<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>吞吐优化 on LoopRouter Blog</title><link>https://blog.looprouter.com/tags/%E5%90%9E%E5%90%90%E4%BC%98%E5%8C%96/</link><description>Recent content in 吞吐优化 on LoopRouter Blog</description><generator>Hugo</generator><language>zh-cn</language><lastBuildDate>Tue, 08 Sep 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://blog.looprouter.com/tags/%E5%90%9E%E5%90%90%E4%BC%98%E5%8C%96/index.xml" rel="self" type="application/rss+xml"/><item><title>投机解码能把推理速度拉到 2.87 倍:一篇看懂它省的是谁的钱</title><link>https://blog.looprouter.com/docs/speculative-decoding-inference-cost/</link><pubDate>Tue, 08 Sep 2026 00:00:00 +0000</pubDate><guid>https://blog.looprouter.com/docs/speculative-decoding-inference-cost/</guid><description>&lt;blockquote&gt;&#10;&lt;p&gt;本文基于 vLLM 官方博客关于 AMD GPU 上投机解码的公开测试整理与二次分析,加入 LoopRouter 的成本/选型视角;原文见文末链接。文中数据反映特定硬件与配置,建议在你自己的负载上复测。&lt;/p&gt;</description></item></channel></rss>