The RAPL Joules counter is limited in capacity.
Turbostat estimates how soon it can roll-over
based on the max TDP of the processor --
which tells us the maximum increment rate.
eg.
RAPL: 2759 sec. Joule Counter Range, at 95 Watts
So if a sample duration is longer than 2759 seconds on this system,
'**' replace the decimal place in the display to indicate
that the results may be suspect.
But the display had an extra ' ' in this case, throwing off the columns.
Also, the -J "Joules" option appended an extra "time" column
to the display. While this may be useful, it printed the interval time,
which may not be the accurate time per processor. Remove this column,
which appeared only when using '-J',
as we plan to add accurate per-cpu interval times in a future commit.
Signed-off-by: Len Brown <len.brown@intel.com>
Replace MSR_NHM_TURBO_RATIO_LIMIT with MSR_TURBO_RATIO_LIMIT.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Acked-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
When run
make -C tools DESTDIR=/my/nice/dir turbostat_install
get a message
install: cannot create regular file '/usr/bin/turbostat': Permission denied
Allow user to alter DESTDIR and PREFIX variables.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
* pm-core:
PM / wakeirq: fix wakeirq setting after wakup re-configuration from sysfs
PM / runtime: Document steps for device removal
* powercap:
powercap: intel_rapl: Add missing Haswell model
* pm-tools:
tools/power turbostat: work around RC6 counter wrap
tools/power turbostat: initial KBL support
tools/power turbostat: initial SKX support
tools/power turbostat: decode BXT TSC frequency via CPUID
tools/power turbostat: initial BXT support
tools/power turbostat: print IRTL MSRs
tools/power turbostat: SGX state should print only if --debug
Sometimes the rc6 sysfs counter spontaneously resets,
causing turbostat prints a very large number
as it tries to calcuate % = 100 * (old - new) / interval
When we see (old > new), print ***.**% instead
of a bogus huge number.
Note that this detection is not fool-proof, as the counter
could reset several times and still result in new > old.
Signed-off-by: Len Brown <len.brown@intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Hard-code BXT ART to 19200MHz, so turbostat --debug
can fully enumerate TSC:
CPUID(0x15): eax_crystal: 3 ebx_tsc: 186 ecx_crystal_hz: 0
TSC: 1190 MHz (19200000 Hz * 186 / 3 / 1000000)
Signed-off-by: Len Brown <len.brown@intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Some processors use the Interrupt Response Time Limit (IRTL) MSR value
to describe the maximum IRQ response time latency for deep
package C-states. (Though others have the register, but do not use it)
Lets print it out to give insight into the cases where it is used.
IRTL begain in SNB, with PC3/PC6/PC7, and HSW added PC8/PC9/PC10.
Signed-off-by: Len Brown <len.brown@intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
The CPUID.SGX bit was printed, even if --debug was used
Signed-off-by: Len Brown <len.brown@intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
- Redesign of cpufreq governors and the intel_pstate driver to
make them use callbacks invoked by the scheduler to trigger CPU
frequency evaluation instead of using per-CPU deferrable timers
for that purpose (Rafael Wysocki).
- Reorganization and cleanup of cpufreq governor code to make it
more straightforward and fix some concurrency problems in it
(Rafael Wysocki, Viresh Kumar).
- Cleanup and improvements of locking in the cpufreq core (Viresh
Kumar).
- Assorted cleanups in the cpufreq core (Rafael Wysocki, Viresh
Kumar, Eric Biggers).
- intel_pstate driver updates including fixes, optimizations and a
modification to make it enable enable hardware-coordinated P-state
selection (HWP) by default if supported by the processor (Philippe
Longepe, Srinivas Pandruvada, Rafael Wysocki, Viresh Kumar, Felipe
Franciosi).
- Operating Performance Points (OPP) framework updates to improve
its handling of voltage regulators and device clocks and updates
of the cpufreq-dt driver on top of that (Viresh Kumar, Jon Hunter).
- Updates of the powernv cpufreq driver to fix initialization
and cleanup problems in it and correct its worker thread handling
with respect to CPU offline, new powernv_throttle tracepoint
(Shilpasri Bhat).
- ACPI cpufreq driver optimization and cleanup (Rafael Wysocki).
- ACPICA updates including one fix for a regression introduced
by previos changes in the ACPICA code (Bob Moore, Lv Zheng,
David Box, Colin Ian King).
- Support for installing ACPI tables from initrd (Lv Zheng).
- Optimizations of the ACPI CPPC code (Prashanth Prakash, Ashwin
Chaugule).
- Support for _HID(ACPI0010) devices (ACPI processor containers)
and ACPI processor driver cleanups (Sudeep Holla).
- Support for ACPI-based enumeration of the AMBA bus (Graeme Gregory,
Aleksey Makarov).
- Modification of the ACPI PCI IRQ management code to make it treat
255 in the Interrupt Line register as "not connected" on x86 (as
per the specification) and avoid attempts to use that value as
a valid interrupt vector (Chen Fan).
- ACPI APEI fixes related to resource leaks (Josh Hunt).
- Removal of modularity from a few ACPI drivers (BGRT, GHES,
intel_pmic_crc) that cannot be built as modules in practice (Paul
Gortmaker).
- PNP framework update to make it treat ACPI_RESOURCE_TYPE_SERIAL_BUS
as a valid resource type (Harb Abdulhamid).
- New device ID (future AMD I2C controller) in the ACPI driver for
AMD SoCs (APD) and in the designware I2C driver (Xiangliang Yu).
- Assorted ACPI cleanups (Colin Ian King, Kaiyen Chang, Oleg Drokin).
- cpuidle menu governor optimization to avoid a square root
computation in it (Rasmus Villemoes).
- Fix for potential use-after-free in the generic device properties
framework (Heikki Krogerus).
- Updates of the generic power domains (genpd) framework including
support for multiple power states of a domain, fixes and debugfs
output improvements (Axel Haslam, Jon Hunter, Laurent Pinchart,
Geert Uytterhoeven).
- Intel RAPL power capping driver updates to reduce IPI overhead in
it (Jacob Pan).
- System suspend/hibernation code cleanups (Eric Biggers, Saurabh
Sengar).
- Year 2038 fix for the process freezer (Abhilash Jindal).
- turbostat utility updates including new features (decoding of more
registers and CPUID fields, sub-second intervals support, GFX MHz
and RC6 printout, --out command line option), fixes (syscall jitter
detection and workaround, reductioin of the number of syscalls made,
fixes related to Xeon x200 processors, compiler warning fixes) and
cleanups (Len Brown, Hubert Chrzaniuk, Chen Yu).
/
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.22 (GNU/Linux)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=cfty
-----END PGP SIGNATURE-----
Merge tag 'pm+acpi-4.6-rc1-1' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm
Pull power management and ACPI updates from Rafael Wysocki:
"This time the majority of changes go into cpufreq and they are
significant.
First off, the way CPU frequency updates are triggered is different
now. Instead of having to set up and manage a deferrable timer for
each CPU in the system to evaluate and possibly change its frequency
periodically, cpufreq governors set up callbacks to be invoked by the
scheduler on a regular basis (basically on utilization updates). The
"old" governors, "ondemand" and "conservative", still do all of their
work in process context (although that is triggered by the scheduler
now), but intel_pstate does it all in the callback invoked by the
scheduler with no need for any additional asynchronous processing.
Of course, this eliminates the overhead related to the management of
all those timers, but also it allows the cpufreq governor code to be
simplified quite a bit. On top of that, the common code and data
structures used by the "ondemand" and "conservative" governors are
cleaned up and made more straightforward and some long-standing and
quite annoying problems are addressed. In particular, the handling of
governor sysfs attributes is modified and the related locking becomes
more fine grained which allows some concurrency problems to be avoided
(particularly deadlocks with the core cpufreq code).
