<?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>CUDA on 硅基 AGI · 智能体学习与测评</title><link>https://guijiagi.com/tags/cuda/</link><description>Recent content in CUDA on 硅基 AGI · 智能体学习与测评</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>Thu, 02 Jul 2026 10:54:00 +0800</lastBuildDate><atom:link href="https://guijiagi.com/tags/cuda/index.xml" rel="self" type="application/rss+xml"/><item><title>Flash Attention 3原理解析</title><link>https://guijiagi.com/posts/flash-attention-3-explained/</link><pubDate>Thu, 02 Jul 2026 10:54:00 +0800</pubDate><guid>https://guijiagi.com/posts/flash-attention-3-explained/</guid><description>从Flash Attention 1到3，解析注意力计算的性能极限突破之路</description></item><item><title>Flash Attention 3 原理：GPU 内存层次的最优利用</title><link>https://guijiagi.com/posts/flash-attention-3-principles/</link><pubDate>Sun, 28 Jun 2026 11:05:00 +0800</pubDate><guid>https://guijiagi.com/posts/flash-attention-3-principles/</guid><description>深入解析Flash Attention 3的算法原理、GPU硬件协同优化与2026年最新性能基准</description></item></channel></rss>