Mastodon 🐘

Kevin Karhan :verified:I like <a href="https://bitbang.social/@ActionRetro" class="u-url mention" rel="nofollow noopener" target="_blank">@ActionRetro</a> taking a look at the <a href="https://infosec.space/tags/Intel" class="mention hashtag" rel="nofollow noopener" target="_blank">#Intel</a> <a href="https://infosec.space/tags/ComputeStick" class="mention hashtag" rel="nofollow noopener" target="_blank">#ComputeStick</a>... <ul><li>Personally I still like the <a href="https://infosec.space/tags/fanless" class="mention hashtag" rel="nofollow noopener" target="_blank">#fanless</a> versions of these as they are pretty nifty and I'm propably gonna recommend one for my "Cheapest Desktop you can buy" challenge instead of a <a href="https://infosec.space/tags/Pi0W" class="mention hashtag" rel="nofollow noopener" target="_blank">#Pi0W</a>... </li></ul>But I'm confident that these can be useable with <a href="https://linuxrocks.online/@bunsenlabs" class="u-url mention" rel="nofollow noopener" target="_blank">@bunsenlabs</a> or other <a href="https://infosec.space/tags/FrugalComputing" class="mention hashtag" rel="nofollow noopener" target="_blank">#FrugalComputing</a> <a href="https://infosec.space/tags/Linux" class="mention hashtag" rel="nofollow noopener" target="_blank">#Linux</a> distros...<ul><li>Maybe try <a href="https://infosec.space/@OS1337" class="u-url mention" rel="nofollow noopener" target="_blank">@OS1337</a> ? </li></ul><a href="https://www.youtube.com/watch?v=G3WvOzdlpwY" rel="nofollow noopener" translate="no" target="_blank">https://www.youtube.com/watch?v=G3WvOzdlpwY</a>

Jonathan Schofield<a href="https://mastodon.social/@grimalkina" class="u-url mention" rel="nofollow noopener" target="_blank">@grimalkina</a> perhaps <a href="https://scholar.social/@wim_v12e" class="u-url mention" rel="nofollow noopener" target="_blank">@wim_v12e</a> is in the right ballpark? More generally, maybe check out <a href="https://mastodon.social/tags/frugalComputing" class="mention hashtag" rel="nofollow noopener" target="_blank">#frugalComputing</a> <a href="https://mastodon.social/@urlyman/114223143337500275" rel="nofollow noopener" translate="no" target="_blank">https://mastodon.social/@urlyman/114223143337500275</a>

Wim🧮The paper by @maswan and myself "Life Cycle Analysis for Emissions of Scientific Computing Centres" has been published!We develop a detailed model for the LCA of (HPC) data centres, including embodied carbon, server replacement and expansion. It is also applicable to other data centres. We also share the source code.It shows how important embodied carbon becomes when the grid has more renewables.<a href="https://link.springer.com/article/10.1140/epjc/s10052-025-14650-8" rel="nofollow noopener" translate="no" target="_blank">https://link.springer.com/article/10.1140/epjc/s10052-025-14650-8</a><a href="https://scholar.social/tags/FrugalComputing" class="mention hashtag" rel="nofollow noopener" target="_blank">#FrugalComputing</a>

Kevin Karhan :verified:<a href="https://tech.lgbt/@Lydie" class="u-url mention" rel="nofollow noopener" target="_blank">@Lydie</a> guess why I want to push for <a href="https://infosec.space/tags/FrugalComputing" class="mention hashtag" rel="nofollow noopener" target="_blank">#FrugalComputing</a> and work on <a href="https://infosec.space/@OS1337" class="u-url mention" rel="nofollow noopener" target="_blank">@OS1337</a> ?

