https://mariano-gonzalez.com/tags/jcstress/ Recent content in Jcstress on Mariano Gonzalez Hugo en-us <p>Copyright © 2026 Mariano Gonzalez</p><p xmlns:dct='http://purl.org/dc/terms/' xmlns:cc='http://creativecommons.org/ns#' class='license-text'><a rel='cc:attributionURL' property='dct:title' href='https://github.com/eschizoid/eschizoid.github.io/tree/main/content'>Content</a> licensed under <a rel='license' href='https://creativecommons.org/licenses/by/4.0'>CC BY 4.0</a></p> Wed, 22 Jul 2026 00:00:00 +0000 https://mariano-gonzalez.com/posts/post-9/ Wed, 22 Jul 2026 00:00:00 +0000 https://mariano-gonzalez.com/posts/post-9/ <p><p class="md__image"> <img src="logo.png" alt="kpipe" /> </p></p> <blockquote> <p><em>KPipe, part 3 of 3: <a href="https://mariano-gonzalez.com/posts/post-4/">a simpler way to structure consumers</a> · <a href="https://mariano-gonzalez.com/posts/post-5/">the first benchmarks</a> · <strong>fast is the easy part</strong></em></p> </blockquote> <p>Last time I benchmarked KPipe I ended with a warning about my own numbers. Single runs moved 5 to 40 percent between iterations. There was no latency-percentile data. I told you to read the percentages as one-sigma indicators and not to quote the absolute throughput at anyone.</p> <p>That was the honest thing to say at the time, and it was also a promissory note. This post is me paying it. The numbers here come from a quiesced machine - no browser, no IDE indexing, on mains power - with five JMH forks instead of one, the full <code>workMicros</code> sweep out to 100 ms per record, and latency percentiles alongside. The raw JSON and a dated markdown snapshot are committed in <a href="https://github.com/eschizoid/kpipe/tree/main/benchmarks/results"><code>benchmarks/results/</code></a> as always. Where the last post had ±5-40% error bars, this one has ±0.1-6% on nearly every cell.</p>