perf-mem.txt 2.2 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192
  1. perf-mem(1)
  2. ===========
  3. NAME
  4. ----
  5. perf-mem - Profile memory accesses
  6. SYNOPSIS
  7. --------
  8. [verse]
  9. 'perf mem' [<options>] (record [<command>] | report)
  10. DESCRIPTION
  11. -----------
  12. "perf mem record" runs a command and gathers memory operation data
  13. from it, into perf.data. Perf record options are accepted and are passed through.
  14. "perf mem report" displays the result. It invokes perf report with the
  15. right set of options to display a memory access profile. By default, loads
  16. and stores are sampled. Use the -t option to limit to loads or stores.
  17. Note that on Intel systems the memory latency reported is the use-latency,
  18. not the pure load (or store latency). Use latency includes any pipeline
  19. queueing delays in addition to the memory subsystem latency.
  20. OPTIONS
  21. -------
  22. <command>...::
  23. Any command you can specify in a shell.
  24. -i::
  25. --input=<file>::
  26. Input file name.
  27. -f::
  28. --force::
  29. Don't do ownership validation
  30. -t::
  31. --type=<type>::
  32. Select the memory operation type: load or store (default: load,store)
  33. -D::
  34. --dump-raw-samples::
  35. Dump the raw decoded samples on the screen in a format that is easy to parse with
  36. one sample per line.
  37. -x::
  38. --field-separator=<separator>::
  39. Specify the field separator used when dump raw samples (-D option). By default,
  40. The separator is the space character.
  41. -C::
  42. --cpu=<cpu>::
  43. Monitor only on the list of CPUs provided. Multiple CPUs can be provided as a
  44. comma-separated list with no space: 0,1. Ranges of CPUs are specified with -: 0-2. Default
  45. is to monitor all CPUS.
  46. -U::
  47. --hide-unresolved::
  48. Only display entries resolved to a symbol.
  49. -p::
  50. --phys-data::
  51. Record/Report sample physical addresses
  52. RECORD OPTIONS
  53. --------------
  54. -e::
  55. --event <event>::
  56. Event selector. Use 'perf mem record -e list' to list available events.
  57. -K::
  58. --all-kernel::
  59. Configure all used events to run in kernel space.
  60. -U::
  61. --all-user::
  62. Configure all used events to run in user space.
  63. -v::
  64. --verbose::
  65. Be more verbose (show counter open errors, etc)
  66. --ldlat <n>::
  67. Specify desired latency for loads event. (x86 only)
  68. In addition, for report all perf report options are valid, and for record
  69. all perf record options.
  70. SEE ALSO
  71. --------
  72. linkperf:perf-record[1], linkperf:perf-report[1]