1 .. SPDX-License-Identifier: GPL-2.0 !! 1 .. |struct cpufreq_policy| replace:: :c:type:`struct cpufreq_policy <cpufreq_policy>`
2 .. include:: <isonum.txt> <<
3 <<
4 .. |intel_pstate| replace:: :doc:`intel_pstate 2 .. |intel_pstate| replace:: :doc:`intel_pstate <intel_pstate>`
5 3
6 ======================= 4 =======================
7 CPU Performance Scaling 5 CPU Performance Scaling
8 ======================= 6 =======================
9 7
10 :Copyright: |copy| 2017 Intel Corporation !! 8 ::
11 <<
12 :Author: Rafael J. Wysocki <rafael.j.wysocki@in <<
13 9
>> 10 Copyright (c) 2017 Intel Corp., Rafael J. Wysocki <rafael.j.wysocki@intel.com>
14 11
15 The Concept of CPU Performance Scaling 12 The Concept of CPU Performance Scaling
16 ====================================== 13 ======================================
17 14
18 The majority of modern processors are capable 15 The majority of modern processors are capable of operating in a number of
19 different clock frequency and voltage configur 16 different clock frequency and voltage configurations, often referred to as
20 Operating Performance Points or P-states (in A 17 Operating Performance Points or P-states (in ACPI terminology). As a rule,
21 the higher the clock frequency and the higher 18 the higher the clock frequency and the higher the voltage, the more instructions
22 can be retired by the CPU over a unit of time, 19 can be retired by the CPU over a unit of time, but also the higher the clock
23 frequency and the higher the voltage, the more 20 frequency and the higher the voltage, the more energy is consumed over a unit of
24 time (or the more power is drawn) by the CPU i 21 time (or the more power is drawn) by the CPU in the given P-state. Therefore
25 there is a natural tradeoff between the CPU ca 22 there is a natural tradeoff between the CPU capacity (the number of instructions
26 that can be executed over a unit of time) and 23 that can be executed over a unit of time) and the power drawn by the CPU.
27 24
28 In some situations it is desirable or even nec 25 In some situations it is desirable or even necessary to run the program as fast
29 as possible and then there is no reason to use 26 as possible and then there is no reason to use any P-states different from the
30 highest one (i.e. the highest-performance freq 27 highest one (i.e. the highest-performance frequency/voltage configuration
31 available). In some other cases, however, it 28 available). In some other cases, however, it may not be necessary to execute
32 instructions so quickly and maintaining the hi 29 instructions so quickly and maintaining the highest available CPU capacity for a
33 relatively long time without utilizing it enti 30 relatively long time without utilizing it entirely may be regarded as wasteful.
34 It also may not be physically possible to main 31 It also may not be physically possible to maintain maximum CPU capacity for too
35 long for thermal or power supply capacity reas 32 long for thermal or power supply capacity reasons or similar. To cover those
36 cases, there are hardware interfaces allowing 33 cases, there are hardware interfaces allowing CPUs to be switched between
37 different frequency/voltage configurations or 34 different frequency/voltage configurations or (in the ACPI terminology) to be
38 put into different P-states. 35 put into different P-states.
39 36
40 Typically, they are used along with algorithms 37 Typically, they are used along with algorithms to estimate the required CPU
41 capacity, so as to decide which P-states to pu 38 capacity, so as to decide which P-states to put the CPUs into. Of course, since
42 the utilization of the system generally change 39 the utilization of the system generally changes over time, that has to be done
43 repeatedly on a regular basis. The activity b 40 repeatedly on a regular basis. The activity by which this happens is referred
44 to as CPU performance scaling or CPU frequency 41 to as CPU performance scaling or CPU frequency scaling (because it involves
45 adjusting the CPU clock frequency). 42 adjusting the CPU clock frequency).
46 43
47 44
48 CPU Performance Scaling in Linux 45 CPU Performance Scaling in Linux
49 ================================ 46 ================================
50 47
51 The Linux kernel supports CPU performance scal 48 The Linux kernel supports CPU performance scaling by means of the ``CPUFreq``
52 (CPU Frequency scaling) subsystem that consist 49 (CPU Frequency scaling) subsystem that consists of three layers of code: the
53 core, scaling governors and scaling drivers. 50 core, scaling governors and scaling drivers.
54 51
55 The ``CPUFreq`` core provides the common code 52 The ``CPUFreq`` core provides the common code infrastructure and user space
56 interfaces for all platforms that support CPU 53 interfaces for all platforms that support CPU performance scaling. It defines
57 the basic framework in which the other compone 54 the basic framework in which the other components operate.
58 55
59 Scaling governors implement algorithms to esti 56 Scaling governors implement algorithms to estimate the required CPU capacity.
60 As a rule, each governor implements one, possi 57 As a rule, each governor implements one, possibly parametrized, scaling
61 algorithm. 58 algorithm.
62 59
63 Scaling drivers talk to the hardware. They pr 60 Scaling drivers talk to the hardware. They provide scaling governors with
64 information on the available P-states (or P-st 61 information on the available P-states (or P-state ranges in some cases) and
65 access platform-specific hardware interfaces t 62 access platform-specific hardware interfaces to change CPU P-states as requested
66 by scaling governors. 63 by scaling governors.
67 64
68 In principle, all available scaling governors 65 In principle, all available scaling governors can be used with every scaling
69 driver. That design is based on the observati 66 driver. That design is based on the observation that the information used by
70 performance scaling algorithms for P-state sel 67 performance scaling algorithms for P-state selection can be represented in a
71 platform-independent form in the majority of c 68 platform-independent form in the majority of cases, so it should be possible
72 to use the same performance scaling algorithm 69 to use the same performance scaling algorithm implemented in exactly the same
73 way regardless of which scaling driver is used 70 way regardless of which scaling driver is used. Consequently, the same set of
74 scaling governors should be suitable for every 71 scaling governors should be suitable for every supported platform.
75 72
76 However, that observation may not hold for per 73 However, that observation may not hold for performance scaling algorithms
77 based on information provided by the hardware 74 based on information provided by the hardware itself, for example through
78 feedback registers, as that information is typ 75 feedback registers, as that information is typically specific to the hardware
79 interface it comes from and may not be easily 76 interface it comes from and may not be easily represented in an abstract,
80 platform-independent way. For this reason, `` 77 platform-independent way. For this reason, ``CPUFreq`` allows scaling drivers
81 to bypass the governor layer and implement the 78 to bypass the governor layer and implement their own performance scaling
82 algorithms. That is done by the |intel_pstate 79 algorithms. That is done by the |intel_pstate| scaling driver.
83 80
84 81
85 ``CPUFreq`` Policy Objects 82 ``CPUFreq`` Policy Objects
86 ========================== 83 ==========================
87 84
88 In some cases the hardware interface for P-sta 85 In some cases the hardware interface for P-state control is shared by multiple
89 CPUs. That is, for example, the same register 86 CPUs. That is, for example, the same register (or set of registers) is used to
90 control the P-state of multiple CPUs at the sa 87 control the P-state of multiple CPUs at the same time and writing to it affects
91 all of those CPUs simultaneously. 88 all of those CPUs simultaneously.
