1TURBOSTAT(8) System Manager's Manual TURBOSTAT(8)
2
6 turbostat - Report processor frequency and idle statistics
7
9 turbostat [Options] command
10 turbostat [Options] [--interval seconds]
11
13 turbostat reports processor topology, frequency, idle power-state sta‐
14 tistics, temperature and power on X86 processors. There are two ways
15 to invoke turbostat. The first method is to supply a command, which is
16 forked and statistics are printed in one-shot upon its completion. The
17 second method is to omit the command, and turbostat displays statistics
18 every 5 seconds interval. The 5-second interval can be changed using
19 the --interval option.
20 Some information is not available on older processors.
22 Options
24 Options can be specified with a single or double '-', and only as much
25 of the option name as necessary to disambiguate it from others is nec‐
26 essary. Note that options are case-sensitive.
27 --add attributes add column with counter having specified 'attributes'.
29 The 'location' attribute is required, all others are optional.
30 location: {msrDDD | msr0xXXX | /sys/path...}
31 msrDDD is a decimal offset, eg. msr16
32 msr0xXXX is a hex offset, eg. msr0x10
33 /sys/path... is an absolute path to a sysfs attribute
34 scope: {cpu | core | package}
36 sample and print the counter for every cpu, core, or package.
37 default: cpu
38 size: {u32 | u64 }
40 MSRs are read as 64-bits, u32 truncates the displayed value to 32-bits.
41 default: u64
42 format: {raw | delta | percent}
44 'raw' shows the MSR contents in hex.
45 'delta' shows the difference in values during the measurement interval.
46 'percent' shows the delta as a percentage of the cycles elapsed.
47 default: delta
48 name: "name_string"
50 Any string that does not match a key-word above is used
51 as the column header.
52 --cpu cpu-set limit output to system summary plus the specified cpu-
54 set. If cpu-set is the string "core", then the system summary plus the
55 first CPU in each core are printed -- eg. subsequent HT siblings are
56 not printed. Or if cpu-set is the string "package", then the system
57 summary plus the first CPU in each package is printed. Otherwise, the
58 system summary plus the specified set of CPUs are printed. The cpu-set
59 is ordered from low to high, comma delimited with ".." and "-" permit‐
60 ted to denote a range. eg. 1,2,8,14..17,21-44
61 --hide column do not show the specified built-in columns. May be
63 invoked multiple times, or with a comma-separated list of column names.
64 Use "--hide sysfs" to hide the sysfs statistics columns as a group.
65 --enable column show the specified built-in columns, which are other‐
67 wise disabled, by default. Currently the only built-in counters dis‐
68 abled by default are "usec" and "Time_Of_Day_Seconds". The column name
69 "all" can be used to enable all disabled-by-default built-in counters.
70 --show column show only the specified built-in columns. May be invoked
72 multiple times, or with a comma-separated list of column names. Use
73 "--show sysfs" to show the sysfs statistics columns as a group.
74 --Dump displays the raw counter values.
76 --quiet Do not decode and print the system configuration header infor‐
78 mation.
79 --interval seconds overrides the default 5.0 second measurement inter‐
81 val.
82 --num_iterations num number of the measurement iterations.
84 --out output_file turbostat output is written to the specified out‐
86 put_file. The file is truncated if it already exists, and it is cre‐
87 ated if it does not exist.
88 --help displays usage for the most common parameters.
90 --Joules displays energy in Joules, rather than dividing Joules by time
92 to print power in Watts.
93 --list display column header names available for use by --show and
95 --hide, then exit.
96 --Summary limits output to a 1-line System Summary for each interval.
98 --TCC temperature sets the Thermal Control Circuit temperature for sys‐
100 tems which do not export that value. This is used for making sense of
101 the Digital Thermal Sensor outputs, as they return degrees Celsius
102 below the TCC activation temperature.
103 --version displays the version.
105 The command parameter forks command, and upon its exit, displays the
107 statistics gathered since it was forked.
108
110 The system configuration dump (if --quiet is not used) is followed by
111 statistics. The first row of the statistics labels the content of each
112 column (below). The second row of statistics is the system summary
113 line. The system summary line has a '-' in the columns for the Pack‐
114 age, Core, and CPU. The contents of the system summary line depends on
115 the type of column. Columns that count items (eg. IRQ) show the sum
116 across all CPUs in the system. Columns that show a percentage show the
117 average across all CPUs in the system. Columns that dump raw MSR val‐
118 ues simply show 0 in the summary. After the system summary row, each
119 row describes a specific Package/Core/CPU. Note that if the --cpu
120 parameter is used to limit which specific CPUs are displayed, turbostat
121 will still collect statistics for all CPUs in the system and will still
122 show the system summary for all CPUs in the system.
