cperf C performance guide (v0)cperf rule ids (slugs)cperf corpus targets cperf evidence policy

cp-alloc-hot — allocation in loop body cp-alloca-hot — measured stack growth, still hint cp-fopen-hot — measured, still hint cp-gettime-hot — measured, still hint cp-printf-hot — printf / fprintf in a loop cp-realloc-grow-one — realloc grow-by-one cp-regex-hot — measured, still hint cp-sprintf-hot — measured, still hint cp-strcat-hot — measured, stays off cp-strlen-in-cond — measured, still hint cp-struct-by-value — measured, still hint cp-wide-bool-flags — measured, still hint

Project

Contributing Changelog Security Policy Contributor Covenant Code of Conduct AGENT.md — XPerf agent operating contract

LanguagesCEvidence

cp-strcat-hot — measured, stays off

Status: measured. Default stays off (G3 cold CLI and path builders). Not re-escalated.

Claim

strcat of 500 chunks of 32 bytes, repeated 30 times, was about 221× an end pointer plus memcpy (3.097 ms versus 0.014 ms). Lengths matched.

How measured

clang -std=c11 -O2
Raw: cp-strcat-hot.jsonl

Limits

The ratio is this fixed chunk repeated onto an empty buffer. A destination that is already at its end, or a single strcat, is a different shape.

cp-sprintf-hot — measured, still hint

Previous Page

cp-strlen-in-cond — measured, still hint

Next Page

On this page

Claim How measured Limits