turbostat.8 10.2 KB
Newer Older
L
Len Brown 已提交
1 2 3 4 5 6
.TH TURBOSTAT 8
.SH NAME
turbostat \- Report processor frequency and idle statistics
.SH SYNOPSIS
.ft B
.B turbostat
7
.RB [ Options ]
L
Len Brown 已提交
8 9 10
.RB command
.br
.B turbostat
11
.RB [ Options ]
L
Len Brown 已提交
12 13
.RB [ "\-i interval_sec" ]
.SH DESCRIPTION
14 15
\fBturbostat \fP reports processor topology, frequency,
idle power-state statistics, temperature and power on modern X86 processors.
L
Len Brown 已提交
16 17 18 19
Either \fBcommand\fP is forked and statistics are printed
upon its completion, or statistics are printed periodically.

\fBturbostat \fP
20 21
must be run on root, and
minimally requires that the processor
L
Len Brown 已提交
22
supports an "invariant" TSC, plus the APERF and MPERF MSRs.
23
Additional information is reported depending on hardware counter support.
L
Len Brown 已提交
24 25

.SS Options
26
The \fB-p\fP option limits output to the 1st thread in 1st core of each package.
27
.PP
28
The \fB-P\fP option limits output to the 1st thread in each Package.
29
.PP
30
The \fB-S\fP option limits output to a 1-line System Summary for each interval.
L
Len Brown 已提交
31
.PP
L
Len Brown 已提交
32 33
The \fB-v\fP option increases verbosity.
.PP
34 35 36
The \fB-c MSR#\fP option includes the delta of the specified 32-bit MSR counter.
.PP
The \fB-C MSR#\fP option includes the delta of the specified 64-bit MSR counter.
37 38 39 40
.PP
The \fB-m MSR#\fP option includes the the specified 32-bit MSR value.
.PP
The \fB-M MSR#\fP option includes the the specified 64-bit MSR value.
L
Len Brown 已提交
41 42 43 44 45 46 47 48 49
.PP
The \fB-i interval_sec\fP option prints statistics every \fiinterval_sec\fP seconds.
The default is 5 seconds.
.PP
The \fBcommand\fP parameter forks \fBcommand\fP and upon its exit,
displays the statistics gathered since it was forked.
.PP
.SH FIELD DESCRIPTIONS
.nf
50
\fBpk\fP processor package number.
L
Len Brown 已提交
51
\fBcor\fP processor core number.
L
Len Brown 已提交
52
\fBCPU\fP Linux CPU (logical processor) number.
L
Len Brown 已提交
53
Note that multiple CPUs per core indicate support for Intel(R) Hyper-Threading Technology.
L
Len Brown 已提交
54 55 56
\fB%c0\fP percent of the interval that the CPU retired instructions.
\fBGHz\fP average clock rate while the CPU was in c0 state.
\fBTSC\fP average GHz that the TSC ran during the entire interval.
L
Len Brown 已提交
57
\fB%c1, %c3, %c6, %c7\fP show the percentage residency in hardware core idle states.
58 59
\fBCTMP\fP Degrees Celsius reported by the per-core Digital Thermal Sensor.
\fBPTMP\fP Degrees Celsius reported by the per-package Package Thermal Monitor.
L
Len Brown 已提交
60
\fB%pc2, %pc3, %pc6, %pc7\fP percentage residency in hardware package idle states.
61 62 63 64 65 66
\fBPkg_W\fP Watts consumed by the whole package.
\fBCor_W\fP Watts consumed by the core part of the package.
\fBGFX_W\fP Watts consumed by the Graphics part of the package -- available only on client processors.
\fBRAM_W\fP Watts consumed by the DRAM DIMMS -- available only on server processors.
\fBPKG_%\fP percent of the interval that RAPL throttling was active on the Package.
\fBRAM_%\fP percent of the interval that RAPL throttling was active on DRAM.
L
Len Brown 已提交
67 68 69 70 71 72 73
.fi
.PP
.SH EXAMPLE
Without any parameters, turbostat prints out counters ever 5 seconds.
(override interval with "-i sec" option, or specify a command
for turbostat to fork).

L
Len Brown 已提交
74
The first row of statistics is a summary for the entire system.
75 76 77
For residency % columns, the summary is a weighted average.
For Temperature columns, the summary is the column maximum.
For Watts columns, the summary is a system total.
L
Len Brown 已提交
78 79 80
Subsequent rows show per-CPU statistics.

.nf
81 82 83 84 85 86 87 88 89 90 91
[root@sandy]# ./turbostat
cor CPU    %c0  GHz  TSC    %c1    %c3    %c6    %c7 CTMP PTMP   %pc2   %pc3   %pc6   %pc7  Pkg_W  Cor_W GFX_W
          0.06 0.80 2.29   0.11   0.00   0.00  99.83   47   40   0.26   0.01   0.44  98.78   3.49   0.12  0.14
  0   0   0.07 0.80 2.29   0.07   0.00   0.00  99.86   40   40   0.26   0.01   0.44  98.78   3.49   0.12  0.14
  0   4   0.03 0.80 2.29   0.12
  1   1   0.04 0.80 2.29   0.25   0.01   0.00  99.71   40
  1   5   0.16 0.80 2.29   0.13
  2   2   0.05 0.80 2.29   0.06   0.01   0.00  99.88   40
  2   6   0.03 0.80 2.29   0.08
  3   3   0.05 0.80 2.29   0.08   0.00   0.00  99.87   47
  3   7   0.04 0.84 2.29   0.09
L
Len Brown 已提交
92 93 94 95 96 97
.fi
.SH SUMMARY EXAMPLE
The "-s" option prints the column headers just once,
and then the one line system summary for each sample interval.

