https://pessolato.github.io/tags/benchmarking/Recent content in Benchmarking on Push to MasteryHugoen<a href='https://creativecommons.org/licenses/by-nc/4.0/' target='_blank' rel='noopener'>CC BY-NC 4.0</a>Sun, 21 Sep 2025 12:58:36 +0200https://pessolato.github.io/posts/go-http-drain-close-microbench/Sun, 21 Sep 2025 12:58:36 +0200https://pessolato.github.io/posts/go-http-drain-close-microbench/<p>Recently, a colleague and I were discussing whether draining the response body (my shorthand for “reading to EOF before closing”) was actually necessary when the response contains an error code. In that situation, the client already reads the full body on successful requests, so my colleague argued that skipping the drain on error responses would have negligible performance impact.</p> <p>That reasoning made sense at first glance, but it left me curious. I knew the theory, but I wanted to quantify the actual performance impact of not draining response bodies.</p>https://pessolato.github.io/posts/go-str-map-microbench/Sun, 22 Jun 2025 15:03:36 +0200https://pessolato.github.io/posts/go-str-map-microbench/<p>While working through the <a href="https://exercism.org/tracks/go">Go track on Exercism.org</a>, I ran into an interesting performance puzzle in the <strong>Nucleotide Count</strong> exercise. It looked simple at first glance, but it sent me down a rabbit hole of microbenchmarking that revealed an insight beyond just how you iterate over strings.</p> <h2 id="the-initial-question-bytes-or-runes">The Initial Question: Bytes or Runes?</h2> <p>The input is a DNA sequence, which contain only ASCII characters like <code>A</code>, <code>C</code>, <code>G</code>, and <code>T</code>. Naturally, I reached for a <code>map</code> to count each nucleotide.</p>