turbostat.8 6.3 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173
  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 [ "\-v" ]
  8. .RB [ "\-M MSR#" ]
  9. .RB command
  10. .br
  11. .B turbostat
  12. .RB [ "\-v" ]
  13. .RB [ "\-M MSR#" ]
  14. .RB [ "\-i interval_sec" ]
  15. .SH DESCRIPTION
  16. \fBturbostat \fP reports processor topology, frequency
  17. and idle power state statistics on modern X86 processors.
  18. Either \fBcommand\fP is forked and statistics are printed
  19. upon its completion, or statistics are printed periodically.
  20. \fBturbostat \fP
  21. requires that the processor
  22. supports an "invariant" TSC, plus the APERF and MPERF MSRs.
  23. \fBturbostat \fP will report idle cpu power state residency
  24. on processors that additionally support C-state residency counters.
  25. .SS Options
  26. The \fB-v\fP option increases verbosity.
  27. .PP
  28. The \fB-M MSR#\fP option dumps the specified MSR,
  29. in addition to the usual frequency and idle statistics.
  30. .PP
  31. The \fB-i interval_sec\fP option prints statistics every \fiinterval_sec\fP seconds.
  32. The default is 5 seconds.
  33. .PP
  34. The \fBcommand\fP parameter forks \fBcommand\fP and upon its exit,
  35. displays the statistics gathered since it was forked.
  36. .PP
  37. .SH FIELD DESCRIPTIONS
  38. .nf
  39. \fBpkg\fP processor package number.
  40. \fBcore\fP processor core number.
  41. \fBCPU\fP Linux CPU (logical processor) number.
  42. \fB%c0\fP percent of the interval that the CPU retired instructions.
  43. \fBGHz\fP average clock rate while the CPU was in c0 state.
  44. \fBTSC\fP average GHz that the TSC ran during the entire interval.
  45. \fB%c1, %c3, %c6\fP show the percentage residency in hardware core idle states.
  46. \fB%pc3, %pc6\fP percentage residency in hardware package idle states.
  47. .fi
  48. .PP
  49. .SH EXAMPLE
  50. Without any parameters, turbostat prints out counters ever 5 seconds.
  51. (override interval with "-i sec" option, or specify a command
  52. for turbostat to fork).
  53. The first row of statistics reflect the average for the entire system.
  54. Subsequent rows show per-CPU statistics.
  55. .nf
  56. [root@x980]# ./turbostat
  57. core CPU %c0 GHz TSC %c1 %c3 %c6 %pc3 %pc6
  58. 0.04 1.62 3.38 0.11 0.00 99.85 0.00 95.07
  59. 0 0 0.04 1.62 3.38 0.06 0.00 99.90 0.00 95.07
  60. 0 6 0.02 1.62 3.38 0.08 0.00 99.90 0.00 95.07
  61. 1 2 0.10 1.62 3.38 0.29 0.00 99.61 0.00 95.07
  62. 1 8 0.11 1.62 3.38 0.28 0.00 99.61 0.00 95.07
  63. 2 4 0.01 1.62 3.38 0.01 0.00 99.98 0.00 95.07
  64. 2 10 0.01 1.61 3.38 0.02 0.00 99.98 0.00 95.07
  65. 8 1 0.07 1.62 3.38 0.15 0.00 99.78 0.00 95.07
  66. 8 7 0.03 1.62 3.38 0.19 0.00 99.78 0.00 95.07
  67. 9 3 0.01 1.62 3.38 0.02 0.00 99.98 0.00 95.07
  68. 9 9 0.01 1.62 3.38 0.02 0.00 99.98 0.00 95.07
  69. 10 5 0.01 1.62 3.38 0.13 0.00 99.86 0.00 95.07
  70. 10 11 0.08 1.62 3.38 0.05 0.00 99.86 0.00 95.07
  71. .fi
  72. .SH VERBOSE EXAMPLE
  73. The "-v" option adds verbosity to the output:
  74. .nf
  75. GenuineIntel 11 CPUID levels; family:model:stepping 0x6:2c:2 (6:44:2)
  76. 12 * 133 = 1600 MHz max efficiency
  77. 25 * 133 = 3333 MHz TSC frequency
  78. 26 * 133 = 3467 MHz max turbo 4 active cores
  79. 26 * 133 = 3467 MHz max turbo 3 active cores
  80. 27 * 133 = 3600 MHz max turbo 2 active cores
  81. 27 * 133 = 3600 MHz max turbo 1 active cores
  82. .fi
  83. The \fBmax efficiency\fP frequency, a.k.a. Low Frequency Mode, is the frequency
  84. available at the minimum package voltage. The \fBTSC frequency\fP is the nominal
  85. maximum frequency of the processor if turbo-mode were not available. This frequency
  86. should be sustainable on all CPUs indefinitely, given nominal power and cooling.
