cpufreq: intel_pstate: Support raw epp and energy_efficiency
Subsystems (3)
Srinivas Pandruvada
maintainerv2
Remove raw value support for EPB
Srinivas Pandruvada (2):
cpufreq: intel_pstate: Allow enable/disable energy efficiency
cpufreq: intel_pstate: Allow raw energy performance preference value
Documentation/admin-guide/pm/intel_pstate.rst | 13 ++- drivers/cpufreq/intel_pstate.c | 99 ++++++++++++++++--- 2 files changed, 99 insertions(+), 13 deletions(-)
Signature · 1 line
2.25.4
Srinivas Pandruvada
maintainerPATCH v2 1/2 cpufreq: intel_pstate: Allow enable/disable energy efficiency
By default intel_pstate driver disables energy efficiency by setting MSR_IA32_POWER_CTL bit 19 for Kaby Lake desktop CPU model in HWP mode. This CPU model is also shared by Coffee Lake desktop CPUs. This allows these systems to reach maximum possible frequency. But this adds power penalty, which some customers don't want. They want some way to enable/ disable dynamically.
So, add an additional attribute "energy_efficiency_enable" under /sys/devices/system/cpu/intel_pstate/ for these CPU models. This allows to read and write bit 19 ("Disable Energy Efficiency Optimization") in the MSR IA32_POWER_CTL.
This attribute is present in both HWP and non-HWP mode as this has an effect in both modes. Refer to Intel Software Developer's manual for details. The scope of this bit is package wide.
--- Documentation/admin-guide/pm/intel_pstate.rst | 7 +++ drivers/cpufreq/intel_pstate.c | 49 ++++++++++++++++++- 2 files changed, 54 insertions(+), 2 deletions(-)
Documentation/admin-guide/pm/intel_pstate.rst +7 -0
@@ -431,6 +431,13 @@ argument is passed to the kernel in the command line. supported in the current configuration, writes to this attribute will fail with an appropriate error. +``energy_efficiency_enable`` + This attribute is only present on platforms, which has CPUs matching + Kaby Lake desktop CPU model. By default "energy_efficiency" is disabled + on these CPU models in HWP mode by this driver. Enabling energy + efficiency may limit maximum operating frequency in both HWP and non + HWP mode. + Interpretation of Policy Attributes -----------------------------------
drivers/cpufreq/intel_pstate.c +47 -2
79 lines not on this page · raw message
Signature · 1 line
2.25.4
Srinivas Pandruvada
maintainerPATCH v2 2/2 cpufreq: intel_pstate: Allow raw energy performance preference value
In reply to Srinivas PandruvadaCurrently using attribute "energy_performance_preference", user space can write one of the four per-defined preference string. These preference strings gets mapped to a hard-coded Energy-Performance Preference (EPP) or Energy-Performance Bias (EPB) knob.
These four values supposed to cover broad spectrum of use cases, but they are not uniformly distributed in the range. There are number of cases, where this is not enough. For example:
Suppose user wants more performance when connected to AC. Instead of using default "balance performance", the "performance" setting can be used. This changes EPP value from 0x80 to 0x00. But setting EPP to 0, results in electrical and thermal issues on some platforms. This results in CPU to do aggressive throttling, which causes drop in performance. But some value between 0x80 and 0x00 results in better performance. But that value can't be fixed as the power curve is not linear. In some cases just changing EPP from 0x80 to 0x75 is enough to get significant performance gain.
Similarly on battery EPP 0x80 can be very aggressive in power consumption. But picking up the next choice "balance power" results in too much loss of performance, which cause bad user experience in use case like "Google Hangout". It was observed that some value between these two EPP is optimal.
This change allows fine grain EPP tuning for platform like Chromebooks. Here based on the product and use cases, different EPP values can be set. This change is similar to the change done for:/sys/devices/system/cpu/cpu*/power/energy_perf_bias
where user has choice to write a predefined string or raw value.
The change itself is trivial. When user preference doesn't match predefined string preferences and value is an unsigned integer and in range, use that value for EPP. When the EPP feature is not prsent writing raw value is not supported.
