<?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>KV缓存 on AI 实战派 · 从技术到赚钱</title><link>https://guijiagi.com/tags/kv%E7%BC%93%E5%AD%98/</link><description>Recent content in KV缓存 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 16:00:00 +0800</lastBuildDate><atom:link href="https://guijiagi.com/tags/kv%E7%BC%93%E5%AD%98/index.xml" rel="self" type="application/rss+xml"/><item><title>FP8 KV Cache：用精度换 HBM 容量，这笔账怎么算</title><link>https://guijiagi.com/posts/2026-10-09-11-fp8-kv-cache-capacity/</link><pubDate>Fri, 09 Oct 2026 16:00:00 +0800</pubDate><guid>https://guijiagi.com/posts/2026-10-09-11-fp8-kv-cache-capacity/</guid><description>vLLM V1 已支持 FP8 KV Cache。把 KV 从 BF16 压到 FP8，显存占用直接减半，但要付出精度代价。本文给换算公式与取舍建议。</description></item><item><title>llm-d：让路由层「认识」KV 缓存，prefix 命中不再靠运气</title><link>https://guijiagi.com/posts/2026-10-09-09-llmd-prefix-aware-routing/</link><pubDate>Fri, 09 Oct 2026 15:48:00 +0800</pubDate><guid>https://guijiagi.com/posts/2026-10-09-09-llmd-prefix-aware-routing/</guid><description>llm-d 读取 vLLM 的 KV-cache 事件做 prefix-aware 路由，并用全局索引与 CPU、SSD 分层卸载管理分布式 KV 缓存。</description></item></channel></rss>