Kevin Karhan :verified:<a href="https://fosstodon.org/@cleantext" class="u-url mention" rel="nofollow noopener" target="_blank">@cleantext</a> well, it depends on what your productivity tool is.<ul><li>99% of my productivity is in <a href="https://infosec.space/tags/SSH" class="mention hashtag" rel="nofollow noopener" target="_blank">#SSH</a> so any <a href="https://infosec.space/tags/i486" class="mention hashtag" rel="nofollow noopener" target="_blank">#i486</a> "<a href="https://infosec.space/tags/shitbox" class="mention hashtag" rel="nofollow noopener" target="_blank">#shitbox</a>" with <a href="https://infosec.space/tags/SSE2" class="mention hashtag" rel="nofollow noopener" target="_blank">#SSE2</a> acceleration is gonna be sufficiently fast (I could propably work from a <a href="https://infosec.space/tags/Pi0W" class="mention hashtag" rel="nofollow noopener" target="_blank">#Pi0W</a> with screen & keyboard attached.</li></ul>It's also an exercise in <a href="https://infosec.space/tags/FrugalComputing" class="mention hashtag" rel="nofollow noopener" target="_blank">#FrugalComputing</a> and asking "What do I actually need to be able to work?" and the answer is pretty simple: An SSH-<a href="https://youtube.com/watch?v=RuZUPpmXfT0&t=3m15s" rel="nofollow noopener" target="_blank">Terminal</a>!<ul><li>My mere customizations would be to do like a <a href="https://infosec.space/@kkarhan/115073462064349843" rel="nofollow noopener" target="_blank">"status bar"…</a></li></ul>

Wim🧮At the risk of restating the obvious: the AI companies are spending $100 billion in a single quarter on data centres because they plan to use them at capacity to maximise profits. Therefore the actual energy and water cost of an individual query is irrelevant: the purpose of installing >10 GW of new data centre capacity per year globally is to use it fully. Ditto for the generation capacity needed to power them: it will be used, if not for data centres, then for something else.<a href="https://scholar.social/tags/FrugalComputing" class="mention hashtag" rel="nofollow noopener" target="_blank">#FrugalComputing</a>

Wim🧮The purpose of Google's paper on energy and water consumption of their LLMs is to absolve individual users of their responsibility for the overall emissions and water usage. That way, they hope adoption will rise or at least not drop. The need the adoption to justify continued huge investments in data centres. And the damage of this investment is done even if the AI bubble would burst tomorrow. <a href="https://limited.systems/articles/the-real-problem-with-AI/" rel="nofollow noopener" translate="no" target="_blank">https://limited.systems/articles/the-real-problem-with-AI/</a> <a href="https://limited.systems/articles/the-insatiable-hunger-of-openai/" rel="nofollow noopener" translate="no" target="_blank">https://limited.systems/articles/the-insatiable-hunger-of-openai/</a><a href="https://scholar.social/tags/FrugalComputing" class="mention hashtag" rel="nofollow noopener" target="_blank">#FrugalComputing</a>

Wim🧮My paper with @maswan "Life Cycle Analysis for Emissions of Scientific Computing Centres" has been accepted fro publication in European Physical Journal C !We develop a detailed model for the LCA of (HPC) data centres, including embodied carbon, server replacement and expansion. It is also applicable to other data centres. We also share the source code.It shows how important embodied carbon becomes when the grid has more renewables.<a href="https://arxiv.org/abs/2506.14365" rel="nofollow noopener" translate="no" target="_blank">https://arxiv.org/abs/2506.14365</a><a href="https://scholar.social/tags/FrugalComputing" class="mention hashtag" rel="nofollow noopener" target="_blank">#FrugalComputing</a>

Wim🧮I'm doing a deep dive into the embodied carbon of GPU servers for AI training and inference, such as the nvidia DGX-A100 and DGX-B300. I was surprised to find that the embodied carbon is dominated by the RAM (i.e it's more than 50% of the total). And the big difference between the DGX-A100 (2020) and the DGX-B300 (2025) is the amount of RAM. The embodied carbon of that RAM has more than doubled. Second largest contribution is the SSD, but that has not increased in size.<a href="https://scholar.social/tags/FrugalComputing" class="mention hashtag" rel="nofollow noopener" target="_blank">#FrugalComputing</a>