123
125 usec For each CPU, the number of microseconds elapsed during counter collection, including thread migration -- if any. This counter is disabled by default, and is enabled with "--enable usec", or --debug. On the summary row, usec refers to the total elapsed time to collect the counters on all cpus.
126 Time_Of_Day_Seconds For each CPU, the gettimeofday(2) value (seconds.subsec since Epoch) when the counters ending the measurement interval were collected. This column is disabled by default, and can be enabled with "--enable Time_Of_Day_Seconds" or "--debug". On the summary row, Time_Of_Day_Seconds refers to the timestamp following collection of counters on the last CPU.
127 Core processor core number. Note that multiple CPUs per core indicate support for Intel(R) Hyper-Threading Technology (HT).
128 CPU Linux CPU (logical processor) number. Yes, it is okay that on many systems the CPUs are not listed in numerical order -- for efficiency reasons, turbostat runs in topology order, so HT siblings appear together.
129 Package processor package number -- not present on systems with a single processor package.
130 Avg_MHz number of cycles executed divided by time elapsed. Note that this includes idle-time when 0 instructions are executed.
131 Busy% percent of the measurement interval that the CPU executes instructions, aka. % of time in "C0" state.
132 Bzy_MHz average clock rate while the CPU was not idle (ie. in "c0" state).
133 TSC_MHz average MHz that the TSC ran during the entire interval.
134 IRQ The number of interrupts serviced by that CPU during the measurement interval. The system total line is the sum of interrupts serviced across all CPUs. turbostat parses /proc/interrupts to generate this summary.
135 SMI The number of System Management Interrupts serviced CPU during the measurement interval. While this counter is actually per-CPU, SMI are triggered on all processors, so the number should be the same for all CPUs.
136 C1, C2, C3... The number times Linux requested the C1, C2, C3 idle state during the measurement interval. The system summary line shows the sum for all CPUs. These are C-state names as exported in /sys/devices/system/cpu/cpu*/cpuidle/state*/name. While their names are generic, their attributes are processor specific. They the system description section of output shows what MWAIT sub-states they are mapped to on each system.
137 C1%, C2%, C3% The residency percentage that Linux requested C1, C2, C3.... The system summary is the average of all CPUs in the system. Note that these are software, reflecting what was requested. The hardware counters reflect what was actually achieved.
138 CPU%c1, CPU%c3, CPU%c6, CPU%c7 show the percentage residency in hardware core idle states. These numbers are from hardware residency counters.
139 CoreTmp Degrees Celsius reported by the per-core Digital Thermal Sensor.
140 PkgTtmp Degrees Celsius reported by the per-package Package Thermal Monitor.
141 GFX%rc6 The percentage of time the GPU is in the "render C6" state, rc6, during the measurement interval. From /sys/class/drm/card0/power/rc6_residency_ms.
142 GFXMHz Instantaneous snapshot of what sysfs presents at the end of the measurement interval. From /sys/class/graphics/fb0/device/drm/card0/gt_cur_freq_mhz.
143 Pkg%pc2, Pkg%pc3, Pkg%pc6, Pkg%pc7 percentage residency in hardware package idle states. These numbers are from hardware residency counters.
144 PkgWatt Watts consumed by the whole package.
145 CorWatt Watts consumed by the core part of the package.
146 GFXWatt Watts consumed by the Graphics part of the package -- available only on client processors.
147 RAMWatt Watts consumed by the DRAM DIMMS -- available only on server processors.
148 PKG_% percent of the interval that RAPL throttling was active on the Package.
149 RAM_% percent of the interval that RAPL throttling was active on DRAM.
150
152 By default, turbostat dumps all possible information -- a system con‐
153 figuration header, followed by columns for all counters. This is ideal
154 for remote debugging, use the "--out" option to save everything to a
155 text file, and get that file to the expert helping you debug.
156
158 When you are not interested in all that information, and there are sev‐
159 eral ways to see only what you want. First the "--quiet" option will
160 skip the configuration information, and turbostat will show only the
161 counter columns. Second, you can reduce the columns with the "--hide"
162 and "--show" options. If you use the "--show" option, then turbostat
163 will show only the columns you list. If you use the "--hide" option,
164 turbostat will show all columns, except the ones you list.
