Illumos had a really nice stack for running containers inside jails and zones... I wonder if any of that ever made it into the linux world. If you broke out of the container you'd just be inside a jail which is even more hardened.
It'll be interesting to see if they leverage any new hardware features I've heard of in the past 3 years. One is AMD got the feature that allows PCIe to be DMA'd directly to L2|L3 working. Another is something about PCIe devices being able to talk "directly" to each other. Pretty sure it's actually mediated by the PCIe controller though.
Alas, this is beyond any level I expect to ever see 1st hand, so I didn't retain more detail than I've given.
Here's another one: golf courses across the US are estimated to use 2 billion gallons per DAY.
700,000 gallons per acre per growing season for Corn, need to look up cover crop water for a per year figure.
500-2000gallons per pound of beef- and usda estimates place domestic production at 27Billion pounds per year.
We should be good stewards of our clean water (aquafers probably shouldn't be used unless they are of the self-filling variety), nor should down-river be deprived of their share. It's just Water use for forced convection evaporative cooling is not that much in the grand scheme, and most of it is used at the power plant rather than the DC.
The beef water quote is assuming the cattle are being fed irrigated crops. If you graze cattle on land you don't irrigate or feed cattle corn in places where you don't irrigate did you use thousands of gallons per animal or zero?
>700,000 gallons per acre per growing season for Corn, need to look up cover crop water for a per year figure.
You rewrote this comment, again this is very misleading.
My family has grown corn for 140ish years. In that time we have used exactly 0 gallons of water to grow corn. We don't irrigate, we don't have the mechanisms to irrigate, nobody in 100 miles irrigates (they do in far western Iowa and Nebraska and other marginal-to-grow-corn places) 80% of the corn grown in the US is grown with no irrigation.
It rains. That's it.
Even people who irrigate don't use that much water, they supplement the rain.
700,000 gallons per acre per year is the amount of water you'd have to use to grow corn in the desert which nobody does (I'm sure one or two crazies can be found on very small scales)
It is a disingenuous argument made either out of ignorance or manipulation.
I don't think 2 feet of irrigation per year is an outrageous claim, certainly you see that in western Kansas, and maybe half that amount in Nebraska. In California we need 2-3 feet per year, but California only has a small fraction of corn land compared to the corn belt states. And yes, that is stupid, but isn't that the point of this thread? People do outrageous stuff with water all over America to a far larger extent than anyone is proposing to do with data centers.
I'm somewhere south-ish in the US- We get the same amount of rain, but in torrents over a couple hours every 6 weeks with triple digit temperatures in between.
So, nothing disingenuous or even really ignorant on my part- just the idea that consistent rainfall can be trusted so much that crop failure due to no rain is an acceptable risk is alien enough it didn't cross my mind... Especially when I was mainly trying to put it in perspective. People space out when Billions come into play. So, I find comparisons with a factor <100 more digestible, and thought I'd share.
You're missing the point. We don't irrigate. 80% of corngrowers don't irrigate.
We don't have the irrigation hardware, we don't have the wells to irrigate, there is no irrigation.
It's an outrageous claim because it's false.
The folks who do irrigate corn more than a foot per year shouldn't be allowed. It is a huge unnecessary waste of water and we aren't lacking for corn production.
At least your stance on this topic is internally consistent, setting you apart from most of the commentariat. For that, I salute you.
As you have established, eastern Nebraska and western Iowa have an abundance of rain, surface water, and shallow aquifers. Would you then agree that it is a perfectly appropriate place for data centers?
It really feels like this article is trying to force the narrative that "reductionism"(a few simple rules) and "emergence"(complicated results) are oppositional and expects its wordiness to be convincing.
This is why I think Conway's Game of Life is beautiful, as a toy model of simple rules giving rise to spectacular complexity. And how the "resolution" at which it's able to achieve replicators and factories is impressively basic, and if you grant that you can scale up several orders of magnitude from the game of life to entire systems of interacting "factories", it's not a stretch to see how those mechanisms can make up something like cells that drive biology.
A Conway "glider" has been found in nature(!!) [1], and if you combine that with an intuition from the "Size of Life" model also recently posted to hn [2], it's easy to get an intuition for how simple principles can lead to astonishing complexity. Even just scaling from DNA to Mitochondria is scaling up in size 200x. An amoeba is another 300x scaling up and if my chatGPTing is right, scaling up to a mouse is like 63,000x larger than a mitochondrion.
If you imagine zooming out from a grid of dots, like zooming out from Google Maps from a street level to planet level, there's plenty of room for physical interaction to work its magic. You just have to have to be able to comprehend the staggering scales. I worry that the "more is different" crowd haven't really appreciated the staggering scales involved or get lost in the act of romanticising it that "it can't be reductionist" is a lost persons way of saying they appreciate complexity.
Edit: Conway's glider was not found in nature, my mistake. But for a different example serving the same purpose, cellular automata modeling a "triforce" pattern of repeating triangles has been found in nature, off the top of my head.
I do not grasp the spite put forth whenever a LLM is used for a non-critical point in an argument.
1) LLMs have the potential to, on the whole, raise the base "correctness" of people's opinions. (Ex: asking Gemini flash why seed oils are unhealthy begines with "that's a highly debated and complex topic." Which, is infinitely better than believing some short form shopping channel)
2) They offer a softer emotional impact when it's inevitably corrected, making for a less toxic environment, and increasing the odds a topic will be discussed and possibly an opinion corrected.
3) more often than not, I've found them to not only be correct- but they'll offer nuance in the answer. Ex: mitochondria aren't the same size across everything that has them.
We should be pointing out flawed usage, not a wholesale assault on all usage.
I'm perfectly comfortable saying that I don't have an intuitive grasp of the conversion between nanometers, micrometers, and inches, which represent length according to different units, across different magnitudes, and especially in a context where I'm switching between talk of scales in terms of orders of magnitude and in terms of multiples (e.g. 100x is also just two orders of magnitude).
Honestly I find it kind of nuts to insist that knowing the difference in size ratio between a mitochondria and a mouse off the top of your head and to convert between them in different units is something the average person knows but whatever, maybe everyone knows that.
You're right that I was wrong about the glider but it's kind of a pennywise, pound foolish argument in this case if you think the upshot is supposed to be a blanket denial of cellular automata ever being reflected in nature. Off the top of my head I know the triforce repeating triangle pattern, perhaps the most famous cellular automata model, is found on seashells as one of many examples, so the point stands that we find in nature remarkable similarities to structures modeled in cellular automata.
Just so people this far down can look it up the term is Canonicalization, and its cousin collation.
These are non-trivial issues that, thankfully, some very smart and/or experienced people have usually handled for us. However, they still frequently lead to all sorts of vulnerabilities. "Stuffing" attacks sometimes rely on these issues, as have several major crypto incidents.
Doing some basic research... hard water is overwhelmingly various carbonate and bicarbonates of magnesium, calcium, sulfur, iron, maganese, and aluminum. All of which are essential nutrients and readily soluable in water.
The other proposed problem was pathogen aerosols- however I was unable to access anything but an abstract. So, I don't know if they survived being aerosolized, produced more and/or worse pathogen count than evaporative humidifiers, Nor the size of the pathogens.
It seems to me the known risk is mostly mechanical (Asthma, exacerbated COPD, etc) and nonpersistent (particles dissolve and are used or excreted via the same pathways as when consumed). With an unknown risk on the pathogen side.
reply