92 89
93 Sets of CPUs sharing hardware P-state control 90 Sets of CPUs sharing hardware P-state control interfaces are represented by
94 ``CPUFreq`` as struct cpufreq_policy objects. !! 91 ``CPUFreq`` as |struct cpufreq_policy| objects. For consistency,
95 struct cpufreq_policy is also used when there !! 92 |struct cpufreq_policy| is also used when there is only one CPU in the given
96 set. 93 set.
97 94
98 The ``CPUFreq`` core maintains a pointer to a !! 95 The ``CPUFreq`` core maintains a pointer to a |struct cpufreq_policy| object for
99 every CPU in the system, including CPUs that a 96 every CPU in the system, including CPUs that are currently offline. If multiple
100 CPUs share the same hardware P-state control i 97 CPUs share the same hardware P-state control interface, all of the pointers
101 corresponding to them point to the same struct !! 98 corresponding to them point to the same |struct cpufreq_policy| object.
102 99
103 ``CPUFreq`` uses struct cpufreq_policy as its !! 100 ``CPUFreq`` uses |struct cpufreq_policy| as its basic data type and the design
104 of its user space interface is based on the po 101 of its user space interface is based on the policy concept.
105 102
106 103
107 CPU Initialization 104 CPU Initialization
108 ================== 105 ==================
109 106
110 First of all, a scaling driver has to be regis 107 First of all, a scaling driver has to be registered for ``CPUFreq`` to work.
111 It is only possible to register one scaling dr 108 It is only possible to register one scaling driver at a time, so the scaling
112 driver is expected to be able to handle all CP 109 driver is expected to be able to handle all CPUs in the system.
113 110
114 The scaling driver may be registered before or 111 The scaling driver may be registered before or after CPU registration. If
115 CPUs are registered earlier, the driver core i 112 CPUs are registered earlier, the driver core invokes the ``CPUFreq`` core to
116 take a note of all of the already registered C 113 take a note of all of the already registered CPUs during the registration of the
117 scaling driver. In turn, if any CPUs are regi 114 scaling driver. In turn, if any CPUs are registered after the registration of
118 the scaling driver, the ``CPUFreq`` core will 115 the scaling driver, the ``CPUFreq`` core will be invoked to take note of them
119 at their registration time. 116 at their registration time.
120 117
121 In any case, the ``CPUFreq`` core is invoked t 118 In any case, the ``CPUFreq`` core is invoked to take note of any logical CPU it
122 has not seen so far as soon as it is ready to 119 has not seen so far as soon as it is ready to handle that CPU. [Note that the
123 logical CPU may be a physical single-core proc 120 logical CPU may be a physical single-core processor, or a single core in a
124 multicore processor, or a hardware thread in a 121 multicore processor, or a hardware thread in a physical processor or processor
125 core. In what follows "CPU" always means "log 122 core. In what follows "CPU" always means "logical CPU" unless explicitly stated
126 otherwise and the word "processor" is used to 123 otherwise and the word "processor" is used to refer to the physical part
127 possibly including multiple logical CPUs.] 124 possibly including multiple logical CPUs.]
128 125
129 Once invoked, the ``CPUFreq`` core checks if t 126 Once invoked, the ``CPUFreq`` core checks if the policy pointer is already set
130 for the given CPU and if so, it skips the poli 127 for the given CPU and if so, it skips the policy object creation. Otherwise,
131 a new policy object is created and initialized 128 a new policy object is created and initialized, which involves the creation of
132 a new policy directory in ``sysfs``, and the p 129 a new policy directory in ``sysfs``, and the policy pointer corresponding to
133 the given CPU is set to the new policy object' 130 the given CPU is set to the new policy object's address in memory.
134 131
135 Next, the scaling driver's ``->init()`` callba 132 Next, the scaling driver's ``->init()`` callback is invoked with the policy
136 pointer of the new CPU passed to it as the arg 133 pointer of the new CPU passed to it as the argument. That callback is expected
137 to initialize the performance scaling hardware 134 to initialize the performance scaling hardware interface for the given CPU (or,
138 more precisely, for the set of CPUs sharing th 135 more precisely, for the set of CPUs sharing the hardware interface it belongs
139 to, represented by its policy object) and, if 136 to, represented by its policy object) and, if the policy object it has been
140 called for is new, to set parameters of the po 137 called for is new, to set parameters of the policy, like the minimum and maximum
141 frequencies supported by the hardware, the tab 138 frequencies supported by the hardware, the table of available frequencies (if
142 the set of supported P-states is not a continu 139 the set of supported P-states is not a continuous range), and the mask of CPUs
143 that belong to the same policy (including both 140 that belong to the same policy (including both online and offline CPUs). That
144 mask is then used by the core to populate the 141 mask is then used by the core to populate the policy pointers for all of the
145 CPUs in it. 142 CPUs in it.
146 143
147 The next major initialization step for a new p 144 The next major initialization step for a new policy object is to attach a
148 scaling governor to it (to begin with, that is 145 scaling governor to it (to begin with, that is the default scaling governor
149 determined by the kernel command line or confi !! 146 determined by the kernel configuration, but it may be changed later
150 later via ``sysfs``). First, a pointer to the !! 147 via ``sysfs``). First, a pointer to the new policy object is passed to the
151 the governor's ``->init()`` callback which is !! 148 governor's ``->init()`` callback which is expected to initialize all of the
152 data structures necessary to handle the given 149 data structures necessary to handle the given policy and, possibly, to add
153 a governor ``sysfs`` interface to it. Next, t 150 a governor ``sysfs`` interface to it. Next, the governor is started by
154 invoking its ``->start()`` callback. 151 invoking its ``->start()`` callback.
155 152
156 That callback is expected to register per-CPU !! 153 That callback it expected to register per-CPU utilization update callbacks for
157 all of the online CPUs belonging to the given 154 all of the online CPUs belonging to the given policy with the CPU scheduler.
158 The utilization update callbacks will be invok 155 The utilization update callbacks will be invoked by the CPU scheduler on
159 important events, like task enqueue and dequeu 156 important events, like task enqueue and dequeue, on every iteration of the
160 scheduler tick or generally whenever the CPU u 157 scheduler tick or generally whenever the CPU utilization may change (from the
161 scheduler's perspective). They are expected t 158 scheduler's perspective). They are expected to carry out computations needed
162 to determine the P-state to use for the given 159 to determine the P-state to use for the given policy going forward and to
163 invoke the scaling driver to make changes to t 160 invoke the scaling driver to make changes to the hardware in accordance with
164 the P-state selection. The scaling driver may 161 the P-state selection. The scaling driver may be invoked directly from
165 scheduler context or asynchronously, via a ker 162 scheduler context or asynchronously, via a kernel thread or workqueue, depending
166 on the configuration and capabilities of the s 163 on the configuration and capabilities of the scaling driver and the governor.
