GPROF(1)

📦
version 2.47+3build2 · 7 versions
, in ubuntu stonking

✨ GPROF(1) — GNU ✨

okay bestie, buckle up 💖 this is gprof from the GNU fam, and she's the one who tells you where your program is actually spending its time. like, the receipts of your runtime. let's go 🫶

🎀 NAME 🎀

gprof — display call graph profile data 💅

🌸 SYNOPSIS 🌸

gprof [ -[abcDhilLrsTvwxyz] ] [ -[ABCeEfFJnNOpPqQRStZ][name] ]
 [ -I dirs ] [ -d[num] ] [ -k from/to ]
 [ -m min-count ] [ -R map_file ] [ -t table-length ]
 [ --[no-]annotated-source[=name] ]
 [ --[no-]exec-counts[=name] ]
 [ --[no-]flat-profile[=name] ] [ --[no-]graph[=name] ]
 [ --[no-]time=name] [ --all-lines ] [ --brief ]
 [ --debug[=level] ] [ --function-ordering ]
 [ --file-ordering map_file ] [ --directory-path=dirs ]
 [ --display-unused-functions ] [ --file-format=name ]
 [ --file-info ] [ --help ] [ --line ] [ --inline-file-names ]
 [ --min-count=n ] [ --no-static ] [ --print-path ]
 [ --separate-files ] [ --static-call-graph ] [ --sum ]
 [ --table-length=len ] [ --traditional ] [ --version ]
 [ --width=n ] [ --ignore-non-functions ]
 [ --demangle[=STYLE] ] [ --no-demangle ]
 [--external-symbol-table=name]
 [ image-file ] [ profile-file ... ]

💖 DESCRIPTION 💖

so basically, gprof spits out an execution profile of your C, Pascal, or Fortran77 program. the way it works is iconic: the effects of the routines you called get folded into the profile of whoever called them. main character energy for every caller fr fr 👑

the profile data itself comes from the call graph profile file — and by default that's gmon.out — which gets created by programs compiled with the -pg option of cc, pc, and f77. and not me just casually mentioning that -pg also links in versions of the library routines that are compiled for profiling. periodt. 💅

gprof reads the object file you hand it (default is a.out) and links up its symbol table with the call graph profile from gmon.out. and if you pass more than one profile file? the gprof output shows the sum of the profile info across all of them. math queen ✨

if you're using gcc 2.95.x or 3.0 to compile your binaries, you may need to add -fprofile-arcs to the compile command line so the call graphs get properly stored in gmon.out. no cap, just add it.

gprof then calculates how much time was spent in each routine. next, those times get propagated along the edges of the call graph. cycles get discovered, and calls into a cycle are made to share the time of the cycle. it's giving collaborative budgeting fr 😭

and there are several forms of output available from all this analysis:

  • the flat profile shows how much time your program spent in each function, and how many times that function was called. if you literally just wanna know which functions are burning the most cycles, it's stated concisely here. short and sweet, ate 💅
  • the call graph shows, for each function, which functions called it, which other functions it called, and how many times. there's also an estimate of how much time was spent in the subroutines of each function. this can suggest places where you might try to eliminate function calls that use a lot of time. bestie red flag detector 🚩
  • the annotated source listing is a copy of the program's source code, labeled with the number of times each line of the program was executed. it's giving highlight reel 🌈

💅 OPTIONS 💅

these options specify which of several output formats gprof should produce. choose your fighter ✨

many of these options take an optional symspec to specify functions to be included or excluded. these options can be specified multiple times, with different *symspec*s, to include or exclude sets of symbols. stack 'em, queen 👑

specifying any of these options overrides the default (-p -q), which prints a flat profile and call graph analysis for all functions. so heads up, you're changing the vibe when you touch these.
  • -A[symspec], --annotated-source[=symspec] — the -A option makes gprof print annotated source code. and if symspec is specified? print output only for matching symbols. slay 💖
  • -b, --brief — if the -b option is given, gprof doesn't print the verbose blurbs that try to explain the meaning of all of the fields in the tables. super useful if you're gonna print the output, or if you're just tired of seeing the blurbs. we get it, you know what a field is 🥱
  • -B — the -B option makes gprof print the call graph analysis. that's it, that's the tweet 🐦
  • -C[symspec], --exec-counts[=symspec] — the -C option makes gprof print a tally of functions and the number of times each was called. if symspec is specified, print tally only for matching symbols.

