turbostat.8 10 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252
  1. .TH TURBOSTAT 8
  2. .SH NAME
  3. turbostat \- Report processor frequency and idle statistics
  4. .SH SYNOPSIS
  5. .ft B
  6. .B turbostat
  7. .RB [ Options ]
  8. .RB command
  9. .br
  10. .B turbostat
  11. .RB [ Options ]
  12. .RB [ "\-i interval_sec" ]
  13. .SH DESCRIPTION
  14. \fBturbostat \fP reports processor topology, frequency,
  15. idle power-state statistics, temperature and power on modern X86 processors.
  16. Either \fBcommand\fP is forked and statistics are printed
  17. upon its completion, or statistics are printed periodically.
  18. \fBturbostat \fP
  19. must be run on root, and
  20. minimally requires that the processor
  21. supports an "invariant" TSC, plus the APERF and MPERF MSRs.
  22. Additional information is reported depending on hardware counter support.
  23. .SS Options
  24. The \fB-p\fP option limits output to the 1st thread in 1st core of each package.
  25. .PP
  26. The \fB-P\fP option limits output to the 1st thread in each Package.
  27. .PP
  28. The \fB-S\fP option limits output to a 1-line System Summary for each interval.
  29. .PP
  30. The \fB-v\fP option increases verbosity.
  31. .PP
  32. The \fB-c MSR#\fP option includes the delta of the specified 32-bit MSR counter.
  33. .PP
  34. The \fB-C MSR#\fP option includes the delta of the specified 64-bit MSR counter.
  35. .PP
  36. The \fB-m MSR#\fP option includes the the specified 32-bit MSR value.
  37. .PP
  38. The \fB-M MSR#\fP option includes the the specified 64-bit MSR value.
  39. .PP
  40. The \fB-i interval_sec\fP option prints statistics every \fiinterval_sec\fP seconds.
  41. The default is 5 seconds.
  42. .PP
  43. The \fBcommand\fP parameter forks \fBcommand\fP and upon its exit,
  44. displays the statistics gathered since it was forked.
  45. .PP
  46. .SH FIELD DESCRIPTIONS
  47. .nf
  48. \fBpk\fP processor package number.
  49. \fBcor\fP processor core number.
  50. \fBCPU\fP Linux CPU (logical processor) number.
  51. Note that multiple CPUs per core indicate support for Intel(R) Hyper-Threading Technology.
  52. \fB%c0\fP percent of the interval that the CPU retired instructions.
  53. \fBGHz\fP average clock rate while the CPU was in c0 state.
  54. \fBTSC\fP average GHz that the TSC ran during the entire interval.
  55. \fB%c1, %c3, %c6, %c7\fP show the percentage residency in hardware core idle states.
  56. \fBCTMP\fP Degrees Celsius reported by the per-core Digital Thermal Sensor.
  57. \fBPTMP\fP Degrees Celsius reported by the per-package Package Thermal Monitor.
  58. \fB%pc2, %pc3, %pc6, %pc7\fP percentage residency in hardware package idle states.
  59. \fBPkg_W\fP Watts consumed by the whole package.
  60. \fBCor_W\fP Watts consumed by the core part of the package.
  61. \fBGFX_W\fP Watts consumed by the Graphics part of the package -- available only on client processors.
  62. \fBRAM_W\fP Watts consumed by the DRAM DIMMS -- available only on server processors.
  63. \fBPKG_%\fP percent of the interval that RAPL throttling was active on the Package.
  64. \fBRAM_%\fP percent of the interval that RAPL throttling was active on DRAM.
  65. .fi
  66. .PP
  67. .SH EXAMPLE
  68. Without any parameters, turbostat prints out counters ever 5 seconds.
  69. (override interval with "-i sec" option, or specify a command
  70. for turbostat to fork).
  71. The first row of statistics is a summary for the entire system.
  72. For residency % columns, the summary is a weighted average.
  73. For Temperature columns, the summary is the column maximum.
  74. For Watts columns, the summary is a system total.
  75. Subsequent rows show per-CPU statistics.
  76. .nf
  77. [root@sandy]# ./turbostat
  78. cor CPU %c0 GHz TSC %c1 %c3 %c6 %c7 CTMP PTMP %pc2 %pc3 %pc6 %pc7 Pkg_W Cor_W GFX_W
  79. 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
  80. 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
  81. 0 4 0.03 0.80 2.29 0.12
  82. 1 1 0.04 0.80 2.29 0.25 0.01 0.00 99.71 40
  83. 1 5 0.16 0.80 2.29 0.13
  84. 2 2 0.05 0.80 2.29 0.06 0.01 0.00 99.88 40
  85. 2 6 0.03 0.80 2.29 0.08
  86. 3 3 0.05 0.80 2.29 0.08 0.00 0.00 99.87 47
  87. 3 7 0.04 0.84 2.29 0.09
  88. .fi
  89. .SH SUMMARY EXAMPLE
  90. The "-s" option prints the column headers just once,
  91. and then the one line system summary for each sample interval.