167 164
168 Similar steps are taken for policy objects tha 165 Similar steps are taken for policy objects that are not new, but were "inactive"
169 previously, meaning that all of the CPUs belon 166 previously, meaning that all of the CPUs belonging to them were offline. The
170 only practical difference in that case is that 167 only practical difference in that case is that the ``CPUFreq`` core will attempt
171 to use the scaling governor previously used wi 168 to use the scaling governor previously used with the policy that became
172 "inactive" (and is re-initialized now) instead 169 "inactive" (and is re-initialized now) instead of the default governor.
173 170
174 In turn, if a previously offline CPU is being 171 In turn, if a previously offline CPU is being brought back online, but some
175 other CPUs sharing the policy object with it a 172 other CPUs sharing the policy object with it are online already, there is no
176 need to re-initialize the policy object at all 173 need to re-initialize the policy object at all. In that case, it only is
177 necessary to restart the scaling governor so t 174 necessary to restart the scaling governor so that it can take the new online CPU
178 into account. That is achieved by invoking th 175 into account. That is achieved by invoking the governor's ``->stop`` and
179 ``->start()`` callbacks, in this order, for th 176 ``->start()`` callbacks, in this order, for the entire policy.
180 177
181 As mentioned before, the |intel_pstate| scalin 178 As mentioned before, the |intel_pstate| scaling driver bypasses the scaling
182 governor layer of ``CPUFreq`` and provides its 179 governor layer of ``CPUFreq`` and provides its own P-state selection algorithms.
183 Consequently, if |intel_pstate| is used, scali 180 Consequently, if |intel_pstate| is used, scaling governors are not attached to
184 new policy objects. Instead, the driver's ``- 181 new policy objects. Instead, the driver's ``->setpolicy()`` callback is invoked
185 to register per-CPU utilization update callbac 182 to register per-CPU utilization update callbacks for each policy. These
186 callbacks are invoked by the CPU scheduler in 183 callbacks are invoked by the CPU scheduler in the same way as for scaling
187 governors, but in the |intel_pstate| case they 184 governors, but in the |intel_pstate| case they both determine the P-state to
188 use and change the hardware configuration acco 185 use and change the hardware configuration accordingly in one go from scheduler
189 context. 186 context.
190 187
191 The policy objects created during CPU initiali 188 The policy objects created during CPU initialization and other data structures
192 associated with them are torn down when the sc 189 associated with them are torn down when the scaling driver is unregistered
193 (which happens when the kernel module containi 190 (which happens when the kernel module containing it is unloaded, for example) or
194 when the last CPU belonging to the given polic 191 when the last CPU belonging to the given policy in unregistered.
195 192
196 193
197 Policy Interface in ``sysfs`` 194 Policy Interface in ``sysfs``
198 ============================= 195 =============================
199 196
200 During the initialization of the kernel, the ` 197 During the initialization of the kernel, the ``CPUFreq`` core creates a
201 ``sysfs`` directory (kobject) called ``cpufreq 198 ``sysfs`` directory (kobject) called ``cpufreq`` under
202 :file:`/sys/devices/system/cpu/`. 199 :file:`/sys/devices/system/cpu/`.
203 200
204 That directory contains a ``policyX`` subdirec 201 That directory contains a ``policyX`` subdirectory (where ``X`` represents an
205 integer number) for every policy object mainta 202 integer number) for every policy object maintained by the ``CPUFreq`` core.
206 Each ``policyX`` directory is pointed to by `` 203 Each ``policyX`` directory is pointed to by ``cpufreq`` symbolic links
207 under :file:`/sys/devices/system/cpu/cpuY/` (w 204 under :file:`/sys/devices/system/cpu/cpuY/` (where ``Y`` represents an integer
208 that may be different from the one represented 205 that may be different from the one represented by ``X``) for all of the CPUs
209 associated with (or belonging to) the given po 206 associated with (or belonging to) the given policy. The ``policyX`` directories
210 in :file:`/sys/devices/system/cpu/cpufreq` eac 207 in :file:`/sys/devices/system/cpu/cpufreq` each contain policy-specific
211 attributes (files) to control ``CPUFreq`` beha 208 attributes (files) to control ``CPUFreq`` behavior for the corresponding policy
212 objects (that is, for all of the CPUs associat 209 objects (that is, for all of the CPUs associated with them).
213 210
214 Some of those attributes are generic. They ar 211 Some of those attributes are generic. They are created by the ``CPUFreq`` core
215 and their behavior generally does not depend o 212 and their behavior generally does not depend on what scaling driver is in use
216 and what scaling governor is attached to the g 213 and what scaling governor is attached to the given policy. Some scaling drivers
217 also add driver-specific attributes to the pol 214 also add driver-specific attributes to the policy directories in ``sysfs`` to
218 control policy-specific aspects of driver beha 215 control policy-specific aspects of driver behavior.
219 216
220 The generic attributes under :file:`/sys/devic 217 The generic attributes under :file:`/sys/devices/system/cpu/cpufreq/policyX/`
221 are the following: 218 are the following:
222 219
223 ``affected_cpus`` 220 ``affected_cpus``
224 List of online CPUs belonging to this 221 List of online CPUs belonging to this policy (i.e. sharing the hardware
225 performance scaling interface represen 222 performance scaling interface represented by the ``policyX`` policy
226 object). 223 object).
227 224
228 ``bios_limit`` 225 ``bios_limit``
229 If the platform firmware (BIOS) tells 226 If the platform firmware (BIOS) tells the OS to apply an upper limit to
230 CPU frequencies, that limit will be re 227 CPU frequencies, that limit will be reported through this attribute (if
231 present). 228 present).
232 229
233 The existence of the limit may be a re 230 The existence of the limit may be a result of some (often unintentional)
234 BIOS settings, restrictions coming fro 231 BIOS settings, restrictions coming from a service processor or another
235 BIOS/HW-based mechanisms. 232 BIOS/HW-based mechanisms.
236 233
237 This does not cover ACPI thermal limit 234 This does not cover ACPI thermal limitations which can be discovered
238 through a generic thermal driver. 235 through a generic thermal driver.
239 236
240 This attribute is not present if the s 237 This attribute is not present if the scaling driver in use does not
241 support it. 238 support it.
242 239
243 ``cpuinfo_cur_freq`` <<
244 Current frequency of the CPUs belongin <<
245 the hardware (in KHz). <<
246 <<
247 This is expected to be the frequency t <<
248 If that frequency cannot be determined <<
249 be present. <<
250 <<
251 ``cpuinfo_max_freq`` 240 ``cpuinfo_max_freq``
252 Maximum possible operating frequency t 241 Maximum possible operating frequency the CPUs belonging to this policy
253 can run at (in kHz). 242 can run at (in kHz).
