![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to timing CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [CSRG BSD Unix] / 43BSDTahoe / new / help / src / f77|
Return to timing CVS log | Up to [CSRG BSD Unix] / 43BSDTahoe / new / help / src / f77 |
1.1 root 1: .TI F77/TIMING "Sep. 15, 1984"
2: Timing Programs, Profiles
3:
4: UNIX divides the processor time charged to a job into user time and
5: system time. User time is the time the processor spends computing
6: for the process; system time is the time the processor spends
7: executing system calls (I/O requests, forks, execs, etc.)
8: on your behalf.
9:
10: Both kinds of time can vary according to the system load
11: because processor time spent handling interrupts is charged
12: to the current user even if the activity was caused by some
13: other user or some system function.
14: In practise the user time varies much less than the system time.
15: The times are kept in 60th of a second units.
16: You are charged for both user time and system time.
17:
18: Program segments may be timed using the dtime() library subroutine
19: (see "man 3f dtime"):
20:
21: .nf
22: dimension tarray(2)
23: ...
24: call dtime(tarray)
25: ... program segment to be timed ...
26: call dtime(tarray)
27: print 8000, tarray
28: 8000 format(" user time = ",f8.3," system time = ",f8.3)
29: .fi
30:
31: Dtime() in this example will put the user time used by the
32: program segment in tarray(1) and the system time used in tarray(2).
33:
34: It is best to exclude I/O operations from the segment being timed.
35: Since the times are kept in units of a 60th of a second and vary by
36: system load, do not try to time any computation that takes less than 10
37: or 20 seconds. If the computation is fast, put it in a loop and
38: time 10 or 100 repetitions.
39:
40: It is also possible to profile an entire program and find out
41: how often each subprogram was called and how much time it used.
42: First compile the program with the '-p' flag, run it and then
43: execute the prof utility to print out the profile:
44:
45: .nf
46: % f77 -p prog.f
47: prog.f:
48: MAIN:
49: sub1:
50: sub2:
51: sub3:
52: Loading a.out ...
53: %
54: % a.out
55: ... program output ...
56: % prof
57:
58: %time cumsecs #call ms/call name
59: 41.1 3.37 2 1683.48 _sub1_
60: 34.2 6.17 1 2800.24 _sub2_
61: 12.2 7.17 1 1000.09 _MAIN__
62: 11.8 8.13 2 483.37 _sub3_
63: 0.4 8.17 _fstat
64: 0.2 8.18 _x_putc
65: .fi
66:
67: In this example, sub1 was called twice and used 3.37 seconds of processor
68: time (41% of the total), sub2 was called once and used
69: 6.17 - 3.37 = 2.80 seconds (34%), etc. Prof keeps track of the sum
70: of user and system times. The last two lines of the output are for
71: library routines which were not compiled with the profile option.
72:
73: For more extensive profiling statistics, use the '-pg' option and the
74: gprof command. See "man f77" and "man gprof" for details.
75:
76: It is also possible to time entire commands. You can do it directly
77: by invoking the C Shell's builtin time command:
78:
79: .nf
80: % time a.out
81: 6.5u 4.6s 0:23 48% 12+47k 690+688io 17pf+0w
82: .fi
83:
84: This is partially documented in "man csh" and more fully in "An
85: Introduction to the C shell", section 2.8. It tells us the command
86: used 6.5 seconds of user time, 4.6 seconds of system time and took 23
87: seconds of real (clock) time to finish. During those 23 seconds, it
88: used 48% of the cpu cycles( 48% = (6.5+4.6)/23. ). On the average, it
89: used 12k bytes of program space and 47k bytes of data space. The
90: program did 690 disk reads and 688 disk writes, had 17 page faults and
91: did not have any pages swapped out.
92:
93: A second method for timing a program is to use /bin/time:
94:
95: .nf
96: % /bin/time a.out
97: 46.0 real 5.5 user 3.8 sys
98: .fi
99:
100: It is similar to the C shell time, but gives much less information.
101: Note that both commands are named time. If you type "man time",
102: you will be told about /bin/time rather than the command built
103: into the C shell.
104:
105: A third way is to set the C shell variable "time" to a minimum.
106: Thereafter, if a command takes more than that amount of cpu time
107: (system plus user time),
108: the statistics will automatically be listed using the same format as
109: the C shell time command. This is reversed by the unset time command.
110:
111: .nf
112: % set time = 4
113: % a.out
114: 5.7u 4.3s 0:59 16% 12+47k 703+654io 4pf+0w
115: %
116: % unset time
117: % a.out
118: %
This archive runs on limited infrastructure. Preserving old code on modern bandwidth. Automated agents are requested to crawl responsibly.