TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

DeepSeek R1 671B running on 2 M2 Ultras faster than reading speed

96 点作者 thyrox4 个月前

9 条评论

mythz4 个月前
Someone also got the full Q8 R1 running on a $6K PC without a GPU on 2x EPYC with 768GB DDR5 RAM running at 6-8 tok&#x2F;s [1].<p>Will be interesting to see the value&#x2F;performance compared to next gen M4 Ultra&#x27;s (or Extreme?) vs NVIDIA&#x27;s new DIGITS [2] when they&#x27;re released.<p>[1] <a href="https:&#x2F;&#x2F;x.com&#x2F;carrigmat&#x2F;status&#x2F;1884244369907278106" rel="nofollow">https:&#x2F;&#x2F;x.com&#x2F;carrigmat&#x2F;status&#x2F;1884244369907278106</a><p>[2] <a href="https:&#x2F;&#x2F;www.nvidia.com&#x2F;en-us&#x2F;project-digits&#x2F;" rel="nofollow">https:&#x2F;&#x2F;www.nvidia.com&#x2F;en-us&#x2F;project-digits&#x2F;</a>
评论 #42861550 未加载
评论 #42861646 未加载
评论 #42861366 未加载
评论 #42861692 未加载
danans4 个月前
Check out the power draw metrics. Following the CPU+GPU power consumption, it seems like it averaged 22W for about a minute. Unless I&#x27;m missing something, the inference for this example consumed at most .0004 kWh.<p>That&#x27;s almost nothing. If these models are capable&#x2F;functional enough for most day-to-day uses, then useful LLM-based GenAI is already at the &quot;too cheap to meter&quot; stage.
评论 #42880181 未加载
teruakohatu4 个月前
I am amazed mlx-lm&#x2F;mlx.distributed works that well on prosumer hardware.<p>I don&#x27;t think they specified what they were using for networking, but it was probably Thunderbolt&#x2F;USB4 networking which can reach 40Gbps.
shihab4 个月前
Please note that it’s using pretty aggressive quantization (around 4 bits per weight)
评论 #42861209 未加载
rashidae4 个月前
This is amazing!! What kind of applications are you considering for this? A part from saving variable costs, fine tuning extensively and security… I’m curious to evaluate this in a financial perspective, as variable costs can be daunting, but not too much “yet”.<p>I’m hoping NVIDIA comes up with their new consumer computer soon!
iFred4 个月前
Complete aside, but I think this is the first time I’ve seen Apple’s internal DNS outside of Apple.
评论 #42861305 未加载
creativenolo3 个月前
How is this split between two computers?
DrNosferatu4 个月前
Heavily quantized…<p>Still interesting though.
mrcwinn4 个月前
Fascinating to read the thinking process of a flush vs a straight in poker. It&#x27;s circular nonsense that is not at all grounded in reason — it&#x27;s grounded in the factual memory of the rules of Poker, repeated over and over as it continues to doubt itself and double-check. What nonsense!<p>How many additional nuclear power plants will need to be built because even these incredibly technical achievements are, under the hood, morons? XD