254 243
255 ``cpuinfo_min_freq`` 244 ``cpuinfo_min_freq``
256 Minimum possible operating frequency t 245 Minimum possible operating frequency the CPUs belonging to this policy
257 can run at (in kHz). 246 can run at (in kHz).
258 247
259 ``cpuinfo_transition_latency`` 248 ``cpuinfo_transition_latency``
260 The time it takes to switch the CPUs b 249 The time it takes to switch the CPUs belonging to this policy from one
261 P-state to another, in nanoseconds. 250 P-state to another, in nanoseconds.
262 251
263 If unknown or if known to be so high t 252 If unknown or if known to be so high that the scaling driver does not
264 work with the `ondemand`_ governor, -1 253 work with the `ondemand`_ governor, -1 (:c:macro:`CPUFREQ_ETERNAL`)
265 will be returned by reads from this at 254 will be returned by reads from this attribute.
266 255
267 ``related_cpus`` 256 ``related_cpus``
268 List of all (online and offline) CPUs 257 List of all (online and offline) CPUs belonging to this policy.
269 258
270 ``scaling_available_frequencies`` <<
271 List of available frequencies of the C <<
272 (in kHz). <<
273 <<
274 ``scaling_available_governors`` 259 ``scaling_available_governors``
275 List of ``CPUFreq`` scaling governors 260 List of ``CPUFreq`` scaling governors present in the kernel that can
276 be attached to this policy or (if the 261 be attached to this policy or (if the |intel_pstate| scaling driver is
277 in use) list of scaling algorithms pro 262 in use) list of scaling algorithms provided by the driver that can be
278 applied to this policy. 263 applied to this policy.
279 264
280 [Note that some governors are modular 265 [Note that some governors are modular and it may be necessary to load a
281 kernel module for the governor held by 266 kernel module for the governor held by it to become available and be
282 listed by this attribute.] 267 listed by this attribute.]
283 268
284 ``scaling_cur_freq`` 269 ``scaling_cur_freq``
285 Current frequency of all of the CPUs b 270 Current frequency of all of the CPUs belonging to this policy (in kHz).
286 271
287 In the majority of cases, this is the !! 272 For the majority of scaling drivers, this is the frequency of the last
288 requested by the scaling driver from t !! 273 P-state requested by the driver from the hardware using the scaling
289 interface provided by it, which may or 274 interface provided by it, which may or may not reflect the frequency
290 the CPU is actually running at (due to 275 the CPU is actually running at (due to hardware design and other
291 limitations). 276 limitations).
292 277
293 Some architectures (e.g. ``x86``) may !! 278 Some scaling drivers (e.g. |intel_pstate|) attempt to provide
294 more precisely reflecting the current !! 279 information more precisely reflecting the current CPU frequency through
295 attribute, but that still may not be t !! 280 this attribute, but that still may not be the exact current CPU
296 seen by the hardware at the moment. !! 281 frequency as seen by the hardware at the moment.
297 282
298 ``scaling_driver`` 283 ``scaling_driver``
299 The scaling driver currently in use. 284 The scaling driver currently in use.
300 285
301 ``scaling_governor`` 286 ``scaling_governor``
302 The scaling governor currently attache 287 The scaling governor currently attached to this policy or (if the
303 |intel_pstate| scaling driver is in us 288 |intel_pstate| scaling driver is in use) the scaling algorithm
304 provided by the driver that is current 289 provided by the driver that is currently applied to this policy.
305 290
306 This attribute is read-write and writi 291 This attribute is read-write and writing to it will cause a new scaling
307 governor to be attached to this policy 292 governor to be attached to this policy or a new scaling algorithm
308 provided by the scaling driver to be a 293 provided by the scaling driver to be applied to it (in the
309 |intel_pstate| case), as indicated by 294 |intel_pstate| case), as indicated by the string written to this
310 attribute (which must be one of the na 295 attribute (which must be one of the names listed by the
311 ``scaling_available_governors`` attrib 296 ``scaling_available_governors`` attribute described above).
312 297
313 ``scaling_max_freq`` 298 ``scaling_max_freq``
314 Maximum frequency the CPUs belonging t 299 Maximum frequency the CPUs belonging to this policy are allowed to be
315 running at (in kHz). 300 running at (in kHz).
316 301
317 This attribute is read-write and writi 302 This attribute is read-write and writing a string representing an
318 integer to it will cause a new limit t 303 integer to it will cause a new limit to be set (it must not be lower
319 than the value of the ``scaling_min_fr 304 than the value of the ``scaling_min_freq`` attribute).
320 305
321 ``scaling_min_freq`` 306 ``scaling_min_freq``
322 Minimum frequency the CPUs belonging t 307 Minimum frequency the CPUs belonging to this policy are allowed to be
323 running at (in kHz). 308 running at (in kHz).
324 309
325 This attribute is read-write and writi 310 This attribute is read-write and writing a string representing a
326 non-negative integer to it will cause 311 non-negative integer to it will cause a new limit to be set (it must not
327 be higher than the value of the ``scal 312 be higher than the value of the ``scaling_max_freq`` attribute).
328 313
329 ``scaling_setspeed`` 314 ``scaling_setspeed``
330 This attribute is functional only if t 315 This attribute is functional only if the `userspace`_ scaling governor
331 is attached to the given policy. 316 is attached to the given policy.
332 317
333 It returns the last frequency requeste 318 It returns the last frequency requested by the governor (in kHz) or can
334 be written to in order to set a new fr 319 be written to in order to set a new frequency for the policy.
335 320
336 321
337 Generic Scaling Governors 322 Generic Scaling Governors
338 ========================= 323 =========================
339 324
340 ``CPUFreq`` provides generic scaling governors 325 ``CPUFreq`` provides generic scaling governors that can be used with all
341 scaling drivers. As stated before, each of th 326 scaling drivers. As stated before, each of them implements a single, possibly
342 parametrized, performance scaling algorithm. 327 parametrized, performance scaling algorithm.
343 328
344 Scaling governors are attached to policy objec 329 Scaling governors are attached to policy objects and different policy objects
345 can be handled by different scaling governors 330 can be handled by different scaling governors at the same time (although that
346 may lead to suboptimal results in some cases). 331 may lead to suboptimal results in some cases).
347 332
348 The scaling governor for a given policy object 333 The scaling governor for a given policy object can be changed at any time with
349 the help of the ``scaling_governor`` policy at 334 the help of the ``scaling_governor`` policy attribute in ``sysfs``.
350 335
351 Some governors expose ``sysfs`` attributes to 336 Some governors expose ``sysfs`` attributes to control or fine-tune the scaling
352 algorithms implemented by them. Those attribu 337 algorithms implemented by them. Those attributes, referred to as governor
353 tunables, can be either global (system-wide) o 338 tunables, can be either global (system-wide) or per-policy, depending on the
354 scaling driver in use. If the driver requires 339 scaling driver in use. If the driver requires governor tunables to be
355 per-policy, they are located in a subdirectory 340 per-policy, they are located in a subdirectory of each policy directory.