165 To find out what columns are available for --show and --hide, the
167 "--list" option is available. For convenience, the special strings
168 "sysfs" can be used to refer to all of the sysfs C-state counters at
169 once:
170 sudo ./turbostat --show sysfs --quiet sleep 10
171 10.003837 sec
172 C1 C1E C3 C6 C7s C1% C1E% C3% C6% C7s%
173 4 21 2 2 459 0.14 0.82 0.00 0.00 98.93
174 1 17 2 2 130 0.00 0.02 0.00 0.00 99.80
175 0 0 0 0 31 0.00 0.00 0.00 0.00 99.95
176 2 1 0 0 52 1.14 6.49 0.00 0.00 92.21
177 1 2 0 0 52 0.00 0.08 0.00 0.00 99.86
178 0 0 0 0 71 0.00 0.00 0.00 0.00 99.89
179 0 0 0 0 25 0.00 0.00 0.00 0.00 99.96
180 0 0 0 0 74 0.00 0.00 0.00 0.00 99.94
181 0 1 0 0 24 0.00 0.00 0.00 0.00 99.84
182
184 If turbostat is invoked with a command, it will fork that command and
185 output the statistics gathered after the command exits. In this case,
186 turbostat output goes to stderr, by default. Output can instead be
187 saved to a file using the --out option. In this example, the "sleep
188 10" command is forked, and turbostat waits for it to complete before
189 saving all statistics into "ts.out". Note that "sleep 10" is not part
190 of turbostat, but is simply an example of a command that turbostat can
191 fork. The "ts.out" file is what you want to edit in a very wide win‐
192 dow, paste into a spreadsheet, or attach to a bugzilla entry.
193 [root@hsw]# ./turbostat -o ts.out sleep 10
195 [root@hsw]#
196
199 Without a command to fork, turbostat displays statistics ever 5 sec‐
200 onds. Periodic output goes to stdout, by default, unless --out is used
201 to specify an output file. The 5-second interval can be changed with
202 the "-i sec" option.
203 sudo ./turbostat --quiet --hide sysfs,IRQ,SMI,CoreTmp,PkgTmp,GFX%rc6,GFXMHz,PkgWatt,CorWatt,GFXWatt
204 Core CPU Avg_MHz Busy% Bzy_MHz TSC_MHz CPU%c1 CPU%c3 CPU%c6 CPU%c7
205 - - 488 12.52 3900 3498 12.50 0.00 0.00 74.98
206 0 0 5 0.13 3900 3498 99.87 0.00 0.00 0.00
207 0 4 3897 99.99 3900 3498 0.01
208 1 1 0 0.00 3856 3498 0.01 0.00 0.00 99.98
209 1 5 0 0.00 3861 3498 0.01
210 2 2 1 0.02 3889 3498 0.03 0.00 0.00 99.95
211 2 6 0 0.00 3863 3498 0.05
212 3 3 0 0.01 3869 3498 0.02 0.00 0.00 99.97
213 3 7 0 0.00 3878 3498 0.03
214 Core CPU Avg_MHz Busy% Bzy_MHz TSC_MHz CPU%c1 CPU%c3 CPU%c6 CPU%c7
215 - - 491 12.59 3900 3498 12.42 0.00 0.00 74.99
216 0 0 27 0.69 3900 3498 99.31 0.00 0.00 0.00
217 0 4 3898 99.99 3900 3498 0.01
218 1 1 0 0.00 3883 3498 0.01 0.00 0.00 99.99
219 1 5 0 0.00 3898 3498 0.01
220 2 2 0 0.01 3889 3498 0.02 0.00 0.00 99.98
221 2 6 0 0.00 3889 3498 0.02
222 3 3 0 0.00 3856 3498 0.01 0.00 0.00 99.99
223 3 7 0 0.00 3897 3498 0.01
224 This example also shows the use of the --hide option to skip columns
225 that are not wanted. Note that cpu4 in this example is 99.99% busy,
226 while the other CPUs are all under 1% busy. Notice that cpu4's HT sib‐
227 ling is cpu0, which is under 1% busy, but can get into CPU%c1 only,
228 because its cpu4's activity on shared hardware keeps it from entering a
229 deeper C-state.
