LLVM-PROFDATA(1)

📦
version 19.1.7 · 12 versions
, in opensuse-leap 16.0

💅✨ LLVM-PROFDATA(1) — llvm-profdata, the profile data tool ✨💅

Welcome to the LLVM-PROFDATA(1) page, bestie — this is part 1, and yes there is more coming, so stay perched 💖

🌸 NAME 🌸

llvm-profdata — profile data tool. Short name, huge main character energy, literally the whole point 🎀

✨ SYNOPSIS ✨

llvm-profdata command [args...]

🦋 DESCRIPTION 🦋

The llvm-profdata tool is a small utility for working with profile data files. Small utility, very cutesy, but it is lowkey essential when you're doing profile stuff 💜

👑 COMMANDS 👑

  • merge
  • show
  • overlap
  • order

Those are the commands, periodt 💅

🔥 MERGE 🔥

🌈 SYNOPSIS 🌈

llvm-profdata merge [options] [filename...]

💜 DESCRIPTION 💜

llvm-profdata merge takes several profile data files generated by PGO instrumentation and merges them together into a single indexed profile data file. It is the gluing-together command, and it ate, no cap ✨

By default, profile data is merged without modification. That means the relative importance of each input file is proportional to the number of samples or counts it contains. In general, input from a longer training run will be interpreted as relatively more important than a shorter run. Depending on the nature of the training runs, it may be useful to adjust the weight given to each input file by using the -weighted-input option. Basically: longer training run, more influence; shorter run, less influence, and if that's not the vibe you want, reweight it 🎀

Profiles passed in via -weighted-input, -input-files, or via positional arguments are processed once for each time they are seen. Seen twice? Processed twice. That's the rule, fr fr 💫

💅 OPTIONS 💅

  • --help — Print a summary of command line options. The classic little helper, we love to see it 🫶
  • --output=<output>, -o — Specify the output file name.
🚨 Bestie, output cannot be - — the resulting indexed profile data can't be written to standard output. Do not try to pipe this cutie to stdout, it's not that kind of file 💀
  • --weighted-input=<weight,filename> — Specify an input file name along with a weight. The profile counts of the supplied filename will be scaled (multiplied) by the supplied weight, where weight is a decimal integer >= 1. Input files specified without using this option are assigned a default weight of 1. Examples are shown below, bestie 👇
  • --input-files=<path>, -f — Specify a file which contains a list of files to merge. The entries in this file are newline-separated. Lines starting with # are skipped. Entries may be of the form <filename> or <weight>,<filename>. Very list, very organized, slay ✨
  • --remapping-file=<path>, -r — Specify a file which contains a remapping from symbol names in the input profile to the symbol names that should be used in the output profile. The file should consist of lines of the form <input-symbol> <output-symbol>. Blank lines and lines starting with # are skipped.

The llvm-cxxmap tool can be used to generate the symbol remapping file. Iconic teamwork 💖

  • --instr (default) — Specify that the input profile is an instrumentation-based profile. This is the default, so it's giving "no notes" 💅
  • --sample — Specify that the input profile is a sample-based profile.

The format of the generated file can be generated in one of three ways — and the page then lists four, I'm just reading it 💀:

  • --binary (default) — Emit the profile using a binary encoding. For instrumentation-based profile the output format is the indexed binary format.
  • --extbinary — Emit the profile using an extensible binary encoding. This option can only be used with sample-based profile. The extensible binary encoding can be more compact with compression enabled and can be loaded faster than the default binary encoding. Faster loading? Obsessed 🦋
  • --text — Emit the profile in text mode. This option can also be used with both sample-based and instrumentation-based profile. When this option is used the profile will be dumped in the text format that is parsable by the profile reader.
  • --gcc — Emit the profile using GCC's gcov format (Not yet supported). The page says Not yet supported, so yeah: not yet, bestie 🌸
  • --sparse[=true|false] — Do not emit function records with 0 execution count. Can only be used in conjunction with -instr. Defaults to false, since it can inhibit compiler optimization during PGO.
🚨 Default is false for a reason, babes — it can inhibit compiler optimization during PGO. Respect the default 💅
  • --num-threads=<N>, -j — Use N threads to perform profile merging. When N=0, llvm-profdata auto-detects an appropriate number of threads to use. This is the default. Parallel merge era 🔥
  • --failure-mode=[any|all] — Set the failure mode. There are two options: any causes the merge command to fail if any profiles are invalid, and all causes the merge command to fail only if all profiles are invalid. If all is set, information from any invalid profiles is excluded from the final merged product. The default failure mode is any. Choose your fighter 🎀
  • --prof-sym-list=<path> — Specify a file which contains a list of symbols to generate profile symbol list in the profile. This option can only be used with sample-based profile in extbinary format. The entries in this file are newline-separated.
  • --compress-all-sections=[true|false] — Compress all sections when writing the profile. This option can only be used with sample-based profile in extbinary format. Compression cutie 💜
  • --use-md5=[true|false] — Use MD5 to represent string in name table when writing the profile. This option can only be used with sample-based profile in extbinary format.
  • --gen-partial-profile=[true|false] — Mark the profile to be a partial profile which only provides partial profile coverage for the optimized target. This option can only be used with sample-based profile in extbinary format.
  • --convert-sample-profile-layout=[nest|flat] — Convert the merged profile into a profile with a new layout. Supported layout are nest (Nested profile, the input should be CS flat profile) and flat (Profile with nested inlinees flattened out). Layout glow-up 💫
  • --supplement-instr-with-sample=<file> — Supplement an instrumentation profile with sample profile. The sample profile is the input of the flag. Output will be in instrumentation format (only works with -instr). Teamwork makes the dream work 🫶
  • --zero-counter-threshold=<float> — For the function which is cold in instr profile but hot in sample profile, if the ratio of the number of zero counters divided by the total number of counters is above the threshold, the profile of the function will be regarded as being harmful for performance and will be dropped. Cold in one profile, hot in another, and potentially harmful? Dropped, periodt 💀
  • --instr-prof-cold-threshold=<int> — User specified cold threshold for instr profile which will override the cold threshold got from profile summary. You said what you said, and it overrides 🎀
  • --suppl-min-size-threshold=<int> — If the size of a function is smaller than the threshold, assume it can be inlined by PGO early inliner and it will not be adjusted based on sample profile. Tiny function behavior, very demure 🌸
  • --debug-info=<path> — Specify the executable or .dSYM that contains debug info for the raw profile. When --debug-info-correlate or --profile-correlate=debug-info was used for instrumentation, use this option to correlate the raw profile.
  • --binary-file=<path> — Specify the executable that contains profile data and profile name sections for the raw profile. When -profile-correlate=binary was used for instrumentation, use this option to correlate the raw profile.
  • --temporal-profile-trace-reservoir-size — The maximum number of temporal profile traces to be stored in the output profile. If more traces are added, we will use reservoir sampling to select which traces to keep. Note that changing this value between different merge invocations on the same indexed profile could result in sample bias. The default value is 100. Reservoir sampling said "I'll decide" 💅
  • --temporal-profile-max-trace-length — The maximum number of functions in a single temporal profile trace. Longer traces will be truncated. The default value is 1000. Long trace? Truncated, bestie ✂️
  • --function=<string> — Only keep functions matching the regex in the output, all others are erased from the profile. Regex said keep, everything else said bye 👋
  • --no-function=<string> — Remove functions matching the regex from the profile. If both --function and --no-function are specified and a function matches both, it is removed. Both match? Removed. No cap 🚨

🎀 EXAMPLES 🎀

🌟 Basic Usage 🌟

Merge three profiles:

llvm-profdata merge foo.profdata bar.profdata baz.profdata -output merged.profdata

💖 Weighted Input 💖

The input file foo.profdata is especially important, multiply its counts by 10:

llvm-profdata merge --weighted-input=10,foo.profdata bar.profdata baz.profdata --output merged.profdata

Exactly equivalent to the previous invocation (explicit form; useful for programmatic invocation):

llvm-profdata merge --weighted-input=10,foo.profdata --weighted-input=1,bar.profdata --weighted-input=1,baz.profdata --output merged.profdata

Part 2 is literally next, bestie — this page keeps going 💖

✨ SHOW ✨

SYNOPSIS

llvm-profdata show [options] [filename]

DESCRIPTION

Okay so llvm-profdata show is literally the "let me look at it" command — it takes a profile data file and shows you all the info about the profile counters in that file, plus the counters for any function(s) you ask about. It's giving detective energy 🔍💅

And if you leave out filename or just pass -? Then llvm-profdata show reads its input from standard input. Slay, piped input works.

