1 This is cpufreq-bench, a microbenchmark for th
2
3 Purpose
4 =======
5
6 What is this benchmark for:
7 - Identify worst case performance loss when
8 scaling using Linux kernel governors
9 - Identify average reaction time of a govern
10 - (Stress) Testing whether a cpufreq low lev
11 as expected
12 - Identify cpufreq related performance regre
13 - Possibly Real time priority testing? -> wh
14 processes with a higher prio than the gove
15 - ...
16
17 What this benchmark does *not* cover:
18 - Power saving related regressions (In fact
19 throughput is, the worse the power savings
20 mostly count more...)
21 - Real world (workloads)
22
23
24 Description
25 ===========
26
27 cpufreq-bench helps to test the condition of a
28 For that purpose, it compares the performance
29 powersave module.
30
31
32 How it works
33 ============
34 You can specify load (100% CPU load) and sleep
35 will be run X time in a row (cycles):
36
37 sleep=25000
38 load=25000
39 cycles=20
40
41 This part of the configuration file will creat
42 repeated 20 times.
43
44 Adding this:
45 sleep_step=25000
46 load_step=25000
47 rounds=5
48 Will increase load and sleep time by 25ms 5 ti
49 Together you get following test:
50 25ms load/sleep time repeated 20 times (cycle
51 50ms load/sleep time repeated 20 times (cycle
52 ..
53 100ms load/sleep time repeated 20 times (cycle
54
55 First it is calibrated how long a specific CPU
56 takes on this machine and needs to be run in a
57 governor.
58 Then the above test runs are processed using t
59 and the governor to test. The time the calcula
60 with the dynamic freq scaling governor is comp
61 on full performance and you get the overall pe
62
63
64 Example of expected results with ondemand gove
65
66 This shows expected results of the first two t
67 above config, you there have:
68
69 100% CPU load (load) | 0 % CPU load (sleep) |
70 25 ms | 25 ms |
71 50 ms | 50 ms |
72
73 For example if ondemand governor is configured
74 sampling rate you get:
75
76 In round 1, ondemand should have rather static
77 won't ever switch up (as long as up_threshold
78
79 In round 2, if the ondemand sampling times exa
80 trigger of the cpufreq-bench, you will see no
81 below possible ondemand sample kick ins (1)):
82
83 But if ondemand always kicks in in the middle
84 will always see 50% loads and you get worst pe
85 switching up (compare with below possible onde
86
87 50 50 50 50ms ->time
88 load -----| |-----| |-----| |-----
89 | | | | | |
90 sleep |-----| |-----| |-----|
91 |-----|-----|-----|-----|-----|-----|-----
92 100 0 100 0 100 0 1
93 |-----|-----|-----|-----|-----|-----|--
94 50 50 50 50 50 50 50
95
96 You can easily test all kind of load/sleep tim
97 governor in average behaves as expected.
98
99
100 ToDo
101 ====
102
103 Provide a gnuplot utility script for easy gene
104 the outcome nicely.
105
106
107 cpufreq-bench Command Usage
108 ===========================
109 -l, --load=<long int> initial load t
110 -s, --sleep=<long int> initial sleep
111 -x, --load-step=<long int> time to be add
112 -y, --sleep-step=<long int> time to be add
113 -c, --cpu=<unsigned int> CPU Number to
114 -p, --prio=<priority> scheduler prio
115 -g, --governor=<governor> cpufreq govern
116 -n, --cycles=<int> load/sleep cyc
117 -r, --rounds<int> load/sleep rou
118 -f, --file=<configfile> config file to
119 -o, --output=<dir> output dir, mu
120 -v, --verbose verbose output
121
122 Due to the high priority, the application may
123 After the benchmark, the logfile is saved in O
124
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.