<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>算力优化 on AI 早报</title><link>https://ai-news.example.com/tags/%E7%AE%97%E5%8A%9B%E4%BC%98%E5%8C%96/</link><description>Recent content in 算力优化 on AI 早报</description><generator>Hugo</generator><language>zh-CN</language><lastBuildDate>Fri, 07 Aug 2026 01:34:02 +0800</lastBuildDate><atom:link href="https://ai-news.example.com/tags/%E7%AE%97%E5%8A%9B%E4%BC%98%E5%8C%96/feed.xml" rel="self" type="application/rss+xml"/><item><title>亿级日活App的算力生死劫：跨云架构如何砍掉75% GPU集群？</title><link>https://ai-news.example.com/articles/app75-gpu/</link><pubDate>Fri, 07 Aug 2026 01:34:02 +0800</pubDate><guid>https://ai-news.example.com/articles/app75-gpu/</guid><description>面对推理成本倒挂的严峻挑战，亿级日活应用通过创新的跨云架构成功削减了75%的GPU集群开销，为高并发AI应用的算力优化提供了新范式。</description></item><item><title>浪潮信息双线突围：单柜承载4万Agent与大模型组队解题</title><link>https://ai-news.example.com/articles/4agent/</link><pubDate>Mon, 20 Jul 2026 23:11:13 +0800</pubDate><guid>https://ai-news.example.com/articles/4agent/</guid><description>浪潮信息在智算领域推出两项创新：单机柜可支持高达4万个智能体（Agent）运行，同时实现多个大模型协同答题，显著提升推理效率。这标志着AI基础设施从“堆数量”向“要质量”的转变，为大规模Agent部署和复杂任务处理提供了新路径。</description></item></channel></rss>