Kevin Karhan :verified:<a href="https://mastodon.social/@biglinter" class="u-url mention" rel="nofollow noopener" target="_blank">@biglinter</a> <a href="https://florp.social/users/snaki" class="u-url mention" rel="nofollow noopener" target="_blank">@snaki</a> yes but not really...More an exercise in <a href="https://infosec.space/tags/ninimalism" class="mention hashtag" rel="nofollow noopener" target="_blank">#ninimalism</a>, <a href="https://infosec.space/tags/FrugalComputing" class="mention hashtag" rel="nofollow noopener" target="_blank">#FrugalComputing</a> and <a href="https://infosec.space/tags/ReproduceableBuilds" class="mention hashtag" rel="nofollow noopener" target="_blank">#ReproduceableBuilds</a>!

Wim🧮<a href="https://mastodon.online/@frebelt" class="u-url mention" rel="nofollow noopener" target="_blank">@frebelt</a> My own estimate (<a href="https://wimvanderbauwhede.codeberg.page/articles/google-search-vs-chatgpt-emissions/" rel="nofollow noopener" translate="no" target="_blank">https://wimvanderbauwhede.codeberg.page/articles/google-search-vs-chatgpt-emissions/</a>) is between 0.002 kWh and 0.005 kWh per query. But that is for short queries (<100 words), and also this is for GPT-3.5. For GPT-4, it is likely to be 3x higher.So all in all I would say the estimate for water could be up to 10x larger than the figure in the study, so it could vary between 10 and 500 ml per query with a response of order of 100 words. I'd say a good ballpark figure would therefore be 50 ml per query.(2/2) <a href="https://scholar.social/tags/FrugalComputing" class="mention hashtag" rel="nofollow noopener" target="_blank">#FrugalComputing</a>

Wim🧮In short, not only are more chips made, the embodied carbon per chip is increasing as well. (3/3) <a href="https://scholar.social/tags/FrugalComputing" class="mention hashtag" rel="nofollow noopener" target="_blank">#FrugalComputing</a>

Wim🧮Unfortunately, this is not what happens. Instead, the chips scale in performance and in storage density as the industry wants to keep Moore's law alive. So the the increased density does not lead to reduced embodied emissions. In other words, the embodied emissions per die have increased by 3x in that period, despite the much higher densities. This is further compounded by the growth in wafer volume. Because of the AI hype, this growth is currently truly exponential. (2/3) <a href="https://scholar.social/tags/FrugalComputing" class="mention hashtag" rel="nofollow noopener" target="_blank">#FrugalComputing</a>

Wim🧮I looked into the growth in embodied carbon (emissions from manufacturing) of the chips, those CPUs, GPUs, RAM and SSDs that fuel the AI hype. For the same silicon area, the embodied carbon of chip production has grown 4x since 2010. At the same time, the density has increased about 100x.At first sight, this is wonderful: if the compute performance and storage capacity would have remained the same as in 2010, the embodied carbon of every chip sold would be 25x lower. (1/3) <a href="https://scholar.social/tags/FrugalComputing" class="mention hashtag" rel="nofollow noopener" target="_blank">#FrugalComputing</a>

Wim🧮So we are looking at a growth of close to 20% which follows the medium-range scenario set out by McKinsey as discussed in my post: <a href="https://wimvanderbauwhede.codeberg.page/articles/the-real-problem-with-AI/" rel="nofollow noopener" translate="no" target="_blank">https://wimvanderbauwhede.codeberg.page/articles/the-real-problem-with-AI/</a>(2/2)<a href="https://scholar.social/tags/FrugalComputing" class="mention hashtag" rel="nofollow noopener" target="_blank">#FrugalComputing</a> <a href="https://scholar.social/tags/AIHype" class="mention hashtag" rel="nofollow noopener" target="_blank">#AIHype</a> <a href="https://scholar.social/tags/ClimateEmergency" class="mention hashtag" rel="nofollow noopener" target="_blank">#ClimateEmergency</a>

