OTFPROFILE(1)

✨ OTFPROFILE ✨

okay bestie, this is the manual page for otfprofile (section 1 of the manual, obvs) 💅 pull up a chair, we're gonna make an OTF trace into a whole look.

---

🌸 NAME 🌸

otfprofile — Generate a profile of an OTF trace

---

🦋 DESCRIPTION 🦋

otfprofile generates a profile of an OTF trace in LaTeX format. yes literally it takes your trace and turns it into a whole LaTeX serve. it's giving academic but make it cute 🎀

Syntax:

otfprofile -i <input file name> [options]

okay so here are the options, and bestie there are a LOT of them, but we're going through every single one, no skipping, periodt:

• -h, --help — shows you this help message. the "help me bestie" button 🫶
• -V — shows the OTF version. version check, lowkey essential.
• -v — cranks up output verbosity. and the way that you can use it more than once?? stack it, stack it, STACK it 🔥
• -i <file> — specify the input trace name. this is the one, the main character argument.
• -p — show progress. so you're not just staring at your terminal like 👁️👄👁️
• -f <n> — max. number of filehandles available per rank (default: 50)
• -b <size> — set buffersize of the reader (default: 1048576)
• -o <prefix> — specify the prefix of output file(s) (default: result)
• -g <n> — max. number of process groups in LaTeX output (range: 1-16, default: 16)
• -c, --cluster — do additional clustering of processes/threads. yes queen, group them ✨
• --cluster-alg <alg> — set comparison algorithm for clustering to "KMEANS" or "CLINKAGE" (default: KMEANS)
• -s <prefix> — calls otfshrink to apply the cluster mapping to input trace and produce a new trace named <prefix> with symbolic links to the original (implies -c)
• -H — use hard groups for CLINKAGE clustering (implies --cluster-alg CLINKAGE)
• -q <0-1> — quality threshold for CLINKAGE clustering (implies --cluster-alg CLINKAGE, default: 0.1)
• -d, --disp <options> — do additional analysis of irregularities using various output options, and you specify them in a comma-separated list. the possible values are:
  • filter — create VampirTrace filter rules from analysis information,
  • info — add information of irregularities to PDF output,
  • marker — add marker information to trace file

(implies --tex, default: info)

• --disp-mode <mode> — set profiling level within the analysis to "per-function" or "per-call-path" (default: per-function)
• --disp-reduction <percentage> — set percentage of call-paths to be filtered. (default: 15)
• --disp-filter <file> — name of the previous filter file that will be added to the new filter file
• -M, --csv-msg-matrix — write communication matrix in CSV file (implies --csv)
• -S, --csv-msg-size — write message length statistics in CSV file (implies --csv)
• --stat — read only summarized information, no events. minimal effort, maximum info, we love her 💜
• --[no]csv — enable/disable producing CSV output (default: disabled)
• --[no]tex — enable/disable producing LaTeX output (default: enabled)

🎀 heads up bestie: PDF creation requires the PGFPLOTS package version >1.4 — http://sourceforge.net/projects/pgfplots/ — no cap, go get it or the PDF is not gonna happen 😭

---

that's the whole thing, you ate that 💅 this was otfprofile 1.12.5, from February 2026, and honestly? iconic. go profile something, bestie 💖✨