<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Xe3p on Best of AI</title><link>https://bestofai.io/tags/xe3p/</link><description>Recent content in Xe3p on Best of AI</description><generator>Hugo</generator><language>en-US</language><lastBuildDate>Sat, 05 Sep 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://bestofai.io/tags/xe3p/index.xml" rel="self" type="application/rss+xml"/><item><title>Intel Crescent Island</title><link>https://bestofai.io/hardware/intel-crescent-island/</link><pubDate>Sat, 05 Sep 2026 00:00:00 +0000</pubDate><guid>https://bestofai.io/hardware/intel-crescent-island/</guid><description>&lt;p&gt;Intel gave a deep technical rundown of its Crescent Island inference GPU at Hot Chips 2026 in August, following up its earlier disclosure of the chip. It packs 32 Xe3P cores, 256 XMX matrix engines, and 32MB of unified L2 cache, and leans on capacity-dense LPDDR5X rather than HBM: the standard air-cooled card ships with 160GB at 350W, with ODM designs supporting up to 480GB for customers who need to fit larger models without a memory-bandwidth-first design. Intel is targeting AI inference and agentic workloads specifically, trading some raw throughput for lower power and higher memory capacity per card. Customer sampling is expected in the second half of 2026, with a full launch slated for 2027.&lt;/p&gt;</description></item></channel></rss>