Lines Matching full:of

5 .\"	The Regents of the University of California.  All rights reserved.
10 .\" 1. Redistributions of source code must retain the above copyright
11 .\"    notice, this list of conditions and the following disclaimer.
13 .\"    notice, this list of conditions and the following disclaimer in the
15 .\" 3. Neither the name of the University nor the names of its contributors
21 .\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24 .\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25 .\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26 .\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28 .\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
42 A call of the monitoring routine has an overhead comparable with a
43 call of a regular routine, and is therefore only suited
59 Use of the monitoring routine requires no planning on part of a
63 We are interested in gathering three pieces of information during
65 each profiled routine, and the arcs of the dynamic call graph
66 traversed by this execution of the program.
67 By post-processing of this data we can build the dynamic call
68 graph for this execution of the program and propagate times along
69 the edges of this graph to attribute times for routines to the
72 Gathering of the profiling information should not greatly
73 interfere with the running of the program.
76 The volume of data thus produced would be unmanageably large,
78 time of most programs.
79 Similarly, the monitoring routine can not do the analysis of
87 listing of the profile data.
88 An advantage of this approach is that the profile data for
89 several executions of a program can be combined by the
90 post-processing to provide a profile of many
93 The execution time monitoring consists of three parts.
97 prologue of each profiled routine.
103 The \fBgprof\fP monitoring routine counts the number of times
113 This address is in the prologue of some profiled routine that is
114 the destination of an arc in the dynamic call graph.
116 routine, thus identifying the call site, or source of the arc.
117 The source of the arc is in the \fIcaller\fP, and the destination is in
121 The prologue of B will include a call to the monitoring routine
127 Therefore, the monitoring routine maintains a table of all the
128 arcs discovered, with counts of the numbers of times each is
139 reduce (usually to one) the number of minor lookups based on the callee.
142 Such an organization has the advantage of associating callers with
143 callees, at the expense of longer lookups in the monitoring
146 and (for the sake of speed) were able to allocate enough space
154 Further, the number of dynamic call sites and callees is not known during
155 execution of the profiled program.
160 will not call the monitoring routine as part of their prologue,
167 Thus the monitoring routine may know the destination of an arc
170 impossible to determine the source of the arc (the caller).
171 Often in these cases the apparent source of the arc is not a call
178 One method measures the execution time of a routine by measuring
181 systems by the time-slicing of the program.
182 A second method samples the value of the program counter at some
183 interval, and infers execution time from the distribution of the
195 sampling of the program counter rather than the duration of the
196 interval is the basis of the distribution.
198 program counter will overwhelm the running of the profiled program.
200 intervals that the distribution of the samples accurately
201 represents the distribution of time for the execution of the
207 histogram of the location of the program counter at the end of
208 each clock tick (1/60th of a second) in which a program runs.
211 We have adjusted the granularity of the histogram so that
214 routine require the same amount of time to execute.
221 the arc table and the histogram of
223 The arc table is condensed to consist of the source and destination
224 addresses of the arc and the count of the number of times the arc
225 was traversed by this execution of the program.
226 The recorded histogram consists of counters of the number of
227 times the program counter was found to be in each of the ranges covered