356 Otherwise, they are located in a subdirectory 341 Otherwise, they are located in a subdirectory under
357 :file:`/sys/devices/system/cpu/cpufreq/`. In 342 :file:`/sys/devices/system/cpu/cpufreq/`. In either case the name of the
358 subdirectory containing the governor tunables 343 subdirectory containing the governor tunables is the name of the governor
359 providing them. 344 providing them.
360 345
361 ``performance`` 346 ``performance``
362 --------------- 347 ---------------
363 348
364 When attached to a policy object, this governo 349 When attached to a policy object, this governor causes the highest frequency,
365 within the ``scaling_max_freq`` policy limit, 350 within the ``scaling_max_freq`` policy limit, to be requested for that policy.
366 351
367 The request is made once at that time the gove 352 The request is made once at that time the governor for the policy is set to
368 ``performance`` and whenever the ``scaling_max 353 ``performance`` and whenever the ``scaling_max_freq`` or ``scaling_min_freq``
369 policy limits change after that. 354 policy limits change after that.
370 355
371 ``powersave`` 356 ``powersave``
372 ------------- 357 -------------
373 358
374 When attached to a policy object, this governo 359 When attached to a policy object, this governor causes the lowest frequency,
375 within the ``scaling_min_freq`` policy limit, 360 within the ``scaling_min_freq`` policy limit, to be requested for that policy.
376 361
377 The request is made once at that time the gove 362 The request is made once at that time the governor for the policy is set to
378 ``powersave`` and whenever the ``scaling_max_f 363 ``powersave`` and whenever the ``scaling_max_freq`` or ``scaling_min_freq``
379 policy limits change after that. 364 policy limits change after that.
380 365
381 ``userspace`` 366 ``userspace``
382 ------------- 367 -------------
383 368
384 This governor does not do anything by itself. 369 This governor does not do anything by itself. Instead, it allows user space
385 to set the CPU frequency for the policy it is 370 to set the CPU frequency for the policy it is attached to by writing to the
386 ``scaling_setspeed`` attribute of that policy. 371 ``scaling_setspeed`` attribute of that policy.
387 372
388 ``schedutil`` 373 ``schedutil``
389 ------------- 374 -------------
390 375
391 This governor uses CPU utilization data availa 376 This governor uses CPU utilization data available from the CPU scheduler. It
392 generally is regarded as a part of the CPU sch 377 generally is regarded as a part of the CPU scheduler, so it can access the
393 scheduler's internal data structures directly. 378 scheduler's internal data structures directly.
394 379
395 It runs entirely in scheduler context, althoug 380 It runs entirely in scheduler context, although in some cases it may need to
396 invoke the scaling driver asynchronously when 381 invoke the scaling driver asynchronously when it decides that the CPU frequency
397 should be changed for a given policy (that dep 382 should be changed for a given policy (that depends on whether or not the driver
398 is capable of changing the CPU frequency from 383 is capable of changing the CPU frequency from scheduler context).
399 384
400 The actions of this governor for a particular 385 The actions of this governor for a particular CPU depend on the scheduling class
401 invoking its utilization update callback for t 386 invoking its utilization update callback for that CPU. If it is invoked by the
402 RT or deadline scheduling classes, the governo 387 RT or deadline scheduling classes, the governor will increase the frequency to
403 the allowed maximum (that is, the ``scaling_ma 388 the allowed maximum (that is, the ``scaling_max_freq`` policy limit). In turn,
404 if it is invoked by the CFS scheduling class, 389 if it is invoked by the CFS scheduling class, the governor will use the
405 Per-Entity Load Tracking (PELT) metric for the 390 Per-Entity Load Tracking (PELT) metric for the root control group of the
406 given CPU as the CPU utilization estimate (see !! 391 given CPU as the CPU utilization estimate (see the `Per-entity load tracking`_
407 LWN.net article [1]_ for a description of the !! 392 LWN.net article for a description of the PELT mechanism). Then, the new
408 CPU frequency to apply is computed in accordan 393 CPU frequency to apply is computed in accordance with the formula
409 394
410 f = 1.25 * ``f_0`` * ``util`` / ``max` 395 f = 1.25 * ``f_0`` * ``util`` / ``max``
411 396
412 where ``util`` is the PELT number, ``max`` is 397 where ``util`` is the PELT number, ``max`` is the theoretical maximum of
413 ``util``, and ``f_0`` is either the maximum po 398 ``util``, and ``f_0`` is either the maximum possible CPU frequency for the given
414 policy (if the PELT number is frequency-invari 399 policy (if the PELT number is frequency-invariant), or the current CPU frequency
415 (otherwise). 400 (otherwise).
416 401
417 This governor also employs a mechanism allowin 402 This governor also employs a mechanism allowing it to temporarily bump up the
418 CPU frequency for tasks that have been waiting 403 CPU frequency for tasks that have been waiting on I/O most recently, called
419 "IO-wait boosting". That happens when the :c: 404 "IO-wait boosting". That happens when the :c:macro:`SCHED_CPUFREQ_IOWAIT` flag
420 is passed by the scheduler to the governor cal 405 is passed by the scheduler to the governor callback which causes the frequency
421 to go up to the allowed maximum immediately an 406 to go up to the allowed maximum immediately and then draw back to the value
422 returned by the above formula over time. 407 returned by the above formula over time.
423 408
424 This governor exposes only one tunable: 409 This governor exposes only one tunable:
425 410
426 ``rate_limit_us`` 411 ``rate_limit_us``
427 Minimum time (in microseconds) that ha 412 Minimum time (in microseconds) that has to pass between two consecutive
428 runs of governor computations (default !! 413 runs of governor computations (default: 1000 times the scaling driver's
429 transition latency or the maximum 2ms) !! 414 transition latency).
430 415
431 The purpose of this tunable is to redu 416 The purpose of this tunable is to reduce the scheduler context overhead
432 of the governor which might be excessi 417 of the governor which might be excessive without it.
433 418
434 This governor generally is regarded as a repla 419 This governor generally is regarded as a replacement for the older `ondemand`_
435 and `conservative`_ governors (described below 420 and `conservative`_ governors (described below), as it is simpler and more
436 tightly integrated with the CPU scheduler, its 421 tightly integrated with the CPU scheduler, its overhead in terms of CPU context
437 switches and similar is less significant, and 422 switches and similar is less significant, and it uses the scheduler's own CPU
438 utilization metric, so in principle its decisi 423 utilization metric, so in principle its decisions should not contradict the
439 decisions made by the other parts of the sched 424 decisions made by the other parts of the scheduler.
