Well, offhand, no one buys a cloud server for power efficiency, people buy into the cloud for reliability, or performance, or cost -- or some combination thereof.
If you're buying ultra-power, you're forgoing power-efficiency.
> no one buys a cloud server for power efficiency, people buy into the cloud for reliability, or performance, or cost -- or some combination thereof
Low cost and power efficiency are pretty much the same thing for a datacenter though, since cooling is the most expensive part. Hence e.g. AWS pushing Graviton.
I'm not saying you're wrong, but I would like to point out that a phone doesn't need active cooling. I just don't know if Graviton requires active cooling.
And I don't know if that statement is accurate, since passive cooling would be even better than active cooling from an efficiency perspective and particularly if you also consider reliability. Less moving parts would mean higher reliability unless, eg., the thermal paste evaporated on the cpu.
> I would like to point out that a phone doesn't need active cooling.
A single phone, far away from other sources of heat, used only in environments that are comfortable for humans (which may well involve active cooling at the building level), and configured to throttle down when it's used for more than a few minutes, perhaps not (although even then, I've seen phones get uncomfortably hot when gaming, and had my own phone shut down because it's too hot on occasion). If you were setting up banks of phones to run in a datacenter environment, and expecting to run them flat out, you'd probably want to actively cool them.
> And I don't know if that statement is accurate, since passive cooling would be even better than active cooling from an efficiency perspective and particularly if you also consider reliability. Less moving parts would mean higher reliability unless, eg., the thermal paste evaporated on the cpu.
Everything is tradeoffs. Per recent posts here, the likes of AWS are now at the point of cooling CPUs directly with datacenter-scale watercooling; essentially the building's air conditioners, rather than terminating at a fan unit on the inside, feed cooling water directly to a plate mounted onto the CPU heatsink.
A single cloud server could host hundreds of thousands of personal static websites though. I suspect they probably do use less power than 100,000 old phones.
This hits so close to home. I once tried to explain to a manager that a server at 60% utilization had zero room left, and they looked at me like I had two heads. I wish I had this article back then!
Up to a hair over 60% utilization the queuing delays on any work queue remain essentially negligible. At 70 they become noticeable, and at 80% they've doubled. And then it just turns into a shitshow from there on.
The rule of thumb is 60% is zero, and 80% is the inflection point where delays go exponential.
The biggest cluster I ran, we hit about 65% CPU at our target P95 time, which is pretty much right on the theoretical mark.
A big part of this is that CPU utilization metrics are frequently averaged over a long period of time (like a minute), but if your SLO is 100 ms, what you care about is whether there's any ~100 ms period where CPU utilization is at 100%. Measuring p99 (or even p100) CPU utilization can make this a lot more visible.
The vertical for this company was one where the daily traffic was oddly regular. That the two lines matched expectations likely has to do with the smoothness of the load.
The biggest problem was not variance in request rate it was variance in request cost, which is usually where queuing kicks in, unless you're being dumb about things. I think for a lot of apps p98 is probably a better metric to chase, p99 and p100 are useful for understanding your application better, but I'm not sure you want your bosses to fixate on them.
But our contracts were for p95, which was fortunate given the workload, or at least whoever made the contracts got good advice from the engineering team.
If your SLO is 100 ms you need far more granular measurement periods than that. You should measure the p99 or p100 utilization for every 5-ms interval or so.
Lately I've been using umu-launcher [1] which is a wrapper and compatibility tool for the proton used by steam and I'm quite satisfied about it. It's shipped with configurations for 3000+ games, might be worth a try if you're having trouble with the configs.
Take it with a grain of salt, I don't have much experience in running windows application on Linux, I just had some problems with wine and I discovered umu to be pretty straightforward and easy to use.
Same here. I used to have great success with PlayOnLinux, then it seems worked on it ceased 3-4 years ago? It was very duplicative, making a fresh Wine environment for each application, but for that cost it seemed to be less prone to issues than other approaches in the 2010-2020 era.
This doesn't seem to be using Wine though, as it requires Docker it's likely something lower level than a compatibility layer that translates syscalls.
Don't underestimate how nice and legitimately useful it is to organize real physical objects in real physical space as opposed to dragging icons around on a computer screen. Not just for some vague feel-good or nostalgia reasons but the user experience is really just significantly better for some 10s or low-100s of objects.