OPTIONS

  • --all-functions — print the details for every function, no filter, no skips, all of them 👑
  • --binary-ids — print the embedded binary ids that live inside a profile.
  • --counts — print out the actual counter values for the functions being displayed.
  • --show-format=<text|json|yaml> — emit the output in the chosen format, but only if the profile type you've got actually supports it. iykyk.
  • --function=<string> — print details for a function whenever that function's name contains the given string 🎀
  • --help — print a summary of the command line options.
  • --output=<output>, -o — specify the output file name. If output is -, or you just don't specify it at all, the output goes to standard output.
  • --instr (default) — says the input profile is an instrumentation-based profile. This is the default, bestie.
  • --text — tells the profile dumper to show the profile counts in the text format of the instrumentation-based profile data representation. By default the profile info gets dumped in a more human-readable form (also text, just with annotations). Cute either way 💖
  • --topn=<n> — tells the profile dumper to show the top n functions with the hottest basic blocks up in the summary section. By default those topn functions are not dumped.
  • --sample — says the input profile is a sample-based profile.
  • --memop-sizes — show the profiled sizes of the memory intrinsic calls for the functions being shown.
  • --value-cutoff=<n> — only show functions whose max count values are greater than or equal to n. Default cutoff is 0.
  • --list-below-cutoff — only output the names of functions whose max count value falls below the cutoff value. Flip side of the coin 🪙
  • --profile-version — print the profile version.
  • --showcs — only show context sensitive profile counts. The default filters all context sensitive profile counts out.
  • --show-prof-sym-list=[true|false] — show the profile symbol list if it exists in the profile. Heads up: this option only means anything for a sample-based profile in extbinary format ⚠️
  • --show-sec-info-only=[true|false] — show basic info about each section in the profile. Again, only meaningful for a sample-based profile in extbinary format.
  • --debug-info=<path> — specify the executable or .dSYM that holds the debug info for the raw profile. So if --debug-info-correlate or --profile-correlate=debug-info was used for instrumentation, use this option to show the correlated functions from the raw profile 🔗
  • --covered — show only the functions that have actually been executed, i.e. functions with non-zero counts. We love a nonzero era.

🦋 OVERLAP 🦋

SYNOPSIS

llvm-profdata overlap [options] [base profile file] [test profile file]

DESCRIPTION

llvm-profdata overlap takes two profile data files and shows you the overlap of the counter distribution — both across the whole files and across any of the specified functions. It's the compatibility-check command, fr fr 💜

In this command, overlap is defined like this: suppose the base profile file has these counts:

{c1_1, c1_2, ..., c1_n, c1_u_1, c2_u_2, ..., c2_u_s}

and the test profile file has:

{c2_1, c2_2, ..., c2_n, c2_v_1, c2_v_2, ..., c2_v_t}

Here c{1|2}_i (i = 1 .. n) are matched counters, and c1_u_i (i = 1 .. s) and c2_v_i (i = 1 .. v) are unmatched counters (or counters only existing in) the base profile file and the test profile file, respectively. Then:

sum_1 = c1_1 + c1_2 +  ... + c1_n +  c1_u_1 + c2_u_2 + ... + c2_u_s
sum_2 = c2_1 + c2_2 + ... + c2_n + c2_v_1 + c2_v_2 + ... + c2_v_t
overlap = min(c1_1/sum_1, c2_1/sum_2) + min(c1_2/sum_1, c2_2/sum_2) + ...  + min(c1_n/sum_1, c2_n/sum_2)

The result overlap distribution is a percentage number, ranging from 0.0% to 100.0%, where 0.0% means there's no overlap at all and 100.0% means a perfect overlap. Periodt 💅

Here's an example: if the base profile file has counts of {400, 600}, and the test profile file has matched counts of {60000, 40000}, the overlap is 80%. Math ate that one up ✨

OPTIONS

  • --function=<string> — print details for a function if that function's name contains the given string.
  • --help — print a summary of the command line options.
  • --output=<output>, -o — specify the output file name. If output is -, or it isn't specified, the output goes to standard output.
  • --value-cutoff=<n> — only show functions whose max count values are greater than or equal to n. By default the value-cutoff is set to the max of unsigned long long 😭 (yes, that big)
  • --cs — only show overlap for the context sensitive profile counts. The default is to show non-context sensitive profile counts.

🌸 ORDER 🌸

SYNOPSIS

llvm-profdata order [options] [filename]

DESCRIPTION

llvm-profdata order uses temporal profiling traces from a profile and finds a function order that reduces the number of page faults for those traces. Basically it's out here rearranging your functions so your program stops tripping over itself 🧚‍♀️

This output can be passed directly to lld via --symbol-ordering-file= for ELF, or -order-file for Mach-O. And if the traces found in the profile are actually representative of the real world, then this order should improve startup performance. Iconic behaviour honestly 🔥

OPTIONS

  • --help — print a summary of the command line options.
  • --output=<output>, -o — specify the output file name. If output is -, or it isn't specified, the output goes to standard output.

💀 EXIT STATUS 💀

llvm-profdata returns 1 if the command is omitted or is invalid, if it can't read the input files, or if there's a mismatch between their data. So a 1 here is the "something's off bestie" signal — no cap 🙈

👑 AUTHOR 👑

Maintained by the LLVM Team (https://llvm.org/).

2003-2025, LLVM Project

Okay that's the whole page, you made it to the end!! 🫶 This was page 19 of LLVM-PROFDATA(1), dated 2025-01-17 — go forth and profile responsibly, bestie 💖✨🌈

📦 where this page comes from