440 425
441 ``ondemand`` 426 ``ondemand``
442 ------------ 427 ------------
443 428
444 This governor uses CPU load as a CPU frequency 429 This governor uses CPU load as a CPU frequency selection metric.
445 430
446 In order to estimate the current CPU load, it 431 In order to estimate the current CPU load, it measures the time elapsed between
447 consecutive invocations of its worker routine 432 consecutive invocations of its worker routine and computes the fraction of that
448 time in which the given CPU was not idle. The 433 time in which the given CPU was not idle. The ratio of the non-idle (active)
449 time to the total CPU time is taken as an esti 434 time to the total CPU time is taken as an estimate of the load.
450 435
451 If this governor is attached to a policy share 436 If this governor is attached to a policy shared by multiple CPUs, the load is
452 estimated for all of them and the greatest res 437 estimated for all of them and the greatest result is taken as the load estimate
453 for the entire policy. 438 for the entire policy.
454 439
455 The worker routine of this governor has to run 440 The worker routine of this governor has to run in process context, so it is
456 invoked asynchronously (via a workqueue) and C 441 invoked asynchronously (via a workqueue) and CPU P-states are updated from
457 there if necessary. As a result, the schedule 442 there if necessary. As a result, the scheduler context overhead from this
458 governor is minimum, but it causes additional 443 governor is minimum, but it causes additional CPU context switches to happen
459 relatively often and the CPU P-state updates t 444 relatively often and the CPU P-state updates triggered by it can be relatively
460 irregular. Also, it affects its own CPU load 445 irregular. Also, it affects its own CPU load metric by running code that
461 reduces the CPU idle time (even though the CPU 446 reduces the CPU idle time (even though the CPU idle time is only reduced very
462 slightly by it). 447 slightly by it).
463 448
464 It generally selects CPU frequencies proportio 449 It generally selects CPU frequencies proportional to the estimated load, so that
465 the value of the ``cpuinfo_max_freq`` policy a 450 the value of the ``cpuinfo_max_freq`` policy attribute corresponds to the load of
466 1 (or 100%), and the value of the ``cpuinfo_mi 451 1 (or 100%), and the value of the ``cpuinfo_min_freq`` policy attribute
467 corresponds to the load of 0, unless when the 452 corresponds to the load of 0, unless when the load exceeds a (configurable)
468 speedup threshold, in which case it will go st 453 speedup threshold, in which case it will go straight for the highest frequency
469 it is allowed to use (the ``scaling_max_freq`` 454 it is allowed to use (the ``scaling_max_freq`` policy limit).
470 455
471 This governor exposes the following tunables: 456 This governor exposes the following tunables:
472 457
473 ``sampling_rate`` 458 ``sampling_rate``
474 This is how often the governor's worke 459 This is how often the governor's worker routine should run, in
475 microseconds. 460 microseconds.
476 461
477 Typically, it is set to values of the !! 462 Typically, it is set to values of the order of 10000 (10 ms). Its
478 default value is to add a 50% breathin !! 463 default value is equal to the value of ``cpuinfo_transition_latency``
479 to ``cpuinfo_transition_latency`` on e !! 464 for each policy this governor is attached to (but since the unit here
480 attached to. The minimum is typically !! 465 is greater by 1000, this means that the time represented by
481 ticks. !! 466 ``sampling_rate`` is 1000 times greater than the transition latency by
>> 467 default).
482 468
483 If this tunable is per-policy, the fol 469 If this tunable is per-policy, the following shell command sets the time
484 represented by it to be 1.5 times as h !! 470 represented by it to be 750 times as high as the transition latency::
485 (the default):: <<
486 471
487 # echo `$(($(cat cpuinfo_transition_la !! 472 # echo `$(($(cat cpuinfo_transition_latency) * 750 / 1000)) > ondemand/sampling_rate
>> 473
>> 474
>> 475 ``min_sampling_rate``
>> 476 The minimum value of ``sampling_rate``.
>> 477
>> 478 Equal to 10000 (10 ms) if :c:macro:`CONFIG_NO_HZ_COMMON` and
>> 479 :c:data:`tick_nohz_active` are both set or to 20 times the value of
>> 480 :c:data:`jiffies` in microseconds otherwise.
488 481
489 ``up_threshold`` 482 ``up_threshold``
490 If the estimated CPU load is above thi 483 If the estimated CPU load is above this value (in percent), the governor
491 will set the frequency to the maximum 484 will set the frequency to the maximum value allowed for the policy.
492 Otherwise, the selected frequency will 485 Otherwise, the selected frequency will be proportional to the estimated
493 CPU load. 486 CPU load.
494 487
495 ``ignore_nice_load`` 488 ``ignore_nice_load``
496 If set to 1 (default 0), it will cause 489 If set to 1 (default 0), it will cause the CPU load estimation code to
497 treat the CPU time spent on executing 490 treat the CPU time spent on executing tasks with "nice" levels greater
498 than 0 as CPU idle time. 491 than 0 as CPU idle time.
499 492
500 This may be useful if there are tasks 493 This may be useful if there are tasks in the system that should not be
501 taken into account when deciding what 494 taken into account when deciding what frequency to run the CPUs at.
502 Then, to make that happen it is suffic 495 Then, to make that happen it is sufficient to increase the "nice" level
503 of those tasks above 0 and set this at 496 of those tasks above 0 and set this attribute to 1.
504 497
505 ``sampling_down_factor`` 498 ``sampling_down_factor``
506 Temporary multiplier, between 1 (defau 499 Temporary multiplier, between 1 (default) and 100 inclusive, to apply to
507 the ``sampling_rate`` value if the CPU 500 the ``sampling_rate`` value if the CPU load goes above ``up_threshold``.
508 501
509 This causes the next execution of the 502 This causes the next execution of the governor's worker routine (after
510 setting the frequency to the allowed m 503 setting the frequency to the allowed maximum) to be delayed, so the
511 frequency stays at the maximum level f 504 frequency stays at the maximum level for a longer time.
512 505
513 Frequency fluctuations in some bursty 506 Frequency fluctuations in some bursty workloads may be avoided this way
514 at the cost of additional energy spent 507 at the cost of additional energy spent on maintaining the maximum CPU
515 capacity. 508 capacity.
516 509
517 ``powersave_bias`` 510 ``powersave_bias``
518 Reduction factor to apply to the origi 511 Reduction factor to apply to the original frequency target of the
519 governor (including the maximum value 512 governor (including the maximum value used when the ``up_threshold``
520 value is exceeded by the estimated CPU 513 value is exceeded by the estimated CPU load) or sensitivity threshold
521 for the AMD frequency sensitivity powe 514 for the AMD frequency sensitivity powersave bias driver
522 (:file:`drivers/cpufreq/amd_freq_sensi 515 (:file:`drivers/cpufreq/amd_freq_sensitivity.c`), between 0 and 1000
523 inclusive. 516 inclusive.