and ooh, here's a fun one: if the profile data file contains basic-block count records, specifying the -l option, along with -C, will cause basic-block execution counts to be tallied and displayed. combo move 🔥

  • -i, --file-info — the -i option makes gprof display summary information about the profile data file(s) and then exit. the number of histogram, call graph, and basic-block count records is displayed. one and done, love that for her 💅
  • -I dirs, --directory-path=dirs — the -I option specifies a list of search directories in which to find source files. the environment variable GPROF_PATH can also be used to convey this information. used mostly for annotated source output. the search party 📂
  • -J[symspec], --no-annotated-source[=symspec] — the -J option makes gprof not print annotated source code. but plot twist: if symspec is specified, gprof prints annotated source, but excludes matching symbols. so you flip the whole thing 🔄
  • -L, --print-path — normally, source filenames are printed with the path component suppressed. the -L option makes gprof print the full pathname of source filenames, which is determined from symbolic debugging information in the image file and is relative to the directory in which the compiler was invoked. full government name, no nicknames 🗺️
  • -p[symspec], --flat-profile[=symspec] — the -p option makes gprof print a flat profile. if symspec is specified, print flat profile only for matching symbols. the classic ✨
  • -P[symspec], --no-flat-profile[=symspec] — the -P option makes gprof suppress printing a flat profile. and if symspec is specified, gprof prints a flat profile, but excludes matching symbols. reverse uno 🃏
  • -q[symspec], --graph[=symspec] — the -q option makes gprof print the call graph analysis. if symspec is specified, print call graph only for matching symbols and their children. bringing the whole friend group 🫶
  • -Q[symspec], --no-graph[=symspec] — the -Q option makes gprof suppress printing the call graph. if symspec is specified, gprof prints a call graph, but excludes matching symbols. ghosting specific symbols, valid 😌
  • -t, --table-length=num — the -t option makes the num most active source lines in each source file get listed when source annotation is enabled. the default is 10.
  • -y, --separate-files — this option affects annotated source output only. normally, gprof prints annotated source files to standard-output. but if this option is specified, annotated source for a file named path/filename is generated in the file filename-ann. and if the underlying file system would truncate filename-ann so that it overwrites the original filename, gprof generates annotated source in the file filename.ann instead (if the original file name has an extension, that extension is replaced with .ann). it's giving safety first, no accidental overwrites 💀
  • -Z[symspec], --no-exec-counts[=symspec] — the -Z option makes gprof not print a tally of functions and the number of times each was called. if symspec is specified, print tally, but exclude matching symbols. no counting today ✋
  • -r, --function-ordering — the --function-ordering option makes gprof print a suggested function ordering for the program based on profiling data. this option suggests an ordering which may improve paging, tlb and cache behavior for the program on systems which support arbitrary ordering of functions in an executable.
the exact details of how to force the linker to place functions in a particular order is system dependent and out of the scope of this manual. not my problem bestie, that's linker territory 😭
  • -R map_file, --file-ordering map_file — the --file-ordering option makes gprof print a suggested .o link line ordering for the program based on profiling data. this option suggests an ordering which may improve paging, tlb and cache behavior for the program on systems which do not support arbitrary ordering of functions in an executable.

use of the -a argument is highly recommended with this option. do it, trust 💖

the map_file argument is a pathname to a file which provides function name to object file mappings. the format of the file is similar to the output of the program nm.