  92. .nf
  93. [root@wsm]# turbostat -S
  94. %c0 GHz TSC %c1 %c3 %c6 CTMP %pc3 %pc6
  95. 1.40 2.81 3.38 10.78 43.47 44.35 42 13.67 2.09
  96. 1.34 2.90 3.38 11.48 58.96 28.23 41 19.89 0.15
  97. 1.55 2.72 3.38 26.73 37.66 34.07 42 2.53 2.80
  98. 1.37 2.83 3.38 16.95 60.05 21.63 42 5.76 0.20
  99. .fi
  100. .SH VERBOSE EXAMPLE
  101. The "-v" option adds verbosity to the output:
  102. .nf
  103. [root@ivy]# turbostat -v
  104. turbostat v3.0 November 23, 2012 - Len Brown <lenb@kernel.org>
  105. CPUID(0): GenuineIntel 13 CPUID levels; family:model:stepping 0x6:3a:9 (6:58:9)
  106. CPUID(6): APERF, DTS, PTM, EPB
  107. RAPL: 851 sec. Joule Counter Range
  108. cpu0: MSR_NHM_PLATFORM_INFO: 0x81010f0012300
  109. 16 * 100 = 1600 MHz max efficiency
  110. 35 * 100 = 3500 MHz TSC frequency
  111. cpu0: MSR_NHM_SNB_PKG_CST_CFG_CTL: 0x1e008402 (UNdemote-C3, UNdemote-C1, demote-C3, demote-C1, locked: pkg-cstate-limit=2: pc6-noret)
  112. cpu0: MSR_NHM_TURBO_RATIO_LIMIT: 0x25262727
  113. 37 * 100 = 3700 MHz max turbo 4 active cores
  114. 38 * 100 = 3800 MHz max turbo 3 active cores
  115. 39 * 100 = 3900 MHz max turbo 2 active cores
  116. 39 * 100 = 3900 MHz max turbo 1 active cores
  117. cpu0: MSR_IA32_ENERGY_PERF_BIAS: 0x00000006 (balanced)
  118. cpu0: MSR_RAPL_POWER_UNIT: 0x000a1003 (0.125000 Watts, 0.000015 Joules, 0.000977 sec.)
  119. cpu0: MSR_PKG_POWER_INFO: 0x01e00268 (77 W TDP, RAPL 60 - 0 W, 0.000000 sec.)
  120. cpu0: MSR_PKG_POWER_LIMIT: 0x830000148268 (UNlocked)
  121. cpu0: PKG Limit #1: ENabled (77.000000 Watts, 1.000000 sec, clamp DISabled)
  122. cpu0: PKG Limit #2: ENabled (96.000000 Watts, 0.000977* sec, clamp DISabled)
  123. cpu0: MSR_PP0_POLICY: 0
  124. cpu0: MSR_PP0_POWER_LIMIT: 0x00000000 (UNlocked)
  125. cpu0: Cores Limit: DISabled (0.000000 Watts, 0.000977 sec, clamp DISabled)
  126. cpu0: MSR_PP1_POLICY: 0
  127. cpu0: MSR_PP1_POWER_LIMIT: 0x00000000 (UNlocked)
  128. cpu0: GFX Limit: DISabled (0.000000 Watts, 0.000977 sec, clamp DISabled)
  129. cpu0: MSR_IA32_TEMPERATURE_TARGET: 0x00691400 (105 C)
  130. cpu0: MSR_IA32_PACKAGE_THERM_STATUS: 0x884e0000 (27 C)
  131. cpu0: MSR_IA32_THERM_STATUS: 0x88560000 (19 C +/- 1)
  132. cpu1: MSR_IA32_THERM_STATUS: 0x88560000 (19 C +/- 1)
  133. cpu2: MSR_IA32_THERM_STATUS: 0x88540000 (21 C +/- 1)
  134. cpu3: MSR_IA32_THERM_STATUS: 0x884e0000 (27 C +/- 1)
  135. ...
  136. .fi
  137. The \fBmax efficiency\fP frequency, a.k.a. Low Frequency Mode, is the frequency
  138. available at the minimum package voltage. The \fBTSC frequency\fP is the nominal
  139. maximum frequency of the processor if turbo-mode were not available. This frequency
  140. should be sustainable on all CPUs indefinitely, given nominal power and cooling.
  141. The remaining rows show what maximum turbo frequency is possible
  142. depending on the number of idle cores. Note that this information is
  143. not available on all processors.
  144. .SH FORK EXAMPLE
  145. If turbostat is invoked with a command, it will fork that command
  146. and output the statistics gathered when the command exits.