524 517
525 If the AMD frequency sensitivity power 518 If the AMD frequency sensitivity powersave bias driver is not loaded,
526 the effective frequency to apply is gi 519 the effective frequency to apply is given by
527 520
528 f * (1 - ``powersave_bias`` / 521 f * (1 - ``powersave_bias`` / 1000)
529 522
530 where f is the governor's original fre 523 where f is the governor's original frequency target. The default value
531 of this attribute is 0 in that case. 524 of this attribute is 0 in that case.
532 525
533 If the AMD frequency sensitivity power 526 If the AMD frequency sensitivity powersave bias driver is loaded, the
534 value of this attribute is 400 by defa 527 value of this attribute is 400 by default and it is used in a different
535 way. 528 way.
536 529
537 On Family 16h (and later) AMD processo 530 On Family 16h (and later) AMD processors there is a mechanism to get a
538 measured workload sensitivity, between 531 measured workload sensitivity, between 0 and 100% inclusive, from the
539 hardware. That value can be used to e 532 hardware. That value can be used to estimate how the performance of the
540 workload running on a CPU will change 533 workload running on a CPU will change in response to frequency changes.
541 534
542 The performance of a workload with the 535 The performance of a workload with the sensitivity of 0 (memory-bound or
543 IO-bound) is not expected to increase 536 IO-bound) is not expected to increase at all as a result of increasing
544 the CPU frequency, whereas workloads w 537 the CPU frequency, whereas workloads with the sensitivity of 100%
545 (CPU-bound) are expected to perform mu 538 (CPU-bound) are expected to perform much better if the CPU frequency is
546 increased. 539 increased.
547 540
548 If the workload sensitivity is less th 541 If the workload sensitivity is less than the threshold represented by
549 the ``powersave_bias`` value, the sens 542 the ``powersave_bias`` value, the sensitivity powersave bias driver
550 will cause the governor to select a fr 543 will cause the governor to select a frequency lower than its original
551 target, so as to avoid over-provisioni 544 target, so as to avoid over-provisioning workloads that will not benefit
552 from running at higher CPU frequencies 545 from running at higher CPU frequencies.
553 546
554 ``conservative`` 547 ``conservative``
555 ---------------- 548 ----------------
556 549
557 This governor uses CPU load as a CPU frequency 550 This governor uses CPU load as a CPU frequency selection metric.
558 551
559 It estimates the CPU load in the same way as t 552 It estimates the CPU load in the same way as the `ondemand`_ governor described
560 above, but the CPU frequency selection algorit 553 above, but the CPU frequency selection algorithm implemented by it is different.
561 554
562 Namely, it avoids changing the frequency signi 555 Namely, it avoids changing the frequency significantly over short time intervals
563 which may not be suitable for systems with lim 556 which may not be suitable for systems with limited power supply capacity (e.g.
564 battery-powered). To achieve that, it changes 557 battery-powered). To achieve that, it changes the frequency in relatively
565 small steps, one step at a time, up or down - 558 small steps, one step at a time, up or down - depending on whether or not a
566 (configurable) threshold has been exceeded by 559 (configurable) threshold has been exceeded by the estimated CPU load.
567 560
568 This governor exposes the following tunables: 561 This governor exposes the following tunables:
569 562
570 ``freq_step`` 563 ``freq_step``
571 Frequency step in percent of the maxim 564 Frequency step in percent of the maximum frequency the governor is
572 allowed to set (the ``scaling_max_freq 565 allowed to set (the ``scaling_max_freq`` policy limit), between 0 and
573 100 (5 by default). 566 100 (5 by default).
574 567
575 This is how much the frequency is allo 568 This is how much the frequency is allowed to change in one go. Setting
576 it to 0 will cause the default frequen 569 it to 0 will cause the default frequency step (5 percent) to be used
577 and setting it to 100 effectively caus 570 and setting it to 100 effectively causes the governor to periodically
578 switch the frequency between the ``sca 571 switch the frequency between the ``scaling_min_freq`` and
579 ``scaling_max_freq`` policy limits. 572 ``scaling_max_freq`` policy limits.
580 573
581 ``down_threshold`` 574 ``down_threshold``
582 Threshold value (in percent, 20 by def 575 Threshold value (in percent, 20 by default) used to determine the
583 frequency change direction. 576 frequency change direction.
584 577
585 If the estimated CPU load is greater t 578 If the estimated CPU load is greater than this value, the frequency will
586 go up (by ``freq_step``). If the load 579 go up (by ``freq_step``). If the load is less than this value (and the
587 ``sampling_down_factor`` mechanism is 580 ``sampling_down_factor`` mechanism is not in effect), the frequency will
588 go down. Otherwise, the frequency wil 581 go down. Otherwise, the frequency will not be changed.
589 582
590 ``sampling_down_factor`` 583 ``sampling_down_factor``
591 Frequency decrease deferral factor, be 584 Frequency decrease deferral factor, between 1 (default) and 10
592 inclusive. 585 inclusive.
593 586
594 It effectively causes the frequency to 587 It effectively causes the frequency to go down ``sampling_down_factor``
595 times slower than it ramps up. 588 times slower than it ramps up.
596 589
597 590
598 Frequency Boost Support 591 Frequency Boost Support
599 ======================= 592 =======================
600 593
601 Background 594 Background
602 ---------- 595 ----------
603 596
604 Some processors support a mechanism to raise t 597 Some processors support a mechanism to raise the operating frequency of some
605 cores in a multicore package temporarily (and 598 cores in a multicore package temporarily (and above the sustainable frequency
606 threshold for the whole package) under certain 599 threshold for the whole package) under certain conditions, for example if the
607 whole chip is not fully utilized and below its 600 whole chip is not fully utilized and below its intended thermal or power budget.
608 601
609 Different names are used by different vendors 602 Different names are used by different vendors to refer to this functionality.
610 For Intel processors it is referred to as "Tur 603 For Intel processors it is referred to as "Turbo Boost", AMD calls it
611 "Turbo-Core" or (in technical documentation) " 604 "Turbo-Core" or (in technical documentation) "Core Performance Boost" and so on.
612 As a rule, it also is implemented differently 605 As a rule, it also is implemented differently by different vendors. The simple
613 term "frequency boost" is used here for brevit 606 term "frequency boost" is used here for brevity to refer to all of those
614 implementations. 607 implementations.
615 608
616 The frequency boost mechanism may be either ha 609 The frequency boost mechanism may be either hardware-based or software-based.
617 If it is hardware-based (e.g. on x86), the dec 610 If it is hardware-based (e.g. on x86), the decision to trigger the boosting is
618 made by the hardware (although in general it r 611 made by the hardware (although in general it requires the hardware to be put
619 into a special state in which it can control t 612 into a special state in which it can control the CPU frequency within certain
620 limits). If it is software-based (e.g. on ARM 613 limits). If it is software-based (e.g. on ARM), the scaling driver decides
621 whether or not to trigger boosting and when to 614 whether or not to trigger boosting and when to do that.