--- Documentation/admin-guide/pm/intel_pstate.rst | 6 ++- drivers/cpufreq/intel_pstate.c | 50 +++++++++++++++---- 2 files changed, 45 insertions(+), 11 deletions(-)
Documentation/admin-guide/pm/intel_pstate.rst +5 -1
@@ -561,7 +561,11 @@ somewhere between the two extremes: Strings written to the ``energy_performance_preference`` attribute are internally translated to integer values written to the processor's Energy-Performance Preference (EPP) knob (if supported) or its -Energy-Performance Bias (EPB) knob. +Energy-Performance Bias (EPB) knob. It is also possible to write a positive +integer value between 0 to 255, if the EPP feature is present. If the EPP +feature is not present, writing integer value to this attribute is not +supported. In this case, user can use + "/sys/devices/system/cpu/cpu*/power/energy_perf_bias" interface. [Note that tasks may by migrated from one CPU to another by the scheduler's load-balancing algorithm and if different energy vs performance hints are
drivers/cpufreq/intel_pstate.c +40 -10
105 lines not on this page · raw message
Signature · 1 line
2.25.4
RE: [PATCH v2 2/2] cpufreq: intel_pstate: Allow raw energy performance preference value
Hi Srinivas,
I have immediate need for this. I have been using a tool I wrote myself for this which I can now retire. (it wasn't very good anyway).
Yours remembers for each governor, and is way better. Thanks.
Quoted 7 lines · On 2020.06.23 11:27 Srinivas Pandruvada wrote:
Currently using attribute "
energy_performance_preference", user space can write one of the four per-defined preference string. These preference strings gets mapped to a hard-coded Energy-Performance Preference (EPP) or Energy-Performance Bias (EPB) knob.These four values supposed to cover broad spectrum of use cases, but they are not uniformly distributed in the range.
Suggest:
These four values are supposed to cover broad spectrum of use cases, but are not uniformly distributed in the range.
Quoted 10 lines
There are number of cases,
where this is not enough. For example:Suppose user wants more performance when connected to AC. Instead of using default "balance performance", the "performance" setting can be used. This changes EPP value from
0x80to0x00. But setting EPP to 0, results in electrical and thermal issues on some platforms.This results in CPU to do
aggressive throttling, which causes drop in performance.
Suggest:
This results in aggressive throttling, which causes a drop in performance.
And:
Tough.
I consider "performance mode" as sacrosanct, and have always expected these to behave identically and at max CPU freq:
intel_pstate no-hwp / performance intel_cpufreq no-hwp / performance (a.k.a. passive) acpi_cpufreq / performanceintel_pstate hwp / performanceintel_cpufreq hwp / performance (in future)
as was always the case on my i7-2600K (no hwp) based computer and is not the case on my i5-9600K (hwp capable) computer.
Quoted 17 lines
But some value
between0x80and0x00results in better performance. But that value can't be fixed as the power curve is not linear. In some cases just changing EPP from0x80to0x75is enough to get significant performance gain.Similarly on battery EPP
0x80can be very aggressive in power consumption. But picking up the next choice "balance power" results in too much loss of performance, which cause bad user experience in use case like "Google Hangout". It was observed that some value between these two EPP is optimal.This change allows fine grain EPP tuning for platform like Chromebooks. Here based on the product and use cases, different EPP values can be set. This change is similar to the change done for:
/sys/devices/system/cpu/cpu*/power/energy_perf_bias
where user has choice to write a predefined string or raw value.
Quoted 3 lines
The change itself is trivial. When user preference doesn't match predefined string preferences and value is an unsigned integer and in range, use that value for EPP. When the EPP feature is not prsent
^^^^^^
s/prsent/present
Quoted 145 lines
writing raw value is not supported.
