giltene/wrk2 — reverse-engineered prompt
Reverse engineered prompt
Build me a fast command line HTTP benchmarking tool for testing a web server under steady load. I want to give it a URL, number of threads, number of open connections, test duration, and a target requests per second rate, then have it keep sending traffic at that constant rate instead of just going as fast as possible.
The important part is accurate latency reporting. It should measure latency from when each request was supposed to be sent until the response actually arrives, so slowdowns don’t get hidden. At the end, print a clear summary with total requests, transfer rate, requests per second, average latency, max latency, and detailed percentile latency numbers like 50, 75, 90, 99, 99.9, 99.99, and 99.999 percent.
Please make it efficient enough to create serious load from one machine, and keep it scriptable so advanced users can customize requests and reporting. If you need current docs for any low level networking details, look them up online.
Want more depth? Deep Reverse