230
233 By default, turbostat always dumps system configuration information
234 before taking measurements. In the example above, "--quiet" is used to
235 suppress that output. Here is an example of the configuration informa‐
236 tion:
237 turbostat version 2017.02.15 - Len Brown <lenb@kernel.org>
239 CPUID(0): GenuineIntel 13 CPUID levels; family:model:stepping 0x6:3c:3 (6:60:3)
240 CPUID(1): SSE3 MONITOR - EIST TM2 TSC MSR ACPI-TM TM
241 CPUID(6): APERF, TURBO, DTS, PTM, No-HWP, No-HWPnotify, No-HWPwindow, No-HWPepp, No-HWPpkg, EPB
242 cpu4: MSR_IA32_MISC_ENABLE: 0x00850089 (TCC EIST No-MWAIT PREFETCH TURBO)
243 CPUID(7): No-SGX
244 cpu4: MSR_MISC_PWR_MGMT: 0x00400000 (ENable-EIST_Coordination DISable-EPB DISable-OOB)
245 RAPL: 3121 sec. Joule Counter Range, at 84 Watts
246 cpu4: MSR_PLATFORM_INFO: 0x80838f3012300
247 8 * 100.0 = 800.0 MHz max efficiency frequency
248 35 * 100.0 = 3500.0 MHz base frequency
249 cpu4: MSR_IA32_POWER_CTL: 0x0004005d (C1E auto-promotion: DISabled)
250 cpu4: MSR_TURBO_RATIO_LIMIT: 0x25262727
251 37 * 100.0 = 3700.0 MHz max turbo 4 active cores
252 38 * 100.0 = 3800.0 MHz max turbo 3 active cores
253 39 * 100.0 = 3900.0 MHz max turbo 2 active cores
254 39 * 100.0 = 3900.0 MHz max turbo 1 active cores
255 cpu4: MSR_CONFIG_TDP_NOMINAL: 0x00000023 (base_ratio=35)
256 cpu4: MSR_CONFIG_TDP_LEVEL_1: 0x00000000 ()
257 cpu4: MSR_CONFIG_TDP_LEVEL_2: 0x00000000 ()
258 cpu4: MSR_CONFIG_TDP_CONTROL: 0x80000000 ( lock=1)
259 cpu4: MSR_TURBO_ACTIVATION_RATIO: 0x00000000 (MAX_NON_TURBO_RATIO=0 lock=0)
260 cpu4: MSR_PKG_CST_CONFIG_CONTROL: 0x1e000400 (UNdemote-C3, UNdemote-C1, demote-C3, demote-C1, UNlocked: pkg-cstate-limit=0: pc0)
261 cpu4: POLL: CPUIDLE CORE POLL IDLE
262 cpu4: C1: MWAIT 0x00
263 cpu4: C1E: MWAIT 0x01
264 cpu4: C3: MWAIT 0x10
265 cpu4: C6: MWAIT 0x20
266 cpu4: C7s: MWAIT 0x32
267 cpu4: MSR_MISC_FEATURE_CONTROL: 0x00000000 (L2-Prefetch L2-Prefetch-pair L1-Prefetch L1-IP-Prefetch)
268 cpu0: MSR_IA32_ENERGY_PERF_BIAS: 0x00000006 (balanced)
269 cpu0: MSR_CORE_PERF_LIMIT_REASONS, 0x31200000 (Active: ) (Logged: Transitions, MultiCoreTurbo, Amps, Auto-HWP, )
270 cpu0: MSR_GFX_PERF_LIMIT_REASONS, 0x00000000 (Active: ) (Logged: )
271 cpu0: MSR_RING_PERF_LIMIT_REASONS, 0x0d000000 (Active: ) (Logged: Amps, PkgPwrL1, PkgPwrL2, )
272 cpu0: MSR_RAPL_POWER_UNIT: 0x000a0e03 (0.125000 Watts, 0.000061 Joules, 0.000977 sec.)
273 cpu0: MSR_PKG_POWER_INFO: 0x000002a0 (84 W TDP, RAPL 0 - 0 W, 0.000000 sec.)