622 615
623 The ``boost`` File in ``sysfs`` 616 The ``boost`` File in ``sysfs``
624 ------------------------------- 617 -------------------------------
625 618
626 This file is located under :file:`/sys/devices 619 This file is located under :file:`/sys/devices/system/cpu/cpufreq/` and controls
627 the "boost" setting for the whole system. It 620 the "boost" setting for the whole system. It is not present if the underlying
628 scaling driver does not support the frequency 621 scaling driver does not support the frequency boost mechanism (or supports it,
629 but provides a driver-specific interface for c 622 but provides a driver-specific interface for controlling it, like
630 |intel_pstate|). 623 |intel_pstate|).
631 624
632 If the value in this file is 1, the frequency 625 If the value in this file is 1, the frequency boost mechanism is enabled. This
633 means that either the hardware can be put into 626 means that either the hardware can be put into states in which it is able to
634 trigger boosting (in the hardware-based case), 627 trigger boosting (in the hardware-based case), or the software is allowed to
635 trigger boosting (in the software-based case). 628 trigger boosting (in the software-based case). It does not mean that boosting
636 is actually in use at the moment on any CPUs i 629 is actually in use at the moment on any CPUs in the system. It only means a
637 permission to use the frequency boost mechanis 630 permission to use the frequency boost mechanism (which still may never be used
638 for other reasons). 631 for other reasons).
639 632
640 If the value in this file is 0, the frequency 633 If the value in this file is 0, the frequency boost mechanism is disabled and
641 cannot be used at all. 634 cannot be used at all.
642 635
643 The only values that can be written to this fi 636 The only values that can be written to this file are 0 and 1.
644 637
645 Rationale for Boost Control Knob 638 Rationale for Boost Control Knob
646 -------------------------------- 639 --------------------------------
647 640
648 The frequency boost mechanism is generally int 641 The frequency boost mechanism is generally intended to help to achieve optimum
649 CPU performance on time scales below software 642 CPU performance on time scales below software resolution (e.g. below the
650 scheduler tick interval) and it is demonstrabl 643 scheduler tick interval) and it is demonstrably suitable for many workloads, but
651 it may lead to problems in certain situations. 644 it may lead to problems in certain situations.
652 645
653 For this reason, many systems make it possible 646 For this reason, many systems make it possible to disable the frequency boost
654 mechanism in the platform firmware (BIOS) setu 647 mechanism in the platform firmware (BIOS) setup, but that requires the system to
655 be restarted for the setting to be adjusted as 648 be restarted for the setting to be adjusted as desired, which may not be
656 practical at least in some cases. For example 649 practical at least in some cases. For example:
657 650
658 1. Boosting means overclocking the processor 651 1. Boosting means overclocking the processor, although under controlled
659 conditions. Generally, the processor's e 652 conditions. Generally, the processor's energy consumption increases
660 as a result of increasing its frequency a 653 as a result of increasing its frequency and voltage, even temporarily.
661 That may not be desirable on systems that 654 That may not be desirable on systems that switch to power sources of
662 limited capacity, such as batteries, so t 655 limited capacity, such as batteries, so the ability to disable the boost
663 mechanism while the system is running may 656 mechanism while the system is running may help there (but that depends on
664 the workload too). 657 the workload too).
665 658
666 2. In some situations deterministic behavior 659 2. In some situations deterministic behavior is more important than
667 performance or energy consumption (or bot 660 performance or energy consumption (or both) and the ability to disable
668 boosting while the system is running may 661 boosting while the system is running may be useful then.
669 662
670 3. To examine the impact of the frequency bo 663 3. To examine the impact of the frequency boost mechanism itself, it is useful
671 to be able to run tests with and without 664 to be able to run tests with and without boosting, preferably without
672 restarting the system in the meantime. 665 restarting the system in the meantime.
673 666
674 4. Reproducible results are important when r 667 4. Reproducible results are important when running benchmarks. Since
675 the boosting functionality depends on the 668 the boosting functionality depends on the load of the whole package,
676 single-thread performance may vary becaus 669 single-thread performance may vary because of it which may lead to
677 unreproducible results sometimes. That c 670 unreproducible results sometimes. That can be avoided by disabling the
678 frequency boost mechanism before running 671 frequency boost mechanism before running benchmarks sensitive to that
679 issue. 672 issue.
680 673
681 Legacy AMD ``cpb`` Knob 674 Legacy AMD ``cpb`` Knob
682 ----------------------- 675 -----------------------
683 676
684 The AMD powernow-k8 scaling driver supports a 677 The AMD powernow-k8 scaling driver supports a ``sysfs`` knob very similar to
685 the global ``boost`` one. It is used for disa 678 the global ``boost`` one. It is used for disabling/enabling the "Core
686 Performance Boost" feature of some AMD process 679 Performance Boost" feature of some AMD processors.
687 680
688 If present, that knob is located in every ``CP 681 If present, that knob is located in every ``CPUFreq`` policy directory in
689 ``sysfs`` (:file:`/sys/devices/system/cpu/cpuf 682 ``sysfs`` (:file:`/sys/devices/system/cpu/cpufreq/policyX/`) and is called
690 ``cpb``, which indicates a more fine grained c 683 ``cpb``, which indicates a more fine grained control interface. The actual
691 implementation, however, works on the system-w 684 implementation, however, works on the system-wide basis and setting that knob
692 for one policy causes the same value of it to 685 for one policy causes the same value of it to be set for all of the other
693 policies at the same time. 686 policies at the same time.
694 687
695 That knob is still supported on AMD processors 688 That knob is still supported on AMD processors that support its underlying
696 hardware feature, but it may be configured out 689 hardware feature, but it may be configured out of the kernel (via the
697 :c:macro:`CONFIG_X86_ACPI_CPUFREQ_CPB` configu 690 :c:macro:`CONFIG_X86_ACPI_CPUFREQ_CPB` configuration option) and the global
698 ``boost`` knob is present regardless. Thus it 691 ``boost`` knob is present regardless. Thus it is always possible use the
699 ``boost`` knob instead of the ``cpb`` one whic 692 ``boost`` knob instead of the ``cpb`` one which is highly recommended, as that
700 is more consistent with what all of the other 693 is more consistent with what all of the other systems do (and the ``cpb`` knob
701 may not be supported any more in the future). 694 may not be supported any more in the future).
702 695
703 The ``cpb`` knob is never present for any proc 696 The ``cpb`` knob is never present for any processors without the underlying
704 hardware feature (e.g. all Intel ones), even i 697 hardware feature (e.g. all Intel ones), even if the
705 :c:macro:`CONFIG_X86_ACPI_CPUFREQ_CPB` configu 698 :c:macro:`CONFIG_X86_ACPI_CPUFREQ_CPB` configuration option is set.
706 699
707 700
708 References !! 701 .. _Per-entity load tracking: https://lwn.net/Articles/531853/
709 ========== <<
710 <<
711 .. [1] Jonathan Corbet, *Per-entity load track <<
712 https://lwn.net/Articles/531853/ <<
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.