I'm dying laughing. Imagine Lewis and Clark, thinking they're forging this epic, heroic legacy for the ages... and their most permanent, scientifically-proven trail is literally toxic poop. History has a WILD sense of humor.
I do wonder if they went in there thinking they're "forging this epic, heroic legacy for the ages", or if that was hindsight / US historians drumming up the story for the history books.
If they shat themselves to death halfway, would they have been remembered as much?
This is horrifying, but I feel like we're focusing on the wrong thing. The AI wasn't the cause; it was a horrifying amplifier. The real tragedy here is that a man was so isolated he turned to a chatbot for validation in the first place.
I don't think "so isolated he turned to a chatbot for validation" describes this, or why people get unhealthily attached to chatbots.
1) The man became severely mentally ill in middle age, and he lived with his mother because he couldn't take care of himself. Describing him as merely "isolated" makes me wonder if you read the article: meeting new friends was not going to help him very much because he was not capable of maintaining those friendships.
2) Saying people turn to chatbots because of isolation is like saying they turn to drugs because of depression. In many cases that's how it started. But people get addicted to chatbots because they are to social interaction what narcotics are to happiness: in the short term you get all of the pleasure without doing any of the work. Human friends insist on give-and-take, chatbots are all give-give-give.
This man didn't talk to chatbots because he was lonely. He did so because he was totally disconnected from reality, and actual human beings don't indulge delusions with endless patience and encouragement the way ChatGPT does. His case is extreme but "people tell me I'm stupid or crazy, ChatGPT says I'm right" is becoming a common theme on social media. It is precisely why LLMs are so addictive and so dangerous.
Most of us are so isolated that we will turn to validation from bubble brothers sharing our view which is a real tragedy, yes. It may be horrifying but a horrifying normal at that.
In a technical sense, no technology is ever the cause of anything: at the end of the day, humans are the cause. However, technology often unlocks scale, and at some point quantity becomes quality, and I believe that is usually implied when it is said that technology “causes” something.
For example, cryptocurrency and tumblers are not themselves the cause of scams. Scams are a result of a malevolent side of human nature; a result of mental health issues, insecurity and hatred, oppression, etc., whereas cryptocurrencies, as many people are keen to point out, are just like cash, only digital. However, one of the core qualities of cash is that it is unwieldy and very difficult to move in big amounts. Cash would not allow criminals to casually steal a billion USD in one go, or ransomware a dozen of hospitals, causing deaths, subsequently washing the proceeds and maintaining plausible deniability throughout. Removing a constraint on cash makes it a new thing qualitatively. Is there a benefit from it? Sure. However, can we say it caused (see above) a wave of crime? I think so.
Similarly, if there has been a widespread problem of mental health issues for a while, but now people are enabled to “address” these issues by themselves—at humongous scale, worldwide—of course it will be possible to say LLMs would not be the cause of whatever mayhem ensues; but wouldn’t they?
Consider that it used to be that physical constraints made any individual worldview necessarily be tempered and averaged out by surrounding society. If someone had a weird obsession with murdering innocent people, they would not be able to find like-minded people to encourage them very easily (unless they happen to be in a localized cult) to sustain this obsession and transform it.
Then, at some point, the Internet and social media made it easy, for someone who might have otherwise been a pariah or forced to adjust, to find like-minded people (or just people who want to see the world burn) right in their bedrooms and basements, for better and for worse.
Now, a new variety of essentially fancy non-deterministic autocomplete, equipped with enough context to finely tailor its output to each individual, enables us to fool ourselves into thinking that we are speaking to a human-like consciousness—meaning that to fuel one’s weird obsession, no matter how left field, one does not have to find a real human at all.
Humans are social creatures, we model ourselves and become self-aware through other people. As chatbots are becoming normalized and humans want to talk to each other less, we (not individually, but at societal scale) are increasingly at the mercy of how an LLMs happen to (mal)function. In theory, they could heal society at scale as well, but even if we imagine there were no technical limitations preventing that, sadly practice is more likely to show selfish interests prevail and be amplified.
This is basically the 1910s version of 'move fast and break things.' Tossing out a one-month-old machine for a better one is such a Silicon Valley mindset.
