<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Asic on Ben's Notes</title><link>https://notes.ben.reynwar.net/tags/asic/</link><description>Recent content in Asic on Ben's Notes</description><generator>Hugo</generator><language>en-us</language><copyright>Ben Reynwar</copyright><lastBuildDate>Tue, 28 Jul 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://notes.ben.reynwar.net/tags/asic/index.xml" rel="self" type="application/rss+xml"/><item><title>Systolic Arrays for Matrix Multiplication</title><link>https://notes.ben.reynwar.net/systolic-arrays-for-matrix-multiplication/</link><pubDate>Tue, 28 Jul 2026 00:00:00 +0000</pubDate><guid>https://notes.ben.reynwar.net/systolic-arrays-for-matrix-multiplication/</guid><description>&lt;p&gt;Systolic arrays are a hardware architecture for efficiently performing matrix
multiplications. They consist of a grid of processing elements, with each
element multiplying two values and adding the result into an accumulated sum.&lt;/p&gt;
&lt;p&gt;The two dataflow organizations considered here are
sum-stationary, where the sum remains in the same processing element while the
multiplicands stream past, and weight-stationary, where one of the multiplicands
stays stationary while the other multiplicand and the sum stream past.&lt;/p&gt;</description></item></channel></rss>