In principle, the new mechanism for triggering frequency updates
allows utilization information to be passed from the scheduler to
cpufreq. Although the current code doesn't make use of it, in the
works is a new cpufreq governor that will make decisions based on the
scheduler's utilization data. That should allow the scheduler and
cpufreq to work more closely together in the long run.
In addition to the core and governor changes, cpufreq drivers are
updated too. Fixes and optimizations go into intel_pstate, the
cpufreq-dt driver is updated on top of some modification in the
Operating Performance Points (OPP) framework and there are fixes and
other updates in the powernv cpufreq driver.
Apart from the cpufreq updates there is some new ACPICA material,
including a fix for a problem introduced by previous ACPICA updates,
and some less significant changes in the ACPI code, like CPPC code
optimizations, ACPI processor driver cleanups and support for loading
ACPI tables from initrd.
Also updated are the generic power domains framework, the Intel RAPL
power capping driver and the turbostat utility and we have a bunch of
traditional assorted fixes and cleanups.
Specifics:
- Redesign of cpufreq governors and the intel_pstate driver to make
them use callbacks invoked by the scheduler to trigger CPU
frequency evaluation instead of using per-CPU deferrable timers for
that purpose (Rafael Wysocki).
- Reorganization and cleanup of cpufreq governor code to make it more
straightforward and fix some concurrency problems in it (Rafael
Wysocki, Viresh Kumar).
- Cleanup and improvements of locking in the cpufreq core (Viresh
Kumar).
- Assorted cleanups in the cpufreq core (Rafael Wysocki, Viresh
Kumar, Eric Biggers).
- intel_pstate driver updates including fixes, optimizations and a
modification to make it enable enable hardware-coordinated P-state
selection (HWP) by default if supported by the processor (Philippe
Longepe, Srinivas Pandruvada, Rafael Wysocki, Viresh Kumar, Felipe
Franciosi).
- Operating Performance Points (OPP) framework updates to improve its
handling of voltage regulators and device clocks and updates of the
cpufreq-dt driver on top of that (Viresh Kumar, Jon Hunter).
- Updates of the powernv cpufreq driver to fix initialization and
cleanup problems in it and correct its worker thread handling with
respect to CPU offline, new powernv_throttle tracepoint (Shilpasri
Bhat).
- ACPI cpufreq driver optimization and cleanup (Rafael Wysocki).
- ACPICA updates including one fix for a regression introduced by
previos changes in the ACPICA code (Bob Moore, Lv Zheng, David Box,
Colin Ian King).
- Support for installing ACPI tables from initrd (Lv Zheng).
- Optimizations of the ACPI CPPC code (Prashanth Prakash, Ashwin
Chaugule).
- Support for _HID(ACPI0010) devices (ACPI processor containers) and
ACPI processor driver cleanups (Sudeep Holla).
- Support for ACPI-based enumeration of the AMBA bus (Graeme Gregory,
Aleksey Makarov).
- Modification of the ACPI PCI IRQ management code to make it treat
255 in the Interrupt Line register as "not connected" on x86 (as
per the specification) and avoid attempts to use that value as a
valid interrupt vector (Chen Fan).
- ACPI APEI fixes related to resource leaks (Josh Hunt).
- Removal of modularity from a few ACPI drivers (BGRT, GHES,
intel_pmic_crc) that cannot be built as modules in practice (Paul
Gortmaker).
- PNP framework update to make it treat ACPI_RESOURCE_TYPE_SERIAL_BUS
as a valid resource type (Harb Abdulhamid).
- New device ID (future AMD I2C controller) in the ACPI driver for
AMD SoCs (APD) and in the designware I2C driver (Xiangliang Yu).
- Assorted ACPI cleanups (Colin Ian King, Kaiyen Chang, Oleg Drokin).
- cpuidle menu governor optimization to avoid a square root
computation in it (Rasmus Villemoes).
- Fix for potential use-after-free in the generic device properties
framework (Heikki Krogerus).
- Updates of the generic power domains (genpd) framework including
support for multiple power states of a domain, fixes and debugfs
output improvements (Axel Haslam, Jon Hunter, Laurent Pinchart,
Geert Uytterhoeven).
- Intel RAPL power capping driver updates to reduce IPI overhead in
it (Jacob Pan).
- System suspend/hibernation code cleanups (Eric Biggers, Saurabh
Sengar).
- Year 2038 fix for the process freezer (Abhilash Jindal).
- turbostat utility updates including new features (decoding of more
registers and CPUID fields, sub-second intervals support, GFX MHz
and RC6 printout, --out command line option), fixes (syscall jitter
detection and workaround, reductioin of the number of syscalls
made, fixes related to Xeon x200 processors, compiler warning
fixes) and cleanups (Len Brown, Hubert Chrzaniuk, Chen Yu)"
* tag 'pm+acpi-4.6-rc1-1' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm: (182 commits)
tools/power turbostat: bugfix: TDP MSRs print bits fixing
tools/power turbostat: correct output for MSR_NHM_SNB_PKG_CST_CFG_CTL dump
tools/power turbostat: call __cpuid() instead of __get_cpuid()
tools/power turbostat: indicate SMX and SGX support
tools/power turbostat: detect and work around syscall jitter
tools/power turbostat: show GFX%rc6
tools/power turbostat: show GFXMHz
tools/power turbostat: show IRQs per CPU
tools/power turbostat: make fewer systems calls
tools/power turbostat: fix compiler warnings
tools/power turbostat: add --out option for saving output in a file
tools/power turbostat: re-name "%Busy" field to "Busy%"
tools/power turbostat: Intel Xeon x200: fix turbo-ratio decoding
tools/power turbostat: Intel Xeon x200: fix erroneous bclk value
tools/power turbostat: allow sub-sec intervals
ACPI / APEI: ERST: Fixed leaked resources in erst_init
ACPI / APEI: Fix leaked resources
intel_pstate: Do not skip samples partially
intel_pstate: Remove freq calculation from intel_pstate_calc_busy()
intel_pstate: Move intel_pstate_calc_busy() into get_target_pstate_use_performance()
...
MSR_CONFIG_TDP_NOMINAL:
should print all 8 bits of base_ratio (bit 0:7) 0xFF
MSR_CONFIG_TDP_LEVEL_1:
should print all 15 bits of PKG_MIN_PWR_LVL1 (bit 48:62) 0x7FFF
should print all 15 bits of PKG_MAX_PWR_LVL1 (bit 32:46) 0x7FFF
should print all 8 bits of LVL1_RATIO (bit 16:23) 0xFF
should print all 15 bits of PKG_TDP_LVL1 (bit 0:14) 0x7FFF
And the same modification to MSR_CONFIG_TDP_LEVEL_2.
MSR_TURBO_ACTIVATION_RATIO:
should print all 8 bits of MAX_NON_TURBO_RATIO (bit 0:7) 0xFF
Signed-off-by: Chen Yu <yu.c.chen@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
MSR_NHM_SNB_PKG_CST_CFG_CTL: 0x1e008008 (...pkg-cstate-limit=0: unlimited)
should print as
MSR_NHM_SNB_PKG_CST_CFG_CTL: 0x1e008008 (...pkg-cstate-limit=8: unlimited)
Signed-off-by: Len Brown <len.brown@intel.com>
turbostat already checks whether calling each cpuid leavf is legal,
and it doesn't look at the function return value,
so call the simpler gcc intrinsic __cpuid() instead of __get_cpuid().
syntax only, no functional change
Signed-off-by: Len Brown <len.brown@intel.com>
The accuracy of Bzy_Mhz and Busy% depend on reading
the TSC, APERF, and MPERF close together in time.
