tips.txt 2.8 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243
  1. For a higher level overview, try: perf report --sort comm,dso
  2. Sample related events with: perf record -e '{cycles,instructions}:S'
  3. Compare performance results with: perf diff [<old file> <new file>]
  4. Boolean options have negative forms, e.g.: perf report --no-children
  5. Customize output of perf script with: perf script -F event,ip,sym
  6. Generate a script for your data: perf script -g <lang>
  7. Save output of perf stat using: perf stat record <target workload>
  8. Create an archive with symtabs to analyse on other machine: perf archive
  9. Search options using a keyword: perf report -h <keyword>
  10. Use parent filter to see specific call path: perf report -p <regex>
  11. List events using substring match: perf list <keyword>
  12. To see list of saved events and attributes: perf evlist -v
  13. Use --symfs <dir> if your symbol files are in non-standard locations
  14. To see callchains in a more compact form: perf report -g folded
  15. Show individual samples with: perf script
  16. Limit to show entries above 5% only: perf report --percent-limit 5
  17. Profiling branch (mis)predictions with: perf record -b / perf report
  18. To show assembler sample contexts use perf record -b / perf script -F +brstackinsn --xed
  19. Treat branches as callchains: perf report --branch-history
  20. To count events in every 1000 msec: perf stat -I 1000
  21. Print event counts in CSV format with: perf stat -x,
  22. If you have debuginfo enabled, try: perf report -s sym,srcline
  23. For memory address profiling, try: perf mem record / perf mem report
  24. For tracepoint events, try: perf report -s trace_fields
  25. To record callchains for each sample: perf record -g
  26. To record every process run by a user: perf record -u <user>
  27. Skip collecting build-id when recording: perf record -B
  28. To change sampling frequency to 100 Hz: perf record -F 100
  29. See assembly instructions with percentage: perf annotate <symbol>
  30. If you prefer Intel style assembly, try: perf annotate -M intel
  31. For hierarchical output, try: perf report --hierarchy
  32. Order by the overhead of source file name and line number: perf report -s srcline
  33. System-wide collection from all CPUs: perf record -a
  34. Show current config key-value pairs: perf config --list
  35. Show user configuration overrides: perf config --user --list
  36. To add Node.js USDT(User-Level Statically Defined Tracing): perf buildid-cache --add `which node`
  37. To report cacheline events from previous recording: perf c2c report
  38. To browse sample contexts use perf report --sample 10 and select in context menu
  39. To separate samples by time use perf report --sort time,overhead,sym
  40. To set sample time separation other than 100ms with --sort time use --time-quantum
  41. Add -I to perf record to sample register values, which will be visible in perf report sample context.
  42. To show IPC for sampling periods use perf record -e '{cycles,instructions}:S' and then browse context
  43. To show context switches in perf report sample context add --switch-events to perf record.