<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Benchmarking on Shai Asher</title><link>https://slow-is-smooth.io/tags/benchmarking/</link><description>Recent content in Benchmarking on Shai Asher</description><generator>Hugo</generator><language>en-gb</language><lastBuildDate>Sat, 12 Sep 2026 11:54:00 +0300</lastBuildDate><atom:link href="https://slow-is-smooth.io/tags/benchmarking/index.xml" rel="self" type="application/rss+xml"/><item><title>Your profiler said dispatch. It did not say virtual.</title><link>https://slow-is-smooth.io/blog/your-profiler-said-dispatch/</link><pubDate>Sat, 12 Sep 2026 11:54:00 +0300</pubDate><guid>https://slow-is-smooth.io/blog/your-profiler-said-dispatch/</guid><description>&lt;p&gt;A profile says &lt;code&gt;dispatch&lt;/code&gt; is getting expensive at several million packets a second, and all four&#10;fixes the team proposes make the virtual call inside it cheaper. Measured, that call costs about&#10;10 nanoseconds of a 333-nanosecond budget, and the fastest alternative saves 1.7 of them. Finding&#10;which of a hundred handlers to call costs 75. None of the four proposals touches it.&lt;/p&gt;</description></item></channel></rss>