When there is a very short measurement interval,
or a large system is profoundly idle, the changes
in APERF and MPERF may be very small.
They can be small enough that an expensive interrupt
between reading APERF and MPERF can cause the APERF/MPERF
ratio to become inaccurate, resulting in invalid
calculation and display of Bzy_MHz.
A dummy APERF read of APERF makes this problem
much more rare. Apparently this 1st systemn call
after exiting a long stretch of idle is when we
typically see expensive timer interrupts that cause
large jitter.
For the cases that dummy APERF read fails to prevent,
we compare the latency of the APERF and MPERF reads.
If they differ by more than 2x, we re-issue them.
Signed-off-by: Len Brown <len.brown@intel.com>
The column "GFX%c6" show the percentage of time the GPU
is in the "render C6" state, rc6. Deep package C-states on several
systems depend on the GPU being in RC6.
This information comes from the counter
/sys/class/drm/card0/power/rc6_residency_ms,
as read before and after the measurement interval.
Signed-off-by: Len Brown <len.brown@intel.com>
Under the column "GFXMHz", show a snapshot of this attribute:
/sys/class/graphics/fb0/device/drm/card0/gt_cur_freq_mhz
This is an instantaneous snapshot of what sysfs presents
at the end of the measurement interval. turbostat does
not average or otherwise perform any math on this value.
Signed-off-by: Len Brown <len.brown@intel.com>
The new IRQ column shows how many interrupts have occurred on each CPU
during the measurement inteval. This information comes from
the difference between /proc/interrupts shapshots made before
and after the measurement interval.
The first row, the system summary, shows the sum of the IRQS
for all CPUs during that interval.
Signed-off-by: Len Brown <len.brown@intel.com>
skip the open(2)/close(2) on each msr read
by keeping the /dev/cpu/*/msr files open.
The remaining read(2) is generally far fewer cycles
than the removed open(2) system call.
Signed-off-by: Len Brown <len.brown@intel.com>
By default...
Turbostat --debug gconfiguration info goes to stderr.
In FORK mode, turbostat statistics go to stderr.
In PERIODIC mode, turbostat statistics go to stdout.
These defaults do not change, but an option "--out file"
will send all output above only to the specified file.
Signed-off-by: Len Brown <len.brown@intel.com>
some tools processing turbostat output
have difficulty with items that begin with %...
Reported-by: Jacob Pan <jacob.jun.pan@linux.intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Following changes have been made:
- changed MSR_NHM_TURBO_RATIO_LIMIT to MSR_TURBO_RATIO_LIMIT in debug print
for consistency with Developer Manual
- updated definition of bitfields in MSR_TURBO_RATIO_LIMIT and appropriate
parsing code
- added x200 to list of architectures that do not support Nahlem compatible
definition of MSR_TURBO_RATIO_LIMIT register (x200 has the register but
bits definition is custom)
- fixed typo in code that parses MSR_TURBO_RATIO_LIMIT
(logical instead of bitwise operator)
- changed MSR_TURBO_RATIO_LIMIT parsing algorithm so the print out had the
same order as implementations for other platforms
Signed-off-by: Hubert Chrzaniuk <hubert.chrzaniuk@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
x200 does not enable any way to programmatically obtain bus clock
speed. Bclk for the architecture has a fixed value of 100 MHz.
At the same time x200 cannot be included in has_snb_msrs since
it does not support C7 idle state.
prior to this patch, MHz values reported on this chip
were erroneously calculated using bclk of 133MHz,
causing MHz values to be reported 33% higher than actual.
Signed-off-by: Hubert Chrzaniuk <hubert.chrzaniuk@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
turbostat -i interval_sec
will sample and display statistics every interval_sec.
interval_sec used to be a whole number of seconds,
but now we accept a decimal, as small as 0.001 sec (1 ms).
Signed-off-by: Len Brown <len.brown@intel.com>
When building with gcc 6 we're getting various build warnings that just
require some trivial function declaration and call fixes:
turbostat.c: In function ‘dump_cstate_pstate_config_info’:
turbostat.c:1973:1: warning: type of ‘family’ defaults to ‘int’
dump_cstate_pstate_config_info(family, model)
turbostat.c:1973:1: warning: type of ‘model’ defaults to ‘int’
turbostat.c: In function ‘get_tdp’:
turbostat.c:2145:8: warning: type of ‘model’ defaults to ‘int’
double get_tdp(model)
turbostat.c: In function ‘perf_limit_reasons_probe’:
turbostat.c:2259:6: warning: type of ‘family’ defaults to ‘int’
void perf_limit_reasons_probe(family, model)
turbostat.c:2259:6: warning: type of ‘model’ defaults to ‘int’
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Cc: Matt Fleming <matt@codeblueprint.co.uk>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/n/tip-wbicer8n0s9qe6ql8h9x478e@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
for debugging, dump a few more fields:
CPUID(1): SSE3 MONITOR EIST TM2 TSC MSR ACPI-TM TM
cpu0: MSR_IA32_MISC_ENABLE: 0x00850089 (TCC EIST MONITOR)
Signed-off-by: Len Brown <len.brown@intel.com>
MSR_PLATFORM_INFO is the new name for MSR_NHM_PLATFORM_INFO
no functional change
Signed-off-by: Len Brown <len.brown@intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
MSR_TURBO_ACTIVATION_RATIO: 0x00000016 (MAX_NON_TURBO_RATIO=6 lock=0)
should print all 7 bits of MAX_NON_TURBO_RATIO (in decimal):
MSR_TURBO_ACTIVATION_RATIO: 0x00000016 (MAX_NON_TURBO_RATIO=22 lock=0)
Signed-off-by: Len Brown <len.brown@intel.com>
Bzy_MHz = TSC_delta*tsc_tweak/APERF_delta/MPERF_delta/measurement_interval
becomes
Bzy_MHz = base_mhz/APERF_delta/MPERF_delta
on systems which support MSR_NHM_PLATFORM_INFO.
base_mhz is calculated directly from the base_ratio
reported in MSR_NHM_PLATFORM_INFO * bclk,
and bclk is discovered via MSR or cpuid.
This reduces the dependency of Bzy_MHz calculation on the TSC.
Previously, there were 4 TSC readings required in each caculation,
the raw TSC delta combined with the measurement_interval.
This also removes the "tsc_tweak" correction factor used when
TSC runs on a different base clock from the CPU's bclk.
After this change, tsc_tweak is used only for %Busy.
The end-result should be a Bzy_MHz result slightly less prone to jitter.
Signed-off-by: Len Brown <len.brown@intel.com>
On a Skylake with 1500MHz base frequency,
the TSC runs at 1512MHz.
This is because the TSC is no longer in the n*100 MHz BCLK domain,
but is now in the m*24MHz crystal clock domain. (24 MHz * 63 = 1512 MHz)
This adds error to several calculations in turbostat,
unless the TSC sample sizes are adjusted for this difference.
Note that calculations in the time domain are immune
from this issue, as the timing sub-system has already
calibrated the TSC against a known wall clock.
AVG_MHz = APERF_delta/measurement_interval
need no adjustment. APERF_delta is in the BCLK domain,
and measurement_interval is in the time domain.
TSC_MHz = TSC_delta/measurement_interval
needs no adjustment -- as we really do want to report
the actual measured TSC delta here, and measurement_interval
is in the accurate time domain.
%Busy = MPERF_delta/TSC_delta
needs adjustment to use TSC_BCLK_DOMAIN_delta.
