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