Suggested-by Len Brown <[email protected]>Signed-off-by Srinivas Pandruvada <[email protected]>--- Documentation/admin-guide/pm/intel_pstate.rst | 6 ++- drivers/cpufreq/intel_pstate.c | 50 +++++++++++++++---- 2 files changed, 45 insertions(+), 11 deletions(-)Documentation/admin- renamed from Documentation/admin-guide/pm/intel_pstate.rst +0
-0
guide/pm/intel_pstate.rst
index939bfdc53f4f..5e209926e0ed100644Documentation/admin-guide/pm/intel_pstate.rst +5
-1@@ -561,7 +561,11 @@ somewhere between the two extremes: Strings written to the ``energy_performance_preference`` attribute are internally translated to integer values written to the processor's Energy-Performance Preference (EPP) knob (if supported) or its -Energy-Performance Bias (EPB) knob. +Energy-Performance Bias (EPB) knob. It is also possible to write a positive +integer value between 0 to 255, if the EPP feature is present. If the EPP +feature is not present, writing integer value to this attribute is not +supported. In this case, user can use + "/sys/devices/system/cpu/cpu*/power/energy_perf_bias" interface. [Note that tasks may by migrated from one CPU to another by the scheduler's load-balancing algorithm and if different energy vs performance hints aredrivers/cpufreq/intel_pstate.c +40
-11@@ -602,11 +602,12 @@ static const unsigned int epp_values[] = { HWP_EPP_POWERSAVE }; -static int intel_pstate_get_energy_pref_index (struct cpudata *cpu_data) +static int intel_pstate_get_energy_pref_index (struct cpudata *cpu_data,int *raw_epp) {s16 epp;int index = -EINVAL ; + *raw_epp =0 ; epp =intel_pstate_get_epp (cpu_data,0 );if (epp <0 )return epp;@@ -614,12 +615,14 @@ static int intel_pstate_get_energy_pref_index(struct cpudata *cpu_data)if (boot_cpu_has (X86_FEATURE_HWP_EPP )) {if (epp ==HWP_EPP_PERFORMANCE )return 1 ; -if (epp <=HWP_EPP_BALANCE_PERFORMANCE ) +if (epp ==HWP_EPP_BALANCE_PERFORMANCE )return 2 ; -if (epp <=HWP_EPP_BALANCE_POWERSAVE ) +if (epp ==HWP_EPP_BALANCE_POWERSAVE )return 3 ; -else +if (epp ==HWP_EPP_POWERSAVE )return 4 ; + *raw_epp = epp; +return 0 ; }else if (boot_cpu_has (X86_FEATURE_EPB )) {/* * Range: @@ -638,7 +641,8 @@ static int intel_pstate_get_energy_pref_index(struct cpudata *cpu_data) }static int intel_pstate_set_energy_pref_index (struct cpudata *cpu_data, -int pref_index) +int pref_index,bool use_raw, +u32 raw_epp) {int epp = -EINVAL ;int ret;@@ -657,6 +661,16 @@ static int intel_pstate_set_energy_pref_index(struct cpudata *cpu_data, value &= ~GENMASK_ULL (31 ,24 ); +if (use_raw) { +if (raw_epp >255 ) { + ret = -EINVAL ; + goto return_pref; + } + value |= (u64 )raw_epp <<24 ; + ret =wrmsrl_on_cpu (cpu_data->cpu,MSR_HWP_REQUEST , value); + goto return_pref; + } +if (epp == -EINVAL ) epp = epp_values[pref_index -1 ];@@ -694,6 +708,8 @@ static ssize_t store_energy_performance_preference( {struct cpudata *cpu_data = all_cpu_data[policy->cpu];char str_preference[21 ]; +bool raw = false; +u32 epp;int ret; ret =sscanf (buf,"%20s" , str_preference);@@ -701,10 +717,21 @@ static ssize_t store_energy_performance_preference(return -EINVAL ; ret =match_string (energy_perf_strings,-1 , str_preference); -if (ret <0 ) +if (ret <0 ) { +if (!boot_cpu_has (X86_FEATURE_HWP_EPP )) +return ret; + + ret =kstrtouint (buf,10 , &epp); +if (ret) +return ret; + + raw = true; + } + + ret =intel_pstate_set_energy_pref_index (cpu_data, ret, raw, epp); +if (ret)return ret; -intel_pstate_set_energy_pref_index (cpu_data, ret);return count; }@@ -712,13 +739,16 @@ static ssize_t show_energy_performance_preference( struct cpufreq_policy *policy, char *buf) { struct cpudata *cpu_data = all_cpu_data[policy->cpu]; -int preference; +int preference, raw_epp; - preference =intel_pstate_get_energy_pref_index (cpu_data); + preference =intel_pstate_get_energy_pref_index (cpu_data, &raw_epp);if (preference <0 )return preference; -return sprintf (buf,"%s\n" , energy_perf_strings[preference]); +if (raw_epp) +return sprintf (buf,"%d\n" , raw_epp); +else +return sprintf (buf,"%s\n" , energy_perf_strings[preference]); }cpufreq_freq_attr_rw (energy_performance_preference); --2.25.4
srinivas pandruvada
maintainerRe: [PATCH v2 2/2] cpufreq: intel_pstate: Allow raw energy performance preference value
Hi Doug,
Quoted 8 lines · Doug Smythies
Hi Srinivas,
I have immediate need for this. I have been using a tool I wrote myself for this which I can now retire. (it wasn't very good anyway).