TSC_BCLK_DOMAIN_delta = TSC_delta * base_hz / tsc_hz
Bzy_MHz = TSC_delta/APERF_delta/MPERF_delta/measurement_interval
need adjustment as above.
No other metrics in turbostat need to be adjusted.
Before:
CPU Avg_MHz %Busy Bzy_MHz TSC_MHz
- 550 24.84 2216 1512
0 2191 98.73 2219 1514
2 0 0.01 2130 1512
1 9 0.43 2016 1512
3 2 0.08 2016 1512
After:
CPU Avg_MHz %Busy Bzy_MHz TSC_MHz
- 550 25.05 2198 1512
0 2190 99.62 2199 1512
2 0 0.01 2152 1512
1 9 0.46 2000 1512
3 2 0.10 2000 1512
Note that in this example, the "Before" Bzy_MHz
was reported as exceeding the 2200 max turbo rate.
Also, even a pinned spin loop would not be reported
as over 99% busy.
Signed-off-by: Len Brown <len.brown@intel.com>
KNL increments APERF and MPERF every 1024 clocks.
This is compliant with the architecture specification,
which requires that only the ratio of APERF/MPERF need be valid.
However, turbostat takes advantage of the fact that these
two MSRs increment every un-halted clock
at the actual and base frequency:
AVG_MHz = APERF_delta/measurement_interval
%Busy = MPERF_delta/TSC_delta
This quirk is needed for these calculations to also work on KNL,
which would otherwise show a value 1024x smaller than expected.
Signed-off-by: Hubert Chrzaniuk <hubert.chrzaniuk@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Staring in Linux-4.3-rc1,
commit 6fb3143b56 ("tools/power turbostat: dump CONFIG_TDP")
touches MSR 0x648, which is not supported on IVB-Xeon.
This results in "turbostat --debug" exiting on those systems:
turbostat: /dev/cpu/2/msr offset 0x648 read failed: Input/output error
Remove IVB-Xeon from the list of machines supporting with that MSR.
Signed-off-by: Len Brown <len.brown@intel.com>
Pull turbostat changes for v4.3 from Len Brown.
* 'turbostat' of https://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux:
tools/power turbostat: fix typo on DRAM column in Joules-mode
tools/power turbostat: fix parameter passing for forked command
tools/power turbostat: dump CONFIG_TDP
tools/power turbostat: cpu0 is no longer hard-coded, so update output
tools/power turbostat: update turbostat(8)
turbostat supports forked command when sampling cpu state. However,
the forked command is not allowed to be executed with options, otherwise
turbostat might regard these options as invalid turbostat options.
For example:
./turbostat stress -c 4 -t 10
./turbostat: unrecognized option '-t'
Reported-by: Chen Yu <yu.c.chen@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Config TDP is a feature that allows parts to be configured
for different thermal limits after they have left the factory.
This can have an effect on the operation of the part,
particularly in determiniing...
Max Non-turbo Ratio
Turbo Activation Ratio
Signed-off-by: Len Brown <len.brown@intel.com>
The --debug option reads a number of per-package MSRs.
Previously we explicitly read them on cpu0, but recently
turbostat changed to read them on the current "base_cpu".
Update the print-out to reflect base_cpu, rather than
the hard-coded cpu0.
Signed-off-by: Len Brown <len.brown@intel.com>
This header containing all MSRs and respective bit definitions
got exported to userspace in conjunction with the big UAPI
shuffle.
But, it doesn't belong in the UAPI headers because userspace can
do its own MSR defines and exporting them from the kernel blocks
us from doing cleanups/renames in that header. Which is
ridiculous - it is not kernel's job to export such a header and
keep MSRs list and their names stable.
Signed-off-by: Borislav Petkov <bp@suse.de>
Acked-by: H. Peter Anvin <hpa@zytor.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: David Howells <dhowells@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/1433436928-31903-19-git-send-email-bp@alien8.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Remove reference to the original Nehalem Turbo white paper,
since it has moved, and these mechanisms have now long since
been documented in the Software Developer's Manual.
Reported-by: Jeremie Lagraviere <jeremie@simula.no>
Signed-off-by: Len Brown <len.brown@intel.com>
Linux-3.7 added CONFIG_BOOTPARAM_HOTPLUG_CPU0,
allowing systems to offline cpu0.
But when cpu0 is offline, turbostat will not run:
# turbostat ls
turbostat: no /dev/cpu/0/msr
This patch replaces the hard-coded use of cpu0 in turbostat
with the current cpu, allowing it to run without a cpu0.
Fewer cross-calls may also be needed due to use of current cpu,
though this hard-coding was used only for the --debug preamble.
Signed-off-by: Prarit Bhargava <prarit@redhat.com>
Signed-off-by: Len Brown <len.brown@intel.com>
When EPB is 0xF, turbosat was incorrectly describing it as "custom"
instead of calling it "powersave":
< cpu0: MSR_IA32_ENERGY_PERF_BIAS: 0x0000000f (custom)
> cpu0: MSR_IA32_ENERGY_PERF_BIAS: 0x0000000f (powersave)
Signed-off-by: Len Brown <len.brown@intel.com>
Changes mainly to account for minor differences in Knights Landing(KNL):
1. KNL supports C1 and C6 core states.
2. KNL supports PC2, PC3 and PC6 package states.
3. KNL has a different encoding of the TURBO_RATIO_LIMIT MSR
Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Without this update, turbostat displays only 2 threads per core.
Some processors, such as Xeon Phi, have more.
Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
HSW expanded MSR_PKG_CST_CONFIG_CONTROL.Package-C-State-Limit,
from bits[2:0] used by previous implementations, to [3:0].
The value 1000b is unlimited, and is used by BDW and SKL too.
Signed-off-by: Len Brown <len.brown@intel.com>
While not yet documented in the Software Developer's Manual,
the data-sheet for modern Xeon states that DRAM RAPL ENERGY units
are fixed at 15.3 uJ, rather than being discovered via MSR.
Before this patch, DRAM energy on these products is over-stated by turbostat
because the RAPL units are 4x larger.
ref: "Xeon E5-2600 v3/E5-1600 v3 Datasheet Volume 2"
http://www.intel.com/content/dam/www/public/us/en/documents/datasheets/xeon-e5-v3-datasheet-vol-2.pdf
Signed-off-by: Andrey Semin <andrey.semin@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Skylake adds some additional residency counters.
Skylake supports a different mix of RAPL registers
from any previous product.
In most other ways, Skylake is like Broadwell.
Signed-off-by: Len Brown <len.brown@intel.com>
Since commit ee0778a301
("tools/power: turbostat: make Makefile a bit more capable")
turbostat's Makefile is using
[...]
BUILD_OUTPUT := $(PWD)
[...]
which obviously causes trouble when building "turbostat" with
make -C /usr/src/linux/tools/power/x86/turbostat ARCH=x86 turbostat
because GNU make does not update nor guarantee that $PWD is set.
This patch changes the Makefile to use $CURDIR instead, which GNU make
guarantees to set and update (i.e. when using "make -C ...") and also
adds support for the O= option (see "make help" in your root of your
kernel source tree for more details).
Link: https://bugs.gentoo.org/show_bug.cgi?id=533918
Fixes: ee0778a301 ("tools/power: turbostat: make Makefile a bit more capable")
Signed-off-by: Thomas D. <whissi@whissi.de>
Cc: Mark Asselstine <mark.asselstine@windriver.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Some distros (Ubuntu) ship the msr driver as a module.
If turbosat is run as root on those systems, and discovers
that there is no /dev/cpu/cpu0/msr, it will now "modprobe msr"
for the user.