c-parse.o:00000000 T yyparse
c-parse.o:00000004 C yyerrflag
c-lang.o:00000000 T maybe_objc_method_name
c-lang.o:00000000 T print_lang_statistics
c-lang.o:00000000 T recognize_objc_keyword
c-decl.o:00000000 T print_lang_identifier
c-decl.o:00000000 T print_lang_type
...

to create a map_file with GNU nm, type a command like nm --extern-only --defined-only -v --print-file-name program-name. copy that exactly, bestie 📝

  • -T, --traditional — the -T option makes gprof print its output in "traditional" BSD style. vintage aesthetic 🕰️
  • -w width, --width=width — sets width of output lines to width. currently only used when printing the function index at the bottom of the call graph. one job, does it well 👑
  • -x, --all-lines — this option affects annotated source output only. by default, only the lines at the beginning of a basic-block are annotated. if this option is specified, every line in a basic-block is annotated by repeating the annotation for the first line. this behavior is similar to tcov's -a. thorough queen 🌸
  • --demangle[=style], --no-demangle — these options control whether C++ symbol names should be demangled when printing output. the default is to demangle symbols. the --no-demangle option may be used to turn off demangling. different compilers have different mangling styles. the optional demangling style argument can be used to choose an appropriate demangling style for your compiler. pick your style, it's giving customization 💅

🦋 Analysis Options 🦋

  • -a, --no-static — the -a option makes gprof suppress the printing of statically declared (private) functions. (these are functions whose names are not listed as global, and which are not visible outside the file/function/block where they were defined.) time spent in these functions, calls to/from them, etc., will all get attributed to the function that was loaded directly before it in the executable file. this option affects both the flat profile and the call graph. sorry private functions, you're getting absorbed into the group project 😭
  • -c, --static-call-graph — okay so this one is lowkey doing detective work 🕵️♀️ It makes the call graph of your program get augmented by a heuristic that goes through the text space of the object file and sniffs out function calls hiding in the binary machine code. Why tho? Because normal call graph records only get generated when functions are actually entered, so this catches children that could have been called but never were. It also spots calls to functions that weren't compiled with profiling enabled — but only if symbol table entries exist for them. Calls to dynamic library routines? Typically not found by this option, sorry bestie. Parents or children found via this heuristic show up in the call graph with call counts of 0.
  • -D, --ignore-non-functions — this tells gprof to ignore symbols that aren't known to be functions. Gives you more accurate profile data on systems where it's supported (Solaris and HPUX, for example). ✨
  • -k from/to — this one lets you delete from the call graph any arcs going from symbols matching symspec from to those matching symspec to. Clean up that graph, girl 💅
  • -l, --line — enables line-by-line profiling, which charges histogram hits to individual source code lines instead of to functions. Big catch tho: this only works with programs compiled by older versions of the gcc compiler. Newer gcc versions are designed to work with the gcov tool instead. 🫠

If the program was compiled with basic-block counting enabled, this option will also identify how many times each line of code got executed. And yes, line-by-line profiling can help isolate where inside a large function your program is burning time — but it also significantly increases the running time of gprof, and it magnifies statistical inaccuracies. So, trade-offs, babes.

  • --inline-file-names — makes gprof print the source file after each symbol, in both the flat profile and the call graph. And if you pair it with the -L option, you get the full path to the file. 🌸
  • -m num, --min-count=num — this only affects execution count output. Any symbol executed fewer than num times gets suppressed. Quiet, small fry 💀
  • -nsymspec, --time=symspec** — this makes gprof, in its call graph analysis, only propagate times for symbols matching symspec.
  • -Nsymspec, --no-time=symspec — this makes gprof, in its call graph analysis, *not to propagate times for symbols matching *symspec.
  • -Sfilename, --external-symbol-table=filename** — makes gprof read an external symbol table file, like /proc/kallsyms, instead of reading the symbol table from the given object file (the default is a.out). Super useful for profiling kernel modules. 👑
  • -z, --display-unused-functions — give it this and gprof will mention all functions in the flat profile, even the ones that were never called and had no time spent in them. Pairs gorgeously with -c for discovering which routines were never called. It's giving "who even are you" energy 🎀

