<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Intel Arc on jota.log</title><link>https://jotalog.com/tags/intel-arc/</link><description>Recent content in Intel Arc on jota.log</description><generator>Hugo</generator><language>pt-BR</language><lastBuildDate>Sat, 12 Sep 2026 09:30:00 -0300</lastBuildDate><atom:link href="https://jotalog.com/tags/intel-arc/index.xml" rel="self" type="application/rss+xml"/><item><title>Dobrando a geração de tokens na Intel Arc Pro B65 com SYCL</title><link>https://jotalog.com/posts/arc-b65-sycl-llama-cpp/</link><pubDate>Sat, 12 Sep 2026 09:30:00 -0300</pubDate><guid>https://jotalog.com/posts/arc-b65-sycl-llama-cpp/</guid><description>Uma flag de build esquecida foi a diferença entre dobrar a geração de tokens do meu LLM local e regredir a performance usando SYCL numa Intel Arc B65.</description></item></channel></rss>