c2_v_t. overlap = min(c1_1/sum_1, c2_1/sum_2) +
206 min(c1_2/sum_1, c2_2/sum_2) + ... + min(c1_n/sum_1, c2_n/sum_2).
207
208 The result overlap distribution is a percentage number, ranging from
209 0.0% to 100.0%, where 0.0% means there is no overlap and 100.0% means a
210 perfect overlap.
211
212 Here is an example, if base profile file has counts of {400, 600}, and
213 test profile file has matched counts of {60000, 40000}. The overlap is
214 80%.
215
216 OPTIONS
217 -function=string
218 Print details for a function if the function's name contains the
219 given string.
220
221 -help Print a summary of command line options.
222
223 -o=output or -o output
224 Specify the output file name. If output is - or it isn't speci‐
225 fied, then the output is sent to standard output.
226
227 -value-cutoff=n
228 Show only those functions whose max count values are greater or
229 equal to n. By default, the value-cutoff is set to max of
230 unsigned long long.
231
232 -cs Only show overlap for the context sensitive profile counts. The
233 default is to show non-context sensitive profile counts.
234
236 llvm-profdata returns 1 if the command is omitted or is invalid, if it
237 cannot read input files, or if there is a mismatch between their data.
238
240 Maintained by the LLVM Team (https://llvm.org/).
241
243 2003-2020, LLVM Project
244
245
246
247
2489 2020-08-12 LLVM-PROFDATA(1)