274 cpu0: MSR_PKG_POWER_LIMIT: 0x428348001a82a0 (UNlocked)
275 cpu0: PKG Limit #1: ENabled (84.000000 Watts, 8.000000 sec, clamp DISabled)
276 cpu0: PKG Limit #2: ENabled (105.000000 Watts, 0.002441* sec, clamp DISabled)
277 cpu0: MSR_PP0_POLICY: 0
278 cpu0: MSR_PP0_POWER_LIMIT: 0x00000000 (UNlocked)
279 cpu0: Cores Limit: DISabled (0.000000 Watts, 0.000977 sec, clamp DISabled)
280 cpu0: MSR_PP1_POLICY: 0
281 cpu0: MSR_PP1_POWER_LIMIT: 0x00000000 (UNlocked)
282 cpu0: GFX Limit: DISabled (0.000000 Watts, 0.000977 sec, clamp DISabled)
283 cpu0: MSR_IA32_TEMPERATURE_TARGET: 0x00641400 (100 C)
284 cpu0: MSR_IA32_PACKAGE_THERM_STATUS: 0x884c0800 (24 C)
285 cpu0: MSR_IA32_THERM_STATUS: 0x884c0000 (24 C +/- 1)
286 cpu1: MSR_IA32_THERM_STATUS: 0x88510000 (19 C +/- 1)
287 cpu2: MSR_IA32_THERM_STATUS: 0x884e0000 (22 C +/- 1)
288 cpu3: MSR_IA32_THERM_STATUS: 0x88510000 (19 C +/- 1)
289 cpu4: MSR_PKGC3_IRTL: 0x00008842 (valid, 67584 ns)
290 cpu4: MSR_PKGC6_IRTL: 0x00008873 (valid, 117760 ns)
291 cpu4: MSR_PKGC7_IRTL: 0x00008891 (valid, 148480 ns)
292 The max efficiency frequency, a.k.a. Low Frequency Mode, is the fre‐
293 quency available at the minimum package voltage. The TSC frequency is
294 the base frequency of the processor -- this should match the brand
295 string in /proc/cpuinfo. This base frequency should be sustainable on
296 all CPUs indefinitely, given nominal power and cooling. The remaining
297 rows show what maximum turbo frequency is possible depending on the
298 number of idle cores. Note that not all information is available on
299 all processors.
300
302 Here we limit turbostat to showing just the CPU number for cpu0 - cpu3.
303 We add a counter showing the 32-bit raw value of MSR 0x199
304 (MSR_IA32_PERF_CTL), labeling it with the column header, "PRF_CTRL",
305 and display it only once, afte the conclusion of a 0.1 second sleep.
306 sudo ./turbostat --quiet --cpu 0-3 --show CPU --add msr0x199,u32,raw,PRF_CTRL sleep .1
307 0.101604 sec
308 CPU PRF_CTRL
309 - 0x00000000
310 0 0x00000c00
311 1 0x00000800
312 2 0x00000a00
313 3 0x00000800
314 ot including any non-busy idle time.
316
319 For interval-mode, turbostat will immediately end the current interval
320 when it sees a newline on standard input. turbostat will then start
321 the next interval. Control-C will be send a SIGINT to turbostat, which
322 will immediately abort the program with no further processing.
323
325 SIGINT will interrupt interval-mode. The end-of-interval data will be
326 collected and displayed before turbostat exits.
327 SIGUSR1 will end current interval, end-of-interval data will be col‐
329 lected and displayed before turbostat starts a new interval.
330
332 turbostat must be run as root. Alternatively, non-root users can be
333 enabled to run turbostat this way:
334 # setcap cap_sys_rawio=ep ./turbostat
336 # chmod +r /dev/cpu/*/msr
338 turbostat reads hardware counters, but doesn't write them. So it will
340 not interfere with the OS or other programs, including multiple invoca‐
341 tions of itself.
342 turbostat may work poorly on Linux-2.6.20 through 2.6.29, as acpi-
344 cpufreq periodically cleared the APERF and MPERF MSRs in those kernels.
345 AVG_MHz = APERF_delta/measurement_interval. This is the actual number
347 of elapsed cycles divided by the entire sample interval -- including
348 idle time. Note that this calculation is resilient to systems lacking
349 a non-stop TSC.
350 TSC_MHz = TSC_delta/measurement_interval. On a system with an invari‐
352 ant TSC, this value will be constant and will closely match the base
353 frequency value shown in the brand string in /proc/cpuinfo. On a sys‐
354 tem where the TSC stops in idle, TSC_MHz will drop below the proces‐
355 sor's base frequency.
356 Busy% = MPERF_delta/TSC_delta
358 Bzy_MHz = TSC_delta/APERF_delta/MPERF_delta/measurement_interval
360 Note that these calculations depend on TSC_delta, so they are not reli‐
362 able during intervals when TSC_MHz is not running at the base fre‐
363 quency.
364 Turbostat data collection is not atomic. Extremely short measurement
366 intervals (much less than 1 second), or system activity that prevents
367 turbostat from being able to run on all CPUS to quickly collect data,
368 will result in inconsistent results.
369 The APERF, MPERF MSRs are defined to count non-halted cycles. Although
371 it is not guaranteed by the architecture, turbostat assumes that they
372 count at TSC rate, which is true on all processors tested to date.
373
376 Volume 3B: System Programming Guide" http://www.intel.com/products/pro‐
377 cessor/manuals/
378
381 /dev/cpu/*/msr
382
385 msr(4), vmstat(8)
386
388 Written by Len Brown <len.brown@intel.com>
389 TURBOSTAT(8)