.nf
98 99 100 101 102 103
[root@wsm]# turbostat -S
   %c0  GHz  TSC    %c1    %c3    %c6 CTMP   %pc3   %pc6
  1.40 2.81 3.38  10.78  43.47  44.35   42  13.67   2.09
  1.34 2.90 3.38  11.48  58.96  28.23   41  19.89   0.15
  1.55 2.72 3.38  26.73  37.66  34.07   42   2.53   2.80
  1.37 2.83 3.38  16.95  60.05  21.63   42   5.76   0.20
L
Len Brown 已提交
104 105 106 107 108
.fi
.SH VERBOSE EXAMPLE
The "-v" option adds verbosity to the output:

.nf
109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141
[root@ivy]# turbostat -v
turbostat v3.0 November 23, 2012 - Len Brown <lenb@kernel.org>
CPUID(0): GenuineIntel 13 CPUID levels; family:model:stepping 0x6:3a:9 (6:58:9)
CPUID(6): APERF, DTS, PTM, EPB
RAPL: 851 sec. Joule Counter Range
cpu0: MSR_NHM_PLATFORM_INFO: 0x81010f0012300
16 * 100 = 1600 MHz max efficiency
35 * 100 = 3500 MHz TSC frequency
cpu0: MSR_NHM_SNB_PKG_CST_CFG_CTL: 0x1e008402 (UNdemote-C3, UNdemote-C1, demote-C3, demote-C1, locked: pkg-cstate-limit=2: pc6-noret)
cpu0: MSR_NHM_TURBO_RATIO_LIMIT: 0x25262727
37 * 100 = 3700 MHz max turbo 4 active cores
38 * 100 = 3800 MHz max turbo 3 active cores
39 * 100 = 3900 MHz max turbo 2 active cores
39 * 100 = 3900 MHz max turbo 1 active cores
cpu0: MSR_IA32_ENERGY_PERF_BIAS: 0x00000006 (balanced)
cpu0: MSR_RAPL_POWER_UNIT: 0x000a1003 (0.125000 Watts, 0.000015 Joules, 0.000977 sec.)
cpu0: MSR_PKG_POWER_INFO: 0x01e00268 (77 W TDP, RAPL 60 - 0 W, 0.000000 sec.)
cpu0: MSR_PKG_POWER_LIMIT: 0x830000148268 (UNlocked)
cpu0: PKG Limit #1: ENabled (77.000000 Watts, 1.000000 sec, clamp DISabled)
cpu0: PKG Limit #2: ENabled (96.000000 Watts, 0.000977* sec, clamp DISabled)
cpu0: MSR_PP0_POLICY: 0
cpu0: MSR_PP0_POWER_LIMIT: 0x00000000 (UNlocked)
cpu0: Cores Limit: DISabled (0.000000 Watts, 0.000977 sec, clamp DISabled)
cpu0: MSR_PP1_POLICY: 0
cpu0: MSR_PP1_POWER_LIMIT: 0x00000000 (UNlocked)
cpu0: GFX Limit: DISabled (0.000000 Watts, 0.000977 sec, clamp DISabled)
cpu0: MSR_IA32_TEMPERATURE_TARGET: 0x00691400 (105 C)
cpu0: MSR_IA32_PACKAGE_THERM_STATUS: 0x884e0000 (27 C)
cpu0: MSR_IA32_THERM_STATUS: 0x88560000 (19 C +/- 1)
cpu1: MSR_IA32_THERM_STATUS: 0x88560000 (19 C +/- 1)
cpu2: MSR_IA32_THERM_STATUS: 0x88540000 (21 C +/- 1)
cpu3: MSR_IA32_THERM_STATUS: 0x884e0000 (27 C +/- 1)
 ...
L
Len Brown 已提交
142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157
.fi
The \fBmax efficiency\fP frequency, a.k.a. Low Frequency Mode, is the frequency
available at the minimum package voltage.  The \fBTSC frequency\fP is the nominal
maximum frequency of the processor if turbo-mode were not available.  This frequency
should be sustainable on all CPUs indefinitely, given nominal power and cooling.
The remaining rows show what maximum turbo frequency is possible
depending on the number of idle cores.  Note that this information is
not available on all processors.
.SH FORK EXAMPLE
If turbostat is invoked with a command, it will fork that command
and output the statistics gathered when the command exits.
eg. Here a cycle soaker is run on 1 CPU (see %c0) for a few seconds
until ^C while the other CPUs are mostly idle:

.nf
[root@x980 lenb]# ./turbostat cat /dev/zero > /dev/null
L
Len Brown 已提交
158 159
^C
cor CPU    %c0  GHz  TSC    %c1    %c3    %c6   %pc3   %pc6
160 161 162 163 164 165 166 167 168 169 170 171 172
          8.86 3.61 3.38  15.06  31.19  44.89   0.00   0.00
  0   0   1.46 3.22 3.38  16.84  29.48  52.22   0.00   0.00
  0   6   0.21 3.06 3.38  18.09
  1   2   0.53 3.33 3.38   2.80  46.40  50.27
  1   8   0.89 3.47 3.38   2.44
  2   4   1.36 3.43 3.38   9.04  23.71  65.89
  2  10   0.18 2.86 3.38  10.22
  8   1   0.04 2.87 3.38  99.96   0.01   0.00
  8   7  99.72 3.63 3.38   0.27
  9   3   0.31 3.21 3.38   7.64  56.55  35.50
  9   9   0.08 2.95 3.38   7.88
 10   5   1.42 3.43 3.38   2.14  30.99  65.44
 10  11   0.16 2.88 3.38   3.40
L
Len Brown 已提交
173
.fi
174
Above the cycle soaker drives cpu7 up its 3.6 GHz turbo limit
L
Len Brown 已提交
175 176
while the other processors are generally in various states of idle.

177 178 179
Note that cpu1 and cpu7 are HT siblings within core8.
As cpu7 is very busy, it prevents its sibling, cpu1,
from entering a c-state deeper than c1.
L
Len Brown 已提交
180

181
Note that turbostat reports average GHz of 3.63, while
L
Len Brown 已提交
182
the arithmetic average of the GHz column above is lower.
L
Len Brown 已提交
183 184
This is a weighted average, where the weight is %c0.  ie. it is the total number of
un-halted cycles elapsed per time divided by the number of CPUs.
185 186
.SH SMI COUNTING EXAMPLE
On Intel Nehalem and newer processors, MSR 0x34 is a System Management Mode Interrupt (SMI) counter.
187
This counter is shown by default under the "SMI" column.
188
.nf
189 190 191 192 193 194 195 196 197 198 199 200 201 202 203
[root@x980 ~]# turbostat
cor CPU    %c0  GHz  TSC SMI    %c1    %c3    %c6 CTMP   %pc3   %pc6
          0.11 1.91 3.38   0   1.84   0.26  97.79   29   0.82  83.87
  0   0   0.40 1.63 3.38   0  10.27   0.12  89.20   20   0.82  83.88
  0   6   0.06 1.63 3.38   0  10.61
  1   2   0.37 2.63 3.38   0   0.02   0.10  99.51   22
  1   8   0.01 1.62 3.38   0   0.39
  2   4   0.07 1.62 3.38   0   0.04   0.07  99.82   23
  2  10   0.02 1.62 3.38   0   0.09
  8   1   0.23 1.64 3.38   0   0.10   1.07  98.60   24
  8   7   0.02 1.64 3.38   0   0.31
  9   3   0.03 1.62 3.38   0   0.03   0.05  99.89   29
  9   9   0.02 1.62 3.38   0   0.05
 10   5   0.07 1.62 3.38   0   0.08   0.12  99.73   27
 10  11   0.03 1.62 3.38   0   0.13
204 205
^C
.fi
L
Len Brown 已提交
206 207 208 209 210 211 212 213 214 215 216 217 218 219 220
.SH NOTES

.B "turbostat "
must be run as root.

.B "turbostat "
reads hardware counters, but doesn't write them.
So it will not interfere with the OS or other programs, including
multiple invocations of itself.

\fBturbostat \fP
may work poorly on Linux-2.6.20 through 2.6.29,
as \fBacpi-cpufreq \fPperiodically cleared the APERF and MPERF
in those kernels.

221 222 223 224 225 226 227
If the TSC column does not make sense, then
the other numbers will also make no sense.
Turbostat is lightweight, and its data collection is not atomic.
These issues are usually caused by an extremely short measurement
interval (much less than 1 second), or system activity that prevents
turbostat from being able to run on all CPUS to quickly collect data.

L
Len Brown 已提交
228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249
The APERF, MPERF MSRs are defined to count non-halted cycles.
Although it is not guaranteed by the architecture, turbostat assumes
that they count at TSC rate, which is true on all processors tested to date.

.SH REFERENCES
"Intel® Turbo Boost Technology
in Intel® Core™ Microarchitecture (Nehalem) Based Processors"
http://download.intel.com/design/processor/applnots/320354.pdf

"Intel® 64 and IA-32 Architectures Software Developer's Manual
Volume 3B: System Programming Guide"
http://www.intel.com/products/processor/manuals/

.SH FILES
.ta
.nf
/dev/cpu/*/msr
.fi

.SH "SEE ALSO"
msr(4), vmstat(8)
.PP
L
Len Brown 已提交
250
.SH AUTHOR
L
Len Brown 已提交
251 252
.nf
Written by Len Brown <len.brown@intel.com>