✨ Miscellaneous Options ✨

  • -d[num], *--debug[=num*]** — specifies debugging options. If num isn't specified, it enables all debugging. Go off, verbose king.
  • -h, --help — prints command line usage. The classic. 🫶
  • -Oname, --file-format=name** — selects the format of the profile data files. The recognized formats are auto (the default), bsd, 4.4bsd, magic, and prof (not yet supported). Notes taken, no cap.
  • -s, --sum — makes gprof summarize all the info in the profile data files it read in, and write out a profile data file called gmon.sum, which contains everything from the profile data files that gprof read. And plot twist: gmon.sum may itself be one of the specified input files — the effect of which is to merge the data in the other input files into gmon.sum.

Eventually you can run gprof again without -s to analyze the cumulative data inside gmon.sum. The long game. 🌈

  • -v, --version — prints the current version number, then dips. Bye bestie 👋

🚫 Deprecated Options 🚫

These options have been replaced by newer versions that use symspecs. Still documented, still honored, but like... consider upgrading, queen.

  • -e function_name — tells gprof to not print information about the function function_name (and its children...) in the call graph. It still gets listed as a child of any functions that call it, but its index number will be shown as [not printed]. You can give more than one -e option, but only one function_name per -e. 🚫
  • -E function_name — works like -e, but the time spent in the function (and children who weren't called from anywhere else) won't be used to compute the percentages-of-time for the call graph. More than one -E option may be given; only one function_name per -E.
  • -f function_name — limits the call graph to the function function_name and its children (and their children...). More than one -f option may be given; only one function_name per -f.
  • -F function_name — works like -f, but only the time spent in the function and its children (and their children...) will be used to determine total-time and percentages-of-time for the call graph. More than one -F option may be given; only one function_name per -F. And heads up: -F overrides -E — it's the main character here 💅

📁 FILES 📁

  • a.out — the namelist and text space.
  • gmon.out — dynamic call graph and profile.
  • gmon.sum — summarized dynamic call graph and profile.

🐛 BUGS 🐛

The granularity of the sampling is shown, but it remains statistical at best. We assume that the time for each execution of a function can be expressed by the total time for the function divided by the number of times the function is called. So the time propagated along the call graph arcs to the function's parents is directly proportional to the number of times that arc is traversed. Math is math, bestie 😭

Parents that aren't themselves profiled will have the time of their profiled children propagated to them — but they'll appear to be spontaneously invoked in the call graph listing, and their time won't be propagated any further. Similarly, signal catchers, even though profiled, will appear to be spontaneous (for more obscure reasons). Any profiled children of signal catchers should have their times propagated properly — unless the signal catcher was invoked during the execution of the profiling routine, in which case... all is lost. 💀

The profiled program must call exit(2) or return normally for the profiling information to be saved in the gmon.out file. Don't skip this, fr fr.

👀 SEE ALSO 👀

cc(1), prof(1), and the Info entry for gprof.

"An Execution Profiler for Modular Programs", by S. Graham, P. Kessler, M. McKusick; Software - Practice and Experience, Vol. 13, pp. 671-685, 1983.

"gprof: A Call Graph Execution Profiler", by S. Graham, P. Kessler, M. McKusick; Proceedings of the SIGPLAN '82 Symposium on Compiler Construction, SIGPLAN Notices, Vol. 17, No 6, pp. 120-126, June 1982.

Copyright (c) 1988-2026 Free Software Foundation, Inc.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.3 or any later version published by the Free Software Foundation; with no Invariant Sections, with no Front-Cover Texts, and with no Back-Cover Texts. A copy of the license is included in the section entitled "GNU Free Documentation License".

And that's a wrap, babe 🎀 this whole page is from binutils-2.47, dated 2026-09-19 — GPROF(1), iconic, no notes, periodt. 💖✨

📦 where this page comes from