<?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>Gpudirect on The Practical AI Builder</title>
    <link>https://aitechy226.github.io/tags/gpudirect/</link>
    <description>Recent content in Gpudirect on The Practical AI Builder</description>
    <generator>Hugo</generator>
    <language>en-us</language>
    <lastBuildDate>Mon, 10 Aug 2026 00:00:00 +0000</lastBuildDate>
    <atom:link href="https://aitechy226.github.io/tags/gpudirect/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>Beyond GPUDirect: Moving Storage Control to the GPU</title>
      <link>https://aitechy226.github.io/posts/scada-cufile-storage-control-path/</link>
      <pubDate>Mon, 10 Aug 2026 00:00:00 +0000</pubDate>
      <guid>https://aitechy226.github.io/posts/scada-cufile-storage-control-path/</guid>
      <description>NVIDIA is open-sourcing cuFile and formalizing SCADA — GPU-initiated storage where the GPU builds NVMe commands and polls completions in VRAM. What GPUDirect already fixed, what SCADA actually moves, and why the drives have to change too.</description>
    </item>
  </channel>
</rss>
