<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>GPU Systems | Shaoyang Cui</title><link>https://spidermonk7.github.io/tags/gpu-systems/</link><atom:link href="https://spidermonk7.github.io/tags/gpu-systems/index.xml" rel="self" type="application/rss+xml"/><description>GPU Systems</description><generator>Hugo Blox Builder (https://hugoblox.com)</generator><language>en-us</language><lastBuildDate>Sun, 10 May 2026 00:00:00 +0000</lastBuildDate><image><url>https://spidermonk7.github.io/media/icon_hu7729264130191091259.png</url><title>GPU Systems</title><link>https://spidermonk7.github.io/tags/gpu-systems/</link></image><item><title>HelioX - A GPU-Native Framework for Simulation and Training of Biophysically Detailed Networks</title><link>https://spidermonk7.github.io/project/heliox/</link><pubDate>Sun, 10 May 2026 00:00:00 +0000</pubDate><guid>https://spidermonk7.github.io/project/heliox/</guid><description>&lt;p>Accepted at &lt;strong>ICML 2026&lt;/strong>.&lt;/p>
&lt;p>HelioX is a GPU-native framework for simulation and training of biophysically detailed neural networks. It targets the mismatch between irregular dendritic computation and conventional deep learning stacks, and provides an end-to-end pipeline for both forward simulation and gradient-based training.&lt;/p>
&lt;p>Highlights:&lt;/p>
&lt;ul>
&lt;li>Multi-stream GPU execution for ionic current calculation, ODE construction, and conductance update&lt;/li>
&lt;li>Efficient spike handling path designed for sparse spike events&lt;/li>
&lt;li>End-to-end differentiable training for deep biophysical neural architectures&lt;/li>
&lt;li>Strong runtime and scalability gains over traditional simulators in large-scale settings&lt;/li>
&lt;/ul></description></item></channel></rss>