<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: VOID</title>
    <description>The latest articles on DEV Community by VOID (@ecl_void).</description>
    <link>https://dev.to/ecl_void</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4246249%2F6314bba3-c35a-4217-9747-7850d2ff9170.jpg</url>
      <title>DEV Community: VOID</title>
      <link>https://dev.to/ecl_void</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/ecl_void"/>
    <language>en</language>
    <item>
      <title>Silicon Nural Network-</title>
      <dc:creator>VOID</dc:creator>
      <pubDate>Mon, 16 Nov 2026 06:07:11 +0000</pubDate>
      <link>https://dev.to/ecl_void/physical-nural-network--2f9b</link>
      <guid>https://dev.to/ecl_void/physical-nural-network--2f9b</guid>
      <description>&lt;p&gt;The core idea of this concept is building a silicon cheap with nural network in it. We use GPU for multiple short calculation. Why not have multiple calculators at a time instead? The idea is having separate calculating unit for each node of nural network.(ASIC) Why not build it if it's so good? Because we can't. With our current tech it's not possible. AI models high parameters is one reason but not the whole. Our silicon cheap tech can't convert this much parameters into silicone chip.. so it's technology limit? Yes. There's the other problem too. Complex and massive amount of nural connection between nodes of each layer. So theoretically it's a great idea but technically it's not possible.&lt;/p&gt;

</description>
      <category>ai</category>
      <category>deeplearning</category>
      <category>hardware</category>
    </item>
  </channel>
</rss>
