<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0"><channel><title><![CDATA[Mac Studio m5u 512G的mlx-serve benchmark-Deepseek v4.1]]></title><description><![CDATA[<p dir="auto">设备：Mac Studio m5u 512G<br />
使用mlx-serve，<br />
模型选用DeepSeek-V4.1-Flash-1017-oQ4e-ngram-mtp<br />
<img src="https://upload.lcz.me/uploads/50241bc9-21cb-46aa-84a6-0aa324708581.png" alt="Screenshot 2026-11-16 at 9.33.41 AM.png" class="img-markdown  img-fluid" /></p>
]]></description><link>https://lcz.me/topic/3109</link><generator>RSS for Node</generator><lastBuildDate>Thu, 19 Nov 2026 00:16:40 GMT</lastBuildDate><atom:link href="https://lcz.me/topic/3109.rss" rel="self" type="application/rss+xml"/><pubDate>Mon, 16 Nov 2026 01:37:02 GMT</pubDate><ttl>60</ttl><item><title><![CDATA[Reply to Mac Studio m5u 512G的mlx-serve benchmark-Deepseek v4.1 on Mon, 16 Nov 2026 02:41:04 GMT]]></title><description><![CDATA[<p dir="auto"><a class="plugin-mentions-user plugin-mentions-a" href="/user/kop-wang" aria-label="Profile: kop-wang">@<bdi>kop-wang</bdi></a> 你说的很有道理，确实，无论spark还是mac，甚至aimax395等方案都是解决最低代价跑起来的问题。从你提的角度来看，无论mac还是spark都还是很有性价比的。还是专注于应用吧，硬件看起来除非捡垃圾，不然方案性价比大厂都算得很清楚，都不吃亏。</p>
]]></description><link>https://lcz.me/post/33677</link><guid isPermaLink="true">https://lcz.me/post/33677</guid><dc:creator><![CDATA[junyi gu]]></dc:creator><pubDate>Mon, 16 Nov 2026 02:41:04 GMT</pubDate></item><item><title><![CDATA[Reply to Mac Studio m5u 512G的mlx-serve benchmark-Deepseek v4.1 on Mon, 16 Nov 2026 02:24:18 GMT]]></title><description><![CDATA[<p dir="auto"><a class="plugin-mentions-user plugin-mentions-a" href="/user/junyi-gu" aria-label="Profile: junyi-gu">@<bdi>junyi-gu</bdi></a> 肯定是的，4倍价格换来10倍的prefill + 5倍的decode，以及更强的并发能力，如果从总账来讲肯定是显卡更划算。</p>
<p dir="auto">统一内存方案，其实解决的就是能不能在最低的成本下跑起来这个问题。</p>
]]></description><link>https://lcz.me/post/33676</link><guid isPermaLink="true">https://lcz.me/post/33676</guid><dc:creator><![CDATA[kop wang]]></dc:creator><pubDate>Mon, 16 Nov 2026 02:24:18 GMT</pubDate></item><item><title><![CDATA[Reply to Mac Studio m5u 512G的mlx-serve benchmark-Deepseek v4.1 on Mon, 16 Nov 2026 02:15:13 GMT]]></title><description><![CDATA[<p dir="auto"><a class="plugin-mentions-user plugin-mentions-a" href="/user/kop-wang" aria-label="Profile: kop-wang">@<bdi>kop-wang</bdi></a> 多并发mac应该不占优，对于少数人使用的场景，mac的质感还是好的。如果多人使用可能spark不错，但是正经多人的话，会不会考虑pro6000部署更合适。</p>
]]></description><link>https://lcz.me/post/33674</link><guid isPermaLink="true">https://lcz.me/post/33674</guid><dc:creator><![CDATA[junyi gu]]></dc:creator><pubDate>Mon, 16 Nov 2026 02:15:13 GMT</pubDate></item><item><title><![CDATA[Reply to Mac Studio m5u 512G的mlx-serve benchmark-Deepseek v4.1 on Mon, 16 Nov 2026 01:52:15 GMT]]></title><description><![CDATA[<p dir="auto">不错的结果，这样的数据结合价格（17.3万），才有竞争力。毕竟GB10 * 4 + 交换机，价格也差不多。性能也差不多。</p>
]]></description><link>https://lcz.me/post/33668</link><guid isPermaLink="true">https://lcz.me/post/33668</guid><dc:creator><![CDATA[kop wang]]></dc:creator><pubDate>Mon, 16 Nov 2026 01:52:15 GMT</pubDate></item></channel></rss>