If not root, the modprobe attempt will fail, and turbostat will exit as before:
turbostat: no /dev/cpu/0/msr, Try "# modprobe msr" : No such file or directory
Signed-off-by: Len Brown <len.brown@intel.com>
s/MSR_NHM_TURBO_RATIO_LIMIT/MSR_TURBO_RATIO_LIMIT/
s/MSR_IVT_TURBO_RATIO_LIMIT/MSR_TURBO_RATIO_LIMIT1/
syntax only -- use the documented strings describing these registers.
Signed-off-by: Len Brown <len.brown@intel.com>
syntax only.
The cool kids are now using the phrase "base frequency",
where in the past we used "max non-turbo frequency" or "TSC frequency".
This distinction becomes important when a processor has a TSC
that runs at a different speed than the "base frequency".
Signed-off-by: Len Brown <len.brown@intel.com>
cosmetic only.
order the decoding of MSR_PERF_LIMIT_REASONS bits
from MSB to LSB -- which you notice when more than 1 bit is set
and you are, say, comparing the output to the documentation...
Signed-off-by: Len Brown <len.brown@intel.com>
Casual turbostat users generally just want to know MHz.
So by default, just print enough information to make sense of MHz.
All the other configuration data and columns for C-states and temperature etc,
are printed with the --debug option.
Signed-off-by: Len Brown <len.brown@intel.com>
Long format options added, though the short ones should still work.
eg. the new "--Counter 0x10" is the same as the old "-C 0x10"
Note this Incompatibility:
Old:
-v displayed verbose debug output
New:
-v and --version simpaly display version
Additional parameters:
-d and --debug display verbose debug output
-h and --help display a help message
Updated turbosat.8 man page accordingly.
Signed-off-by: Len Brown <len.brown@intel.com>
Replaced previously open-coded Package C-state Limit decoding
with table-driven decoding. In doing so, updated to match January 2015
"Intel(R) 64 and IA-23 Architectures Software Developer's Manual".
In the past, turbostat would print package C-state residency columns
for all package states supported by the model's architecture, even though
a particular SKU may not support them, or they may be disabled by the BIOS.
Now turbostat will skip printing colunns if MSRs indicate that they are not enabled.
eg. many SKUs don't support PC7, and so that column will no longer be printed.
Signed-off-by: Len Brown <len.brown@intel.com>
While turbostat is significantly less useful on systems
with no APERF_MSR, it seems more friendly
to run on such systems and report what we can,
rather than refusing to run.
Update man page to reflect recent changes.
Signed-off-by: Len Brown <len.brown@intel.com>
Turbostat can be useful on systems that do not support invariant TSC,
so allow it to run on those systgems.
All arithmetic in turbostat using the TSC value is per-processsor,
so it does not depend on the TSC values being in sync acrosss processors.
Turbostat uses gettimeofday() for the measurement interval
rather than using the TSC directly, so that key metric
is also immune from variable TSC.
Turbostat prints a TSC sanity check column:
TSC_MHz = TSC_delta/interval
If this column is constant and is close to the processor
base frequency, then the TSC is behaving properly.
The other key turbostat columns are calculated this way:
Avg_Mhz = APERF_delta/interval
%Busy = MPERF_delta/TSC_delta
Bzy_MHz = TSC_delta/APERF_delta/MPERF_delta/interval
Tested on Core2 and Core2-Xeon, and so this patch includes
a few other changes to remove the assumption that target
systems are Nehalem and newer.
Signed-off-by: Len Brown <len.brown@intel.com>
The Processor generation code-named Haswell
added MSR_{CORE | GFX | RING}_PERF_LIMIT_REASONS
to explain when and how the processor limits frequency.
turbostat -v
will now decode these bits.
Each MSR has an "Active" set of bits which describe
current conditions, and a "Logged" set of bits,
which describe what has happened since last cleared.
Turbostat currently doesn't clear the log bits.
Signed-off-by: Len Brown <len.brown@intel.com>
For turbostat to run as non-root, it needs to permissions:
1. read access to /dev/cpu/*/msr
via standard user/group/world file permissions
2. CAP_SYS_RAWIO
eg. # setcap cap_sys_rawio=ep turbostat
Yes, running as root still works.
Signed-off-by: Len Brown <len.brown@intel.com>
The Intel 64 and IA-32 Architectures Software Developer's Manual says
that TjMax is stored in bits 23:16 of MSR_TEMPERATURE TARGET (0x1a2).
That's 8 bits, not 7, so it must be masked with 0xFF rather than 0x7F.
The manual has no mention of which values should be considered valid,
which kind of implies that they all are. Arbitrarily discarding values
outside a specific range is wrong. The upper range check had to be
fixed recently (commit 144b44b1) and the lower range check is just as
wrong. See bug #75071:
https://bugzilla.kernel.org/show_bug.cgi?id=75071
There are many Xeon processor series with TjMax of 70, 71 or 80
degrees Celsius, way below the arbitrary 85 degrees Celsius limit.
There may be other (past or future) models with even lower limits.
So drop this arbitrary check. The only value that would be clearly
invalid is 0. Everything else should be accepted.
After these changes, turbostat is aligned with what the coretemp
driver does.
Signed-off-by: Jean Delvare <jdelvare@suse.de>
Cc: Len Brown <len.brown@intel.com>
Acked-by: Guenter Roeck <linux@roeck-us.net>
Reviewed-by: Josh Triplett <josh@joshtriplett.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Use 8 columns for each number ouput.
We don't fit into 80 columns on most machines,
so keep the format simple.
Print frequency in MHz instead of GHz.
We've got 8 columns now, so use them to
show low frequency in a more natural unit.
Many users didn't understand what %c0 meant,
so re-name it to be %Busy.
Add Avg_MHz column, which is the frequency that many
users expect to see -- the total number of cycles executed
over the measurement interval.
People found the previous GHz to be confusing, since
it was the speed only over the non-idle interval.
That measurement has been re-named Bzy_MHz.
Suggested-by: Dirk J. Brandewie
Signed-off-by: Len Brown <len.brown@intel.com>
The new option allows just run turbostat and get dump of counter values. It's
useful when we have something more than one program to test.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
The -s is not used, let's remove it, and update quick help accordingly.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Add "-J" option to report energy consumed in joules per sample. This option
also adds the sample time to the reported values.
Signed-off-by: Dirk Brandewie <dirk.j.brandewie@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Haswell Xeon has slightly different RAPL support than client HSW,
which prevented the previous version of turbostat from running on HSX.
Signed-off-by: Len Brown <len.brown@intel.com>
Most of turbostat's error handling consists of printing an error (often
including an errno) and exiting. Since perror doesn't support a format
string, those error messages are often ambiguous, such as just showing a
file path, which doesn't uniquely identify which call failed.
turbostat already uses _GNU_SOURCE, so switch to the err and errx
functions from err.h, which take a format string.
Signed-off-by: Josh Triplett <josh@joshtriplett.org>
Signed-off-by: Len Brown <len.brown@intel.com>
Several different functions in turbostat contain the same pattern of
opening a file and exiting on failure. Factor out a common fopen_or_die
function for that.
Signed-off-by: Josh Triplett <josh@joshtriplett.org>
Signed-off-by: Len Brown <len.brown@intel.com>
Many different chunks of code in turbostat open a file, parse a single
int out of it, and close it. Factor that out into a common function.