  147. eg. Here a cycle soaker is run on 1 CPU (see %c0) for a few seconds
  148. until ^C while the other CPUs are mostly idle:
  149. .nf
  150. [root@x980 lenb]# ./turbostat cat /dev/zero > /dev/null
  151. ^C
  152. cor CPU %c0 GHz TSC %c1 %c3 %c6 %pc3 %pc6
  153. 8.86 3.61 3.38 15.06 31.19 44.89 0.00 0.00
  154. 0 0 1.46 3.22 3.38 16.84 29.48 52.22 0.00 0.00
  155. 0 6 0.21 3.06 3.38 18.09
  156. 1 2 0.53 3.33 3.38 2.80 46.40 50.27
  157. 1 8 0.89 3.47 3.38 2.44
  158. 2 4 1.36 3.43 3.38 9.04 23.71 65.89
  159. 2 10 0.18 2.86 3.38 10.22
  160. 8 1 0.04 2.87 3.38 99.96 0.01 0.00
  161. 8 7 99.72 3.63 3.38 0.27
  162. 9 3 0.31 3.21 3.38 7.64 56.55 35.50
  163. 9 9 0.08 2.95 3.38 7.88
  164. 10 5 1.42 3.43 3.38 2.14 30.99 65.44
  165. 10 11 0.16 2.88 3.38 3.40
  166. .fi
  167. Above the cycle soaker drives cpu7 up its 3.6 GHz turbo limit
  168. while the other processors are generally in various states of idle.
  169. Note that cpu1 and cpu7 are HT siblings within core8.
  170. As cpu7 is very busy, it prevents its sibling, cpu1,
  171. from entering a c-state deeper than c1.
  172. Note that turbostat reports average GHz of 3.63, while
  173. the arithmetic average of the GHz column above is lower.
  174. This is a weighted average, where the weight is %c0. ie. it is the total number of
  175. un-halted cycles elapsed per time divided by the number of CPUs.
  176. .SH SMI COUNTING EXAMPLE
  177. On Intel Nehalem and newer processors, MSR 0x34 is a System Management Mode Interrupt (SMI) counter.
  178. This counter is shown by default under the "SMI" column.
  179. .nf
  180. [root@x980 ~]# turbostat
  181. cor CPU %c0 GHz TSC SMI %c1 %c3 %c6 CTMP %pc3 %pc6
  182. 0.11 1.91 3.38 0 1.84 0.26 97.79 29 0.82 83.87
  183. 0 0 0.40 1.63 3.38 0 10.27 0.12 89.20 20 0.82 83.88
  184. 0 6 0.06 1.63 3.38 0 10.61
  185. 1 2 0.37 2.63 3.38 0 0.02 0.10 99.51 22
  186. 1 8 0.01 1.62 3.38 0 0.39
  187. 2 4 0.07 1.62 3.38 0 0.04 0.07 99.82 23
  188. 2 10 0.02 1.62 3.38 0 0.09
  189. 8 1 0.23 1.64 3.38 0 0.10 1.07 98.60 24
  190. 8 7 0.02 1.64 3.38 0 0.31
  191. 9 3 0.03 1.62 3.38 0 0.03 0.05 99.89 29
  192. 9 9 0.02 1.62 3.38 0 0.05
  193. 10 5 0.07 1.62 3.38 0 0.08 0.12 99.73 27
  194. 10 11 0.03 1.62 3.38 0 0.13
  195. ^C
  196. .fi
  197. .SH NOTES
  198. .B "turbostat "
  199. must be run as root.
  200. .B "turbostat "
  201. reads hardware counters, but doesn't write them.
  202. So it will not interfere with the OS or other programs, including
  203. multiple invocations of itself.
  204. \fBturbostat \fP
  205. may work poorly on Linux-2.6.20 through 2.6.29,
  206. as \fBacpi-cpufreq \fPperiodically cleared the APERF and MPERF
  207. in those kernels.
  208. If the TSC column does not make sense, then
  209. the other numbers will also make no sense.
  210. Turbostat is lightweight, and its data collection is not atomic.
  211. These issues are usually caused by an extremely short measurement
  212. interval (much less than 1 second), or system activity that prevents
  213. turbostat from being able to run on all CPUS to quickly collect data.
  214. The APERF, MPERF MSRs are defined to count non-halted cycles.
  215. Although it is not guaranteed by the architecture, turbostat assumes
  216. that they count at TSC rate, which is true on all processors tested to date.
  217. .SH REFERENCES
  218. "Intel® Turbo Boost Technology
  219. in Intel® Core™ Microarchitecture (Nehalem) Based Processors"
  220. http://download.intel.com/design/processor/applnots/320354.pdf
  221. "Intel® 64 and IA-32 Architectures Software Developer's Manual
  222. Volume 3B: System Programming Guide"
  223. http://www.intel.com/products/processor/manuals/
  224. .SH FILES
  225. .ta
  226. .nf
  227. /dev/cpu/*/msr
  228. .fi
  229. .SH "SEE ALSO"
  230. msr(4), vmstat(8)
  231. .PP
  232. .SH AUTHOR
  233. .nf
  234. Written by Len Brown <len.brown@intel.com>