https://aeshift.com/tags/memory-management/ Recent content in Memory-Management on aeshift Hugo -- gohugo.io en &copy; 2026 [Dachary Carey](https://dacharycarey.com) - with agent assistance · Part of the [Agent Ecosystem Research Program](https://agentecosystem.dev) Wed, 11 Mar 2026 00:00:00 +0000 https://aeshift.com/posts/2026-03-11-the-missing-memory-hierarchy-demand-paging-for-llm-context-windows/ Wed, 11 Mar 2026 00:00:00 +0000 https://aeshift.com/posts/2026-03-11-the-missing-memory-hierarchy-demand-paging-for-llm-context-windows/ <p>If you&rsquo;re still trying to &ldquo;fit the prompt,&rdquo; you&rsquo;re solving the wrong problem. The right move is to treat the context window like cache and build paging, because that&rsquo;s what it is. <a href="http://arxiv.org/abs/2603.09023v1" target="_blank" rel="noreferrer">&ldquo;The Missing Memory Hierarchy&rdquo;</a> makes that argument plainly, then backs it with production numbers that are hard to ignore: 21.8% of tokens are structural waste, and a demand-paging proxy cut context consumption by up to 93% with a tiny fault rate. That&rsquo;s not prompt engineering; that&rsquo;s systems engineering.</p>