<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>SlimTCP: It's fast, but not because it's slim · HackerPapers discussions</title><link>https://hackerpaper.kaipfister.de/p/2608.25834/</link><description>Collected HN comments and submissions linking this paper.</description><language>en</language><lastBuildDate>Mon, 02 Nov 2026 11:16:22 GMT</lastBuildDate><atom:link href="https://hackerpaper.kaipfister.de/p/2608.25834/feed.xml" rel="self" type="application/rss+xml" /><item><title>SlimTCP: It's fast, but not because it's slim — mentioned by CajuMD</title><link>https://news.ycombinator.com/item?id=49574906</link><guid isPermaLink="false">urn:hackerpapers:mention:2608.25834:49574906</guid><pubDate>Sat, 05 Sep 2026 09:49:17 GMT</pubDate><description>&lt;p&gt;SlimTCP is a TCP/IP stack written to answer the question: Would a trimmed-down TCP stack run faster in modern datacenters(Ultra Ethernet)? According to my benchmarks, it achieves twice the goodput of mTCP on one connection. It also maintains 99.5% of its goodput on 4096 connections, while mTCP maintains 95%. Yet, I&amp;#x27;m still uncertain whether it&amp;#x27;s faster because it&amp;#x27;s simpler or because I overworked it. Paper: https://arxiv.org/abs/2608.25834 Benchmarks: https://github.com/CajuM/eqds-cloudlab&lt;/p&gt;&lt;p&gt;&lt;a href="https://hackerpaper.kaipfister.de/p/2608.25834/"&gt;Paper and discussion history&lt;/a&gt;. A mention is not an endorsement.&lt;/p&gt;</description><category>cs.NI</category></item></channel></rss>