Wim🧮This post shows a graph of capex for meta, google, microsoft and amazon <a href="https://mastodon.social/@samim/114957102999706720" rel="nofollow noopener" translate="no" target="_blank">https://mastodon.social/@samim/114957102999706720</a>In the last quarter, it was >$100 billion. I checked those numbers, and the order of magnitude is certainly correct.If all that goes to AI data centres, with a capex of $5 billion for a 500 MW data centre (from <a href="https://ravenscraig.co.uk/news/news-post/ravenscraig-plans-to-transform-former-steelworks-into-one-of-the-uks-largest-green-ai-data-centres/" rel="nofollow noopener" translate="no" target="_blank">https://ravenscraig.co.uk/news/news-post/ravenscraig-plans-to-transform-former-steelworks-into-one-of-the-uks-largest-green-ai-data-centres/</a>) that means 10 GW of data centre capacity. Over a year it will be at least double so 20 GW In 2023, global capacity was 55 GW. (1/2) <a href="https://scholar.social/tags/FrugalComputing" class="mention hashtag" rel="nofollow noopener" target="_blank">#FrugalComputing</a> <a href="https://scholar.social/tags/AIHype" class="mention hashtag" rel="nofollow noopener" target="_blank">#AIHype</a>

Earth NotesRSS Podcast Feed Inefficiency - Climate cost of handling feeds ineptly... <a href="https://mastodon.energy/tags/RSS" class="mention hashtag" rel="nofollow noopener" target="_blank">#RSS</a> <a href="https://mastodon.energy/tags/podcast" class="mention hashtag" rel="nofollow noopener" target="_blank">#podcast</a> <a href="https://mastodon.energy/tags/FrugalComputing" class="mention hashtag" rel="nofollow noopener" target="_blank">#FrugalComputing</a> <a href="https://mastodon.energy/tags/LowCarbonComputing" class="mention hashtag" rel="nofollow noopener" target="_blank">#LowCarbonComputing</a> <a href="https://mastodon.energy/tags/greenSoftware" class="mention hashtag" rel="nofollow noopener" target="_blank">#greenSoftware</a> - <a href="https://www.earth.org.uk/RSS-efficiency.html" rel="nofollow noopener" translate="no" target="_blank">https://www.earth.org.uk/RSS-efficiency.html</a>

Wim🧮This is why I write blog posts, as they reach a wider audience and therefore have more potential to change anything. But I know that a blog post does not carry the same weight as a scientific paper, even if it has exactly the same content (I have done the experiment). So that is how we lose.(5/n=5) <a href="https://scholar.social/tags/FrugalComputing" class="mention hashtag" rel="nofollow noopener" target="_blank">#FrugalComputing</a>

Wim🧮This gets me thinking: this is how we lose. The tech industry can push "solutions" like this carbon-aware computing without having to justify anything. But as an academic, if I want to contest them through a scientific publication, the standards have to be very high. So high that I can't meet them in practice. (3/n) <a href="https://scholar.social/tags/FrugalComputing" class="mention hashtag" rel="nofollow noopener" target="_blank">#FrugalComputing</a>

Wim🧮I am writing a paper explaining that carbon-aware computing (moving data centre load to when/where emissions are low) is a distraction. I had the draft reviewed by a colleague, and it is clear that the paper will be rejected for lack of rigour.I am a senior academic and I know about originality, significance and rigour. I know the paper is rigorous, but that is beside the points: the reviewers can easily criticise it for not being so, and that is what matters.<a href="https://scholar.social/tags/FrugalComputing" class="mention hashtag" rel="nofollow noopener" target="_blank">#FrugalComputing</a> (1/n)

Recent searches

Search options

Administered by:

Server stats:

#frugalcomputing