Signed-off-by: Josh Triplett <josh@joshtriplett.org>
Signed-off-by: Len Brown <len.brown@intel.com>
Some systems declare fscanf with the warn_unused_result attribute. On
such systems, turbostat generates the following warnings:
turbostat.c: In function 'get_core_id':
turbostat.c:1203:8: warning: ignoring return value of 'fscanf', declared with attribute warn_unused_result [-Wunused-result]
turbostat.c: In function 'get_physical_package_id':
turbostat.c:1186:8: warning: ignoring return value of 'fscanf', declared with attribute warn_unused_result [-Wunused-result]
turbostat.c: In function 'cpu_is_first_core_in_package':
turbostat.c:1169:8: warning: ignoring return value of 'fscanf', declared with attribute warn_unused_result [-Wunused-result]
turbostat.c: In function 'cpu_is_first_sibling_in_core':
turbostat.c:1148:8: warning: ignoring return value of 'fscanf', declared with attribute warn_unused_result [-Wunused-result]
Fix these by checking the return value of those four calls to fscanf.
Signed-off-by: Josh Triplett <josh@joshtriplett.org>
Signed-off-by: Len Brown <len.brown@intel.com>
turbostat uses inline assembly to call cpuid. On 32-bit x86, on systems
that have certain security features enabled by default that make -fPIC
the default, this causes a build error:
turbostat.c: In function ‘check_cpuid’:
turbostat.c:1906:2: error: PIC register clobbered by ‘ebx’ in ‘asm’
asm("cpuid" : "=a" (fms), "=c" (ecx), "=d" (edx) : "a" (1) : "ebx");
^
GCC provides a header cpuid.h, containing a __get_cpuid function that
works with both PIC and non-PIC. (On PIC, it saves and restores ebx
around the cpuid instruction.) Use that instead.
Signed-off-by: Josh Triplett <josh@joshtriplett.org>
Cc: stable@vger.kernel.org
Signed-off-by: Len Brown <len.brown@intel.com>
turbostat uses the format %zx to print an off_t. However, %zx wants a
size_t, not an off_t. On 32-bit targets, those refer to different
types, potentially even with different sizes. Use %llx and a cast
instead, since printf does not have a length modifier for off_t.
Without this patch, when compiling for a 32-bit target:
turbostat.c: In function 'get_msr':
turbostat.c:231:3: warning: format '%zx' expects argument of type 'size_t', but argument 4 has type 'off_t' [-Wformat]
Signed-off-by: Josh Triplett <josh@joshtriplett.org>
Signed-off-by: Len Brown <len.brown@intel.com>
turbostat's Makefile puts arch/x86/include/uapi/ in the include path, so
that it can include <asm/msr.h> from it. It isn't in general safe to
include even uapi headers directly from the kernel tree without
processing them through scripts/headers_install.sh, but asm/msr.h
happens to work.
However, that include path can break with some versions of system
headers, by overriding some system headers with the unprocessed versions
directly from the kernel source. For instance:
In file included from /build/x86-generic/usr/include/bits/sigcontext.h:28:0,
from /build/x86-generic/usr/include/signal.h:339,
from /build/x86-generic/usr/include/sys/wait.h:31,
from turbostat.c:27:
../../../../arch/x86/include/uapi/asm/sigcontext.h:4:28: fatal error: linux/compiler.h: No such file or directory
This occurs because the system bits/sigcontext.h on that build system
includes <asm/sigcontext.h>, and asm/sigcontext.h in the kernel source
includes <linux/compiler.h>, which scripts/headers_install.sh would have
filtered out.
Since turbostat really only wants a single header, just include that one
header rather than putting an entire directory of kernel headers on the
include path.
In the process, switch from msr.h to msr-index.h, since turbostat just
wants the MSR numbers.
Signed-off-by: Josh Triplett <josh@joshtriplett.org>
Cc: stable@vger.kernel.org
Signed-off-by: Len Brown <len.brown@intel.com>
Support the next generation Intel Atom processor
mirco-architecture, formerly called Silvermont.
The server version, formerly called "Avoton",
is named the "Intel(R) Atom(TM) Processor C2000 Product Family".
The client version, formerly called "Bay Trail",
is named the "Intel Atom Processor Z3000 Series",
as well as various "Intel Pentium Processor"
and "Intel Celeron Processor" brands, depending
on form-factor.
Silvermont has a set of MSRs not far off from NHM,
but the RAPL register set is a sub-set of those previously supported.
Signed-off-by: Len Brown <len.brown@intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
On platforms with C8-C10 support, the additional C-states cause
turbostat to overrun its output buffer of 128 bytes per CPU. Increase
this to 256 bytes per CPU.
[ As a bugfix, this should go into 3.10; however, since the C8-C10
support didn't go in until after 3.9, this need not go into any stable
kernel. ]
Signed-off-by: Josh Triplett <josh@joshtriplett.org>
Cc: Len Brown <len.brown@intel.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Display residency in the new C-states, C8, C9, C10.
C8, C9, C10 are present on some:
"Fourth Generation Intel(R) Core(TM) Processors",
which are based on Intel(R) microarchitecture code name Haswell.
Signed-off-by: Kristen Carlson Accardi <kristen@linux.intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
The SMI counter is popular -- so display it by default
rather than requiring an option. What the heck,
we've blown the 80 column budget on many systems already...
Note that the value displayed is the delta
during the measurement interval.
The absolute value of the counter can still be seen with
the generic 32-bit MSR option, ie. -m 0x34
Signed-off-by: Len Brown <len.brown@intel.com>
When verbose is enabled, print the C1E-Enable
bit in MSR_IA32_POWER_CTL.
also delete some redundant tests on the verbose variable.
Signed-off-by: Len Brown <len.brown@intel.com>
This patch enables turbostat to run properly on the
next-generation Intel(R) Microarchitecture, code named "Haswell" (HSW).
HSW supports the BCLK and counters found in SNB.
Signed-off-by: Len Brown <len.brown@intel.com>
Pull powertool update from Len Brown:
"This updates the tree w/ the latest version of turbostat, which
reports temperature and - on SNB and later - Watts."
Fix up semantic merge conflict as per Len.
* 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux:
tools: Allow tools to be installed in a user specified location
tools/power: turbostat: make Makefile a bit more capable
tools/power x86_energy_perf_policy: close /proc/stat in for_every_cpu()
tools/power turbostat: v3.0: monitor Watts and Temperature
tools/power turbostat: fix output buffering issue
tools/power turbostat: prevent infinite loop on migration error path
x86 power: define RAPL MSRs
tools/power/x86/turbostat: share kernel MSR #defines
When building x86_energy_perf_policy or turbostat within the confines of
a packaging system such as RPM, we need to be able to have it install to
the buildroot and not the root filesystem of the build machine. This
adds a DESTDIR variable that when set will act as a prefix for the
install location of these tools.
Signed-off-by: Josh Boyer <jwboyer@redhat.com>
Signed-off-by: Len Brown <len.brown@intel.com>
The turbostat Makefile is pretty simple, its output is placed in the
same directory as the source, the install rule has no concept of a
prefix or sysroot, and you can set CC to use a specific compiler but
not use the more familiar CROSS_COMPILE. By making a few minor changes
these limitations are removed while leaving the default behavior
matching what it used to be.
Example build with these changes:
make CROSS_COMPILE=i686-wrs-linux-gnu- DESTDIR=/tmp install
or from the tools directory
make CROSS_COMPILE=i686-wrs-linux-gnu- DESTDIR=/tmp turbostat_install
Signed-off-by: Mark Asselstine <mark.asselstine@windriver.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Instead of returning out of for_every_cpu() we should break out of the loop=
which will then tidy up correctly by closing the file /proc/stat.
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Show power in Watts and temperature in Celsius
when hardware support is present.
