<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>Jason Kessler — Writing</title>
    <link>https://kitsapdata.com/blog/</link>
    <atom:link href="https://kitsapdata.com/blog/rss.xml" rel="self" type="application/rss+xml"/>
    <description>Notes on agentic infrastructure, context engineering, and enterprise security.</description>
    <language>en-us</language>
    <item>
      <title>miopenStatusUnknownError isn&#39;t a GPU limitation. It&#39;s a missing header.</title>
      <link>https://kitsapdata.com/blog/miopen-unknown-error-is-a-missing-header/</link>
      <guid isPermaLink="true">https://kitsapdata.com/blog/miopen-unknown-error-is-a-missing-header/</guid>
      <pubDate>Tue, 28 Jul 2026 12:00:00 GMT</pubDate>
      <description>A batch_norm call kept failing on my AMD Strix Halo iGPU, so I moved the workload to CPU. Nine days later I found the real cause: the container had no C++ standard headers. The GPU was fine the whole time.</description>
      <category>rocm</category>
      <category>gpu</category>
      <category>docker</category>
      <category>debugging</category>
    </item>
  </channel>
</rss>