  87. The remaining rows show what maximum turbo frequency is possible
  88. depending on the number of idle cores. Note that this information is
  89. not available on all processors.
  90. .SH FORK EXAMPLE
  91. If turbostat is invoked with a command, it will fork that command
  92. and output the statistics gathered when the command exits.
  93. eg. Here a cycle soaker is run on 1 CPU (see %c0) for a few seconds
  94. until ^C while the other CPUs are mostly idle:
  95. .nf
  96. [root@x980 lenb]# ./turbostat cat /dev/zero > /dev/null
  97. ^Ccore CPU %c0 GHz TSC %c1 %c3 %c6 %pc3 %pc6
  98. 8.49 3.63 3.38 16.23 0.66 74.63 0.00 0.00
  99. 0 0 1.22 3.62 3.38 32.18 0.00 66.60 0.00 0.00
  100. 0 6 0.40 3.61 3.38 33.00 0.00 66.60 0.00 0.00
  101. 1 2 0.11 3.14 3.38 0.19 3.95 95.75 0.00 0.00
  102. 1 8 0.05 2.88 3.38 0.25 3.95 95.75 0.00 0.00
  103. 2 4 0.00 3.13 3.38 0.02 0.00 99.98 0.00 0.00
  104. 2 10 0.00 3.09 3.38 0.02 0.00 99.98 0.00 0.00
  105. 8 1 0.04 3.50 3.38 14.43 0.00 85.54 0.00 0.00
  106. 8 7 0.03 2.98 3.38 14.43 0.00 85.54 0.00 0.00
  107. 9 3 0.00 3.16 3.38 100.00 0.00 0.00 0.00 0.00
  108. 9 9 99.93 3.63 3.38 0.06 0.00 0.00 0.00 0.00
  109. 10 5 0.01 2.82 3.38 0.08 0.00 99.91 0.00 0.00
  110. 10 11 0.02 3.36 3.38 0.06 0.00 99.91 0.00 0.00
  111. 6.950866 sec
  112. .fi
  113. Above the cycle soaker drives cpu9 up 3.6 Ghz turbo limit
  114. while the other processors are generally in various states of idle.
  115. Note that cpu3 is an HT sibling sharing core9
  116. with cpu9, and thus it is unable to get to an idle state
  117. deeper than c1 while cpu9 is busy.
  118. Note that turbostat reports average GHz of 3.61, while
  119. the arithmetic average of the GHz column above is 3.24.
  120. This is a weighted average, where the weight is %c0. ie. it is the total number of
  121. un-halted cycles elapsed per time divided by the number of CPUs.
  122. .SH NOTES
  123. .B "turbostat "
  124. must be run as root.
  125. .B "turbostat "
  126. reads hardware counters, but doesn't write them.
  127. So it will not interfere with the OS or other programs, including
  128. multiple invocations of itself.
  129. \fBturbostat \fP
  130. may work poorly on Linux-2.6.20 through 2.6.29,
  131. as \fBacpi-cpufreq \fPperiodically cleared the APERF and MPERF
  132. in those kernels.
  133. The APERF, MPERF MSRs are defined to count non-halted cycles.
  134. Although it is not guaranteed by the architecture, turbostat assumes
  135. that they count at TSC rate, which is true on all processors tested to date.
  136. .SH REFERENCES
  137. "Intel® Turbo Boost Technology
  138. in Intel® Core™ Microarchitecture (Nehalem) Based Processors"
  139. http://download.intel.com/design/processor/applnots/320354.pdf
  140. "Intel® 64 and IA-32 Architectures Software Developer's Manual
  141. Volume 3B: System Programming Guide"
  142. http://www.intel.com/products/processor/manuals/
  143. .SH FILES
  144. .ta
  145. .nf
  146. /dev/cpu/*/msr
  147. .fi
  148. .SH "SEE ALSO"
  149. msr(4), vmstat(8)
  150. .PP
  151. .SH AUTHORS
  152. .nf
  153. Written by Len Brown <len.brown@intel.com>