<?xml version="1.0" encoding="utf-8" standalone="yes"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>PagedAttention on 办公AI智能小助手</title>
    <link>https://blog.qife122.com/tags/pagedattention/</link>
    <description>Recent content in PagedAttention on 办公AI智能小助手</description>
    <generator>Hugo</generator>
    <language>zh-cn</language>
    <copyright>qife</copyright>
    <lastBuildDate>Wed, 01 Oct 2025 20:35:22 +0800</lastBuildDate>
    <atom:link href="https://blog.qife122.com/tags/pagedattention/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>使用vLLM优化大语言模型服务：从0.68到10请求/秒的性能飞跃</title>
      <link>https://blog.qife122.com/p/%E4%BD%BF%E7%94%A8vllm%E4%BC%98%E5%8C%96%E5%A4%A7%E8%AF%AD%E8%A8%80%E6%A8%A1%E5%9E%8B%E6%9C%8D%E5%8A%A1%E4%BB%8E0.68%E5%88%B010%E8%AF%B7%E6%B1%82/%E7%A7%92%E7%9A%84%E6%80%A7%E8%83%BD%E9%A3%9E%E8%B7%83/</link>
      <pubDate>Wed, 01 Oct 2025 20:35:22 +0800</pubDate>
      <guid>https://blog.qife122.com/p/%E4%BD%BF%E7%94%A8vllm%E4%BC%98%E5%8C%96%E5%A4%A7%E8%AF%AD%E8%A8%80%E6%A8%A1%E5%9E%8B%E6%9C%8D%E5%8A%A1%E4%BB%8E0.68%E5%88%B010%E8%AF%B7%E6%B1%82/%E7%A7%92%E7%9A%84%E6%80%A7%E8%83%BD%E9%A3%9E%E8%B7%83/</guid>
      <description>&lt;h2 id=&#34;llm推理成本&#34;&gt;LLM推理成本&lt;/h2&gt;&#xA;&lt;p&gt;现代GPU对于运行大语言模型至关重要。NVIDIA H100等企业级GPU价格高达3-4万美元。低效的内存管理会导致昂贵的GPU集群只能服务少量用户，而不是数十个用户。&lt;/p&gt;</description>
    </item>
  </channel>
</rss>
