<?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>Program-Structure on 兜忆轩 | Cedric's Compiler Notes</title><link>https://douyixuan.github.io/tags/program-structure/</link><description>Recent content in Program-Structure on 兜忆轩 | Cedric's Compiler Notes</description><image><title>兜忆轩 | Cedric's Compiler Notes</title><url>https://douyixuan.github.io/images/papermod-cover.png</url><link>https://douyixuan.github.io/images/papermod-cover.png</link></image><generator>Hugo -- 0.146.0</generator><language>en-us</language><lastBuildDate>Sun, 18 Oct 2026 12:10:00 +0800</lastBuildDate><atom:link href="https://douyixuan.github.io/tags/program-structure/index.xml" rel="self" type="application/rss+xml"/><item><title>033 · 程序结构帮助生成，为何未必帮助规划？</title><link>https://douyixuan.github.io/posts/ai-silicon-papers/2026-10-18/</link><pubDate>Sun, 18 Oct 2026 12:10:00 +0800</pubDate><guid>https://douyixuan.github.io/posts/ai-silicon-papers/2026-10-18/</guid><description>PRISM 把 AST、CFG、DFG 编码后注入模型，用于 HLS pragma 优化。论文最值得讨论的不是结构信息是否有用，而是它为何可能改善代码生成，却未必改善规划；角色差异的实验又能支持多强的解释。</description></item></channel></rss>