<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Load Bearing</title><link>https://blog.negmbots.com/</link><description>Recent content on Load Bearing</description><generator>Hugo</generator><language>en</language><lastBuildDate>Fri, 20 Nov 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://blog.negmbots.com/index.xml" rel="self" type="application/rss+xml"/><item><title>Is AI latency really fixed? Taking NCM for a spin</title><link>https://blog.negmbots.com/field-tests/2026-11-ncm-on-amd-gpu/</link><pubDate>Fri, 20 Nov 2026 00:00:00 +0000</pubDate><guid>https://blog.negmbots.com/field-tests/2026-11-ncm-on-amd-gpu/</guid><description>Yes, NCM&amp;rsquo;s latency is fixed, but because every answer has the same length, not because of its new method. That buys a cost you can forecast, and it costs quality on long inputs.</description></item><item><title>My lab: the hardware behind the posts</title><link>https://blog.negmbots.com/side-projects/2026-11-my-lab/</link><pubDate>Sun, 01 Nov 2026 00:00:00 +0000</pubDate><guid>https://blog.negmbots.com/side-projects/2026-11-my-lab/</guid><description>The machines behind every field test on this blog, so later posts can just link here.</description></item><item><title>About</title><link>https://blog.negmbots.com/about/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://blog.negmbots.com/about/</guid><description>&lt;p&gt;I’m Mohamed Negm, a staff engineer in on-prem infrastructure and platform organizations at cloud companies. I’ve led many initiatives and delivered products from hardware-specific offerings to AI solutions, with a fair share spent on sovereign AI platforms for EU customers.&#10;I’m drawn to the trade-offs where workloads put hard demands on the platform underneath; that’s where I get hooked, and where I find the fun in my work.&lt;/p&gt;&#10;&lt;h2 id="what-youll-find-here"&gt;What you&amp;rsquo;ll find here&lt;/h2&gt;&#10;&lt;ul&gt;&#10;&lt;li&gt;&lt;strong&gt;Perspectives:&lt;/strong&gt; my arguments about platform and infrastructure (cost models, architecture, teams).&lt;/li&gt;&#10;&lt;li&gt;&lt;strong&gt;Field tests:&lt;/strong&gt; I take a new technology for a spin on my own hardware, starting from a question I would ask in my daily work, and end up with what I&amp;rsquo;d decide. The measurements are at the bottom of each one, for anyone who cares about details.&lt;/li&gt;&#10;&lt;li&gt;&lt;strong&gt;Side projects:&lt;/strong&gt; things I build in my homelab, for fun and to learn.&lt;/li&gt;&#10;&lt;/ul&gt;&#10;&lt;h2 id="how-i-work"&gt;How I work&lt;/h2&gt;&#10;&lt;ul&gt;&#10;&lt;li&gt;Where I can, I test before I form an opinion, and I say so when I can&amp;rsquo;t.&lt;/li&gt;&#10;&lt;li&gt;Numbers are reported as measured, including failures.&lt;/li&gt;&#10;&lt;li&gt;Opinions are my own.&lt;/li&gt;&#10;&lt;/ul&gt;&#10;&lt;h2 id="the-lab"&gt;The lab&lt;/h2&gt;&#10;&lt;p&gt;I have many components I collected and played with over the years; these are the main ones I use and their roles:&lt;/p&gt;</description></item></channel></rss>