Yours remembers for each governor, and is way better. Thanks.
I will incorporate your changes and re-post.
Thanks,
Srinivas
Quoted 253 lines · On 2020.06.23 11:27 Srinivas Pandruvada wrote:
Quoted 9 lines
Currently using attribute "
energy_performance_preference", user space can
write one of the four per-defined preference string. These preference
strings gets mapped to a hard-coded Energy-Performance Preference (EPP) or
Energy-Performance Bias (EPB) knob.Quoted 3 lines
These four values supposed to cover broad spectrum of use cases, but they
are not uniformly distributed in the range.Suggest:
These four values are supposed to cover broad spectrum of use cases, but
are not uniformly distributed in the range.Quoted 12 lines
There are number of cases,
where this is not enough. For example:Suppose user wants more performance when connected to AC. Instead of using
default "balance performance", the "performance" setting can be used. This
changes EPP value from0x80to0x00. But setting EPP to 0, results in
electrical and thermal issues on some platforms.
This results in CPU to do
aggressive throttling, which causes drop in performance.Suggest:
This results in aggressive throttling, which causes a drop in performance.
And:
Tough.
I consider "performance mode" as sacrosanct, and have always expected these to behave identically and at max CPU freq:
intel_pstateno-hwp / performanceintel_cpufreqno-hwp / performance (a.k.a. passive)acpi_cpufreq/ performanceintel_pstatehwp / performanceintel_cpufreqhwp / performance (in future)as was always the case on my i7-2600K (no hwp) based computer and is not the case on my i5-9600K (hwp capable) computer.
Quoted 24 lines
But some value
between0x80and0x00results in better performance. But that value can't
be fixed as the power curve is not linear. In some cases just changing EPP
from0x80to0x75is enough to get significant performance gain.Similarly on battery EPP
0x80can be very aggressive in power consumption.
But picking up the next choice "balance power" results in too much loss
of performance, which cause bad user experience in use case like "Google
Hangout". It was observed that some value between these two EPP is optimal.This change allows fine grain EPP tuning for platform like Chromebooks.
Here based on the product and use cases, different EPP values can be set.
This change is similar to the change done for:/sys/devices/system/cpu/cpu*/power/energy_perf_biaswhere user has choice to write a predefined string or raw value.Quoted 4 lines
The change itself is trivial. When user preference doesn't match predefined string preferences and value is an unsigned integer and in
range, use that value for EPP. When the EPP feature is not prsent^^^^^^s/prsent/present
Quoted 170 lines
writing raw value is not supported.
Suggested-by Len Brown <[email protected]>Signed-off-by Srinivas Pandruvada <--- Documentation/admin-guide/pm/intel_pstate.rst | 6 ++- drivers/cpufreq/intel_pstate.c | 50 +++++++++++++++---- 2 files changed, 45 insertions(+), 11 deletions(-)Documentation/admin- renamed from Documentation/admin-guide/pm/intel_pstate.rst +0
-0
guide/pm/intel_pstate.rst
index939bfdc53f4f..5e209926e0ed100644Documentation/admin-guide/pm/intel_pstate.rst +0
-0@@ -561,7 +561,11 @@ somewhere between the two extremes: Strings written to the ``energy_performance_preference`` attributeare
internally translated to integer values written to the processor's Energy-Performance Preference (EPP) knob (if supported) or itsDocumentation/admin-guide/pm/intel_pstate.rst +1
-1-Energy-Performance Bias (EPB) knob. +Energy-Performance Bias (EPB) knob. It is also possible to write apositive