Intel's Sandy Bridge and Ivy Bridge processor generations support RAPL
(Run-Time-Average-Power-Limiting). Per the Intel SDM
(Intel® 64 and IA-32 Architectures Software Developer Manual)
RAPL provides hardware energy counters and power control MSRs
(Model Specific Registers). RAPL MSRs are designed primarily
as a method to implement power capping. However, they are useful
for monitoring system power whether or not power capping is used.
In addition, Turbostat now shows temperature from DTS
(Digital Thermal Sensor) and PTM (Package Thermal Monitor) hardware,
if present.
As before, turbostat reads MSRs, and never writes MSRs.
New columns are present in turbostat output:
The Pkg_W column shows Watts for each package (socket) in the system.
On multi-socket systems, the system summary on the 1st row shows the sum
for all sockets together.
The Cor_W column shows Watts due to processors cores.
Note that Core_W is included in Pkg_W.
The optional GFX_W column shows Watts due to the graphics "un-core".
Note that GFX_W is included in Pkg_W.
The optional RAM_W column on server processors shows Watts due to DRAM DIMMS.
As DRAM DIMMs are outside the processor package, RAM_W is not included in Pkg_W.
The optional PKG_% and RAM_% columns on server processors shows the % of time
in the measurement interval that RAPL power limiting is in effect on the
package and on DRAM.
Note that the RAPL energy counters have some limitations.
First, hardware updates the counters about once every milli-second.
This is fine for typical turbostat measurement intervals > 1 sec.
However, when turbostat is used to measure events that approach
1ms, the counters are less useful.
Second, the 32-bit energy counters are subject to wrapping.
For example, a counter incrementing 15 micro-Joule units
on a 130 Watt TDP server processor could (in theory)
roll over in about 9 minutes. Turbostat detects and handles
up to 1 counter overflow per measurement interval.
But when the measurement interval exceeds the guaranteed
counter range, we can't detect if more than 1 overflow occured.
So in this case turbostat indicates that the results are
in question by replacing the fractional part of the Watts
in the output with "**":
Pkg_W Cor_W GFX_W
3** 0** 0**
Third, the RAPL counters are energy (Joule) counters -- they sum up
weighted events in the package to estimate energy consumed. They are
not analong power (Watt) meters. In practice, they tend to under-count
because they don't cover every possible use of energy in the package.
The accuracy of the RAPL counters will vary between product generations,
and between SKU's in the same product generation, and with temperature.
turbostat's -v (verbose) option now displays more power and thermal configuration
information -- as shown on the turbostat.8 manual page.
For example, it now displays the Package and DRAM Thermal Design Power (TDP):
cpu0: MSR_PKG_POWER_INFO: 0x2f064001980410 (130 W TDP, RAPL 51 - 200 W, 0.045898 sec.)
cpu0: MSR_DRAM_POWER_INFO,: 0x28025800780118 (35 W TDP, RAPL 15 - 75 W, 0.039062 sec.)
cpu8: MSR_PKG_POWER_INFO: 0x2f064001980410 (130 W TDP, RAPL 51 - 200 W, 0.045898 sec.)
cpu8: MSR_DRAM_POWER_INFO,: 0x28025800780118 (35 W TDP, RAPL 15 - 75 W, 0.039062 sec.)
Signed-off-by: Len Brown <len.brown@intel.com>
In periodic mode, turbostat writes to stdout,
but users were un-able to re-direct stdout, eg.
turbostat > outputfile
would result in an empty outputfile.
Signed-off-by: Len Brown <len.brown@intel.com>
Turbostat assumed if it can't migrate to a CPU, then the CPU
must have gone off-line and turbostat should re-initialize
with the new topology.
But if turbostat can not migrate because it is restricted by
a cpuset, then it will fail to migrate even after re-initialization,
resulting in an infinite loop.
Spit out a warning when we can't migrate
and endure only 2 re-initialize cycles in a row
before giving up and exiting.
Signed-off-by: Len Brown <len.brown@intel.com>
Now that turbostat is built in the kernel tree,
it can share MSR #defines with the kernel.
Signed-off-by: Len Brown <len.brown@intel.com>
Cc: x86@kernel.org
Counting SMIs is popular, so add a dedicated "-s" option to do it,
and juggle some of the other option letters.
-S is now system summary (was -s)
-c is 32 bit counter (was -d)
-C is 64-bit counter (was -D)
-p is 1st thread in core (was -c)
-P is 1st thread in package (was -p)
bump the minor version number
Signed-off-by: Len Brown <len.brown@intel.com>
# turbostat -d 0x34
is useful for printing the number of SMI's within an interval
on Nehalem and newer processors.
where
# turbostat -m 0x34
will simply print out the total SMI count since reset.
Suggested-by: Andi Kleen
Signed-off-by: Len Brown <len.brown@intel.com>
The -M option dumps the specified 64-bit MSR with every sample.
Previously it was output at the end of each line.
However, with the v2 style of printing, the lines are now staggered,
making MSR output hard to read.
So move the MSR output column to the left where things are aligned.
Signed-off-by: Len Brown <len.brown@intel.com>
The "turbo-limit" is the maximum opportunistic processor
speed, assuming no electrical or thermal constraints.
For a given processor, the turbo-limit varies, depending
on the number of active cores. Generally, there is more
opportunity when fewer cores are active.
Under the "-v" verbose option, turbostat would
print the turbo-limits for the four cases
of 1 to 4 cores active.
Expand that capability to cover the cases of turbo
opportunities with up to 16 cores active.
Note that not all hardware platforms supply this information,
and that sometimes a valid limit may be specified for
a core which is not actually present.
Signed-off-by: Len Brown <len.brown@intel.com>
Under some conditions, c1% was displayed as very large number,
much higher than 100%.
c1% is not measured, it is derived as "that, which is left over"
from other counters. However, the other counters are not collected
atomically, and so it is possible for c1% to be calaculagted as
a small negative number -- displayed as very large positive.
There was a check for mperf vs tsc for this already,
but it needed to also include the other counters
that are used to calculate c1.
Signed-off-by: Len Brown <len.brown@intel.com>
Measuring large profoundly-idle configurations
requires turbostat to be more lightweight.
Otherwise, the operation of turbostat itself
can interfere with the measurements.
This re-write makes turbostat topology aware.
Hardware is accessed in "topology order".
Redundant hardware accesses are deleted.
Redundant output is deleted.
Also, output is buffered and
local RDTSC use replaces remote MSR access for TSC.
From a feature point of view, the output
looks different since redundant figures are absent.
Also, there are now -c and -p options -- to restrict
output to the 1st thread in each core, and the 1st
thread in each package, respectively. This is helpful
to reduce output on big systems, where more detail
than the "-s" system summary is desired.
Finally, periodic mode output is now on stdout, not stderr.
Turbostat v2 is also slightly more robust in
handling run-time CPU online/offline events,
as it now checks the actual map of on-line cpus rather
than just the total number of on-line cpus.
Signed-off-by: Len Brown <len.brown@intel.com>
Initial IVB support went into turbostat in Linux-3.1:
553575f1ae
(tools turbostat: recognize and run properly on IVB)
However, when running on IVB, turbostat would fail
to report the new couters added with SNB, c7, pc2 and pc7.
So in scenarios where these counters are non-zero on IVB,
turbostat would report erroneous residencey results.
In particular c7 time would be added to c1 time,
since c1 time is calculated as "that which is left over".
Also, turbostat reports MHz capabilities when passed
the "-v" option, and it would incorrectly report 133MHz
bclk instead of 100MHz bclk for IVB, which would inflate
GHz reported with that option.
This patch is a backport of a fix already included in turbostat v2.
