<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/"><channel><title>NIXL on AI 实战派 · 从技术到赚钱</title><link>https://guijiagi.com/tags/nixl/</link><description>Recent content in NIXL on AI 实战派 · 从技术到赚钱</description><generator>Hugo</generator><language>zh-cn</language><copyright>本站内容采用 CC BY-NC-SA 4.0 国际许可协议授权</copyright><lastBuildDate>Fri, 09 Oct 2026 16:48:00 +0800</lastBuildDate><atom:link href="https://guijiagi.com/tags/nixl/index.xml" rel="self" type="application/rss+xml"/><item><title>当 P/D 分离遇上混合 SSM：vLLM 如何搬运 Mamba 卷积状态</title><link>https://guijiagi.com/posts/2026-10-09-19-vllm-hybrid-ssm-nixl/</link><pubDate>Fri, 09 Oct 2026 16:48:00 +0800</pubDate><guid>https://guijiagi.com/posts/2026-10-09-19-vllm-hybrid-ssm-nixl/</guid><description>vLLM 把 NIXL 预填充与解码分离扩展到混合 SSM 注意力模型，靠双描述符视图与物理逻辑块桥接解决异构缓存布局。</description></item></channel></rss>