+integer value between 0 to 255, if the EPP feature is present. If the EPP
+feature is not present, writing integer value to this attribute is not
+supported. In this case, user can use
+ "/sys/devices/system/cpu/cpu*/power/energy_perf_bias" interface.[Note that tasks may by migrated from one CPU to another by the
scheduler's
load-balancing algorithm and if different energy vs performance
hints aredrivers/cpufreq/intel_pstate.c +2
-1@@ -602,11 +602,12 @@ static const unsigned int epp_values[] = {HWP_EPP_POWERSAVE }; -static int intel_pstate_get_energy_pref_index (struct cpudata *cpu_data) +static int intel_pstate_get_energy_pref_index (struct cpudata *cpu_data,int *raw_epp) {s16 epp;int index = -EINVAL ; + *raw_epp =0 ; epp =intel_pstate_get_epp (cpu_data,0 );drivers/cpufreq/intel_pstate.c +5
-3if (epp <0 )return epp;@@ -614,12 +615,14 @@ static int intel_pstate_get_energy_pref_index(struct cpudata *cpu_data) if (boot_cpu_has(X86_FEATURE_HWP_EPP)) {if (epp ==HWP_EPP_PERFORMANCE )return 1 ; -if (epp <=HWP_EPP_BALANCE_PERFORMANCE ) +if (epp ==HWP_EPP_BALANCE_PERFORMANCE )return 2 ; -if (epp <=HWP_EPP_BALANCE_POWERSAVE ) +if (epp ==HWP_EPP_BALANCE_POWERSAVE )return 3 ; -else +if (epp ==HWP_EPP_POWERSAVE )return 4 ; + *raw_epp = epp; +return 0 ; } else if (boot_cpu_has(X86_FEATURE_EPB)) {/* drivers/cpufreq/intel_pstate.c +2
-1* Range: @@ -638,7 +641,8 @@ static int intel_pstate_get_energy_pref_index(struct cpudata *cpu_data) }static int intel_pstate_set_energy_pref_index (struct cpudata *cpu_data, - int pref_index) +int pref_index,bool use_raw, +u32 raw_epp){
drivers/cpufreq/intel_pstate.c +10
-0int epp = -EINVAL ;int ret;@@ -657,6 +661,16 @@ static int intel_pstate_set_energy_pref_index(struct cpudata *cpu_data, value &= ~GENMASK_ULL(31, 24); +if (use_raw) { +if (raw_epp >255 ) { + ret = -EINVAL ; + goto return_pref; + } + value |= (u64 )raw_epp <<24 ; + ret =wrmsrl_on_cpu (cpu_data->cpu,MSR_HWP_REQUEST , value); + goto return_pref; + } +if (epp == -EINVAL )drivers/cpufreq/intel_pstate.c +0
-0epp = epp_values[pref_index -1 ];@@ -694,6 +708,8 @@ static ssize_tstore_energy_performance_preference(
{drivers/cpufreq/intel_pstate.c +2
-0struct cpudata *cpu_data = all_cpu_data[policy->cpu];char str_preference[21 ]; +bool raw = false; +u32 epp;int ret; ret =sscanf (buf,"%20s" , str_preference);@@ -701,10 +717,21 @@ static ssize_tstore_energy_performance_preference(
drivers/cpufreq/intel_pstate.c +12
-1return -EINVAL ; ret =match_string (energy_perf_strings,-1 , str_preference); -if (ret <0 ) +if (ret <0 ) { +if (!boot_cpu_has (X86_FEATURE_HWP_EPP )) +return ret; + + ret =kstrtouint (buf,10 , &epp); +if (ret) +return ret; + + raw = true; + } + + ret = intel_pstate_set_energy_pref_index(cpu_data, ret, raw,epp);
drivers/cpufreq/intel_pstate.c +1
-1+if (ret)return ret; -intel_pstate_set_energy_pref_index (cpu_data, ret);return count; }@@ -712,13 +739,16 @@ static ssize_tshow_energy_performance_preference(
structcpufreq_policy*policy, char
*buf)
{drivers/cpufreq/intel_pstate.c +2
-2struct cpudata *cpu_data = all_cpu_data[policy->cpu]; -int preference; +int preference, raw_epp; - preference =intel_pstate_get_energy_pref_index (cpu_data); + preference = intel_pstate_get_energy_pref_index(cpu_data,
&raw_epp);drivers/cpufreq/intel_pstate.c +4
-1if (preference <0 )return preference; -return sprintf (buf,"%s\n" , energy_perf_strings[preference]); +if (raw_epp) +return sprintf (buf,"%d\n" , raw_epp); +else + return sprintf(buf, "%s\n",energy_perf_strings[preference]);
}
cpufreq_freq_attr_rw(energy_performance_preference);
--
2.25.4