Signed-off-by: Len Brown <len.brown@intel.com>
Linux 3.4 included a modification to turbostat to
lower cross-call overhead by using scheduler affinity:
15aaa34654
(tools turbostat: reduce measurement overhead due to IPIs)
In the use-case where turbostat forks a child program,
that change had the un-intended side-effect of binding
the child to the last cpu in the system.
This change removed the binding before forking the child.
This is a back-port of a fix already included in turbostat v2.
Signed-off-by: Len Brown <len.brown@intel.com>
Sometimes users have turbostat running in interval mode
when they take processors offline/online.
Previously, turbostat would survive, but not gracefully.
Tighten up the error checking so turbostat notices
changesn sooner, and print just 1 line on change:
turbostat: re-initialized with num_cpus %d
Signed-off-by: Len Brown <len.brown@intel.com>
turbostat uses /dev/cpu/*/msr interface to read MSRs.
For modern systems, it reads 10 MSR/CPU. This can
be observed as 10 "Function Call Interrupts"
per CPU per sample added to /proc/interrupts.
This overhead is measurable on large idle systems,
and as Yoquan Song pointed out, it can even trick
cpuidle into thinking the system is busy.
Here turbostat re-schedules itself in-turn to each
CPU so that its MSR reads will always be local.
This replaces the 10 "Function Call Interrupts"
with a single "Rescheduling interrupt" per sample
per CPU.
On an idle 32-CPU system, this shifts some residency from
the shallow c1 state to the deeper c7 state:
# ./turbostat.old -s
%c0 GHz TSC %c1 %c3 %c6 %c7 %pc2 %pc3 %pc6 %pc7
0.27 1.29 2.29 0.95 0.02 0.00 98.77 20.23 0.00 77.41 0.00
0.25 1.24 2.29 0.98 0.02 0.00 98.75 20.34 0.03 77.74 0.00
0.27 1.22 2.29 0.54 0.00 0.00 99.18 20.64 0.00 77.70 0.00
0.26 1.22 2.29 1.22 0.00 0.00 98.52 20.22 0.00 77.74 0.00
0.26 1.38 2.29 0.78 0.02 0.00 98.95 20.51 0.05 77.56 0.00
^C
i# ./turbostat.new -s
%c0 GHz TSC %c1 %c3 %c6 %c7 %pc2 %pc3 %pc6 %pc7
0.27 1.20 2.29 0.24 0.01 0.00 99.49 20.58 0.00 78.20 0.00
0.27 1.22 2.29 0.25 0.00 0.00 99.48 20.79 0.00 77.85 0.00
0.27 1.20 2.29 0.25 0.02 0.00 99.46 20.71 0.03 77.89 0.00
0.28 1.26 2.29 0.25 0.01 0.00 99.46 20.89 0.02 77.67 0.00
0.27 1.20 2.29 0.24 0.01 0.00 99.48 20.65 0.00 78.04 0.00
cc: Youquan Song <youquan.song@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
turbostat -s
cuts down on the amount of output, per user request.
also treak some output whitespace and the man page.
Signed-off-by: Len Brown <len.brown@intel.com>
This includes initial support for the recently published ACPI 5.0 spec.
In particular, support for the "hardware-reduced" bit that eliminates
the dependency on legacy hardware.
APEI has patches resulting from testing on real hardware.
Plus other random fixes.
* 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux: (52 commits)
acpi/apei/einj: Add extensions to EINJ from rev 5.0 of acpi spec
intel_idle: Split up and provide per CPU initialization func
ACPI processor: Remove unneeded variable passed by acpi_processor_hotadd_init V2
ACPI processor: Remove unneeded cpuidle_unregister_driver call
intel idle: Make idle driver more robust
intel_idle: Fix a cast to pointer from integer of different size warning in intel_idle
ACPI: kernel-parameters.txt : Add intel_idle.max_cstate
intel_idle: remove redundant local_irq_disable() call
ACPI processor: Fix error path, also remove sysdev link
ACPI: processor: fix acpi_get_cpuid for UP processor
intel_idle: fix API misuse
ACPI APEI: Convert atomicio routines
ACPI: Export interfaces for ioremapping/iounmapping ACPI registers
ACPI: Fix possible alignment issues with GAS 'address' references
ACPI, ia64: Use SRAT table rev to use 8bit or 16/32bit PXM fields (ia64)
ACPI, x86: Use SRAT table rev to use 8bit or 32bit PXM fields (x86/x86-64)
ACPI: Store SRAT table revision
ACPI, APEI, Resolve false conflict between ACPI NVS and APEI
ACPI, Record ACPI NVS regions
ACPI, APEI, EINJ, Refine the fix of resource conflict
...
Field names were shortened: "pkg" is now "pk", "core" is now "cr"
Signed-off-by: Arun Thomas <arun.thomas@gmail.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Reduce columns for package number to 1.
If you can afford more than 9 packages,
you can also afford a terminal with more than 80 columns:-)
Also shave a column also off the package C-states
Signed-off-by: Len Brown <len.brown@intel.com>
Follow kernel coding style traditions more closely.
Delete typedef, re-name "per cpu counters" to
simply be counters etc.
This patch changes no functionality.
Suggested-by: Thiago Farina <tfransosi@gmail.com>
Signed-off-by: Len Brown <len.brown@intel.com>
bug could cause false positive on indicating
presence of invarient TSC or APERF support.
Signed-off-by: Thomas Renninger <trenn@suse.de>
Signed-off-by: Len Brown <len.brown@intel.com>
MSR_IA32_ENERGY_PERF_BIAS first became available on Westmere Xeon.
It is implemented in all Sandy Bridge processors -- mobile, desktop and server.
It is expected to become increasingly important in subsequent generations.
x86_energy_perf_policy is a user-space utility to set the
hardware energy vs performance policy hint in the processor.
Most systems would benefit from "x86_energy_perf_policy normal"
at system startup, as the hardware default is maximum performance
at the expense of energy efficiency.
See x86_energy_perf_policy.8 man page for more information.
Background:
Linux-2.6.36 added "epb" to /proc/cpuinfo to indicate
if an x86 processor supports MSR_IA32_ENERGY_PERF_BIAS,
without actually modifying the MSR.
In March, 2010, Venkatesh Pallipadi proposed a small driver
that programmed MSR_IA32_ENERGY_PERF_BIAS, based on
the cpufreq governor in use. It also offered
a boot-time cmdline option to override.
http://lkml.org/lkml/2010/3/4/457
But hiding the hardware policy behind the
governor choice was deemed "kinda icky".
In June, 2010, I proposed a generic user/kernel API to
generalize the power/performance policy trade-off.
"RFC: /sys/power/policy_preference"
http://lkml.org/lkml/2010/6/16/399
That is my preference for implementing this capability,
but I received no support on the list.
So in September, 2010, I sent x86_energy_perf_policy.c to LKML,
a user-space utility that scribbles directly to the MSR.
http://lkml.org/lkml/2010/9/28/246
Here is that same utility, after responding to some review feedback,
to live in tools/power/, where it is easily found.
Signed-off-by: Len Brown <len.brown@intel.com>
turbostat is a Linux tool to observe proper operation
of Intel(R) Turbo Boost Technology.
turbostat displays the actual processor frequency
on x86 processors that include APERF and MPERF MSRs.
Note that turbostat is of limited utility on Linux
kernels 2.6.29 and older, as acpi_cpufreq cleared
APERF/MPERF up through that release.
On Intel Core i3/i5/i7 (Nehalem) and newer processors,
turbostat also displays residency in idle power saving states,
which are necessary for diagnosing any cpuidle issues
that may have an effect on turbo-mode.
See the turbostat.8 man page for example usage.
Signed-off-by: Len Brown <len.brown@intel.com>
Len Brown