The deeper meditation is understanding the politics of Henry Ford, how he too bought a newspaper business, and the worker conditions and economies. See:
Powerful story. But let's be real: after the "survivor's euphoria" fades, how do you actually keep that level of consciousness? I feel like the daily grind would inevitably pull me back to my old self. Has anyone here had a life-changing moment and actually managed to stay changed?
Its a constant practice, like anything. Part of you is changed forever when you go through something like this, the awareness part mostley, you can never go back to normality even when something like that wears off.
I know from experience because I survived a brain hemorrhage. I had a state where I experienced the world differently for many years. I still do. Something cracked open in me and it has stayed that way, other aspects of my physiology are returning to a baseline state, like my nervous system changes which damped my fear responses.
You typically don’t go back to the daily grind, as this kind of event often substantially changes your priorities.
I speak for myself, although I know I am not alone in my trajectory. About a decade ago I was ill enough for long enough with an uncertain enough prognosis that I was getting my affairs in order. At the same time a close friend died of an agressivo cancer, aged 32.
I decided to choose quality over quantity. Fuck my business, fuck my career, fuck stupid status games and absolutely fuck climbing the infinite pile of skulls.
Sold up. Put everything on 00 and gave the wheel a spin.
It’s been almost a decade. I still live in the woods, start my days with a coffee and birdsong and “ein heiliges ‘ja!’”, still have zero temptation to return to my life before.
Surviving something like that (a much less serious adenoma near the brain) is neat because I can mentally use the memory to alter my current state of mind.
I can't recreate the exact feeling, obviously. Just remembering waking up in the world of the living is still powerful enough to improve my mood and put problems into perspective, even after more than a decade. The old me has a new mental tool, forever.
At the same time, I'm not walking around like an enlightened monk either. Whether something counts as life-changing must depend on perspective and personality.
Most enlightened monks aren't walking around like enlightened monks, either! Enlightenment/awakening isn't really described as a "life changing" experience, though a sudden experience of the earliest hints of 'stream entry' can sometimes feel like one. Quite on the contrary, it seems to be connoted as a kind of very practical wisdom.
In Happiness Hypothesis Jonathan Haidt talks about how we all have a basic setting for these kind of things.
We can move the bar around but it always tends back toward that default.
He uses the example of this being why people who read self help book always seem to be reading a new self help book.
That little euphoric moment of clarity and fresh outlook only last a few months or so until you’re back at your regular old self and need a new epiphany.
> Has anyone here had a life-changing moment and actually managed to stay changed?
A life-changing moment changes one's life by definition. Each time a person experiences one, they are changed in a way where who they were before they can remember, perhaps even look fondly upon, but know they are not that person anymore.
> Powerful story. But let's be real: after the "survivor's euphoria" fades, how do you actually keep that level of consciousness?
By living in the moment and remembering how you got there.
I think one of the big bummers about this kind of thing is that it’s not really something that could be planned or chosen for. We tend to change slowly and subconsciously through the things we prioritize and routinely practice, our brains and bodies adapt to our “normal”.
The times we tend to adopt changes quickly and consciously are most often with circumstance and external pressures, and the shortcomings implicit with such rapid adaptations can manifest as neuroses/complexes. In traumatic scenarios this might be something like PTSD, but it isn’t necessarily all downsides, either. People taking therapeutic amounts of MDMA or psilocybin (as in, occasionally, not “micro dosing” or whatever Elon Musk seems to be doing) might experience a durable improvement in subjective happiness and optimism.
Disclaimer: this is my own intuitive and wholly unqualified understanding of this, which was arrived at via discussions with behavioral therapists, but I’m an IT consultant, wtf do I really know about it?
I will say that I’ve found mindful meditation highly effective for treating mild to moderate PTSD. It isn’t fast to get started, but after a few weeks of training, you can deploy your own chemical Xanax directly within your own brain using breathing patterns. It really worked for me. I used the app “Headspace” to start out.
My last attempt at using an AI assistant ended with it trying to book me a hotel in Sydney, Australia instead of Sydney, Nova Scotia. I have trust issues now, lol.
There is stuff in the MCP spec that will allow the MCP server to send a prompt back to the user if there is information or clarification needed. We will get there eventually.
Looking forward to that day. None of the major MCP clients have implemented that part of the spec yet unfortunately. But as you say, we'll get there...
It cracks me up to no end how the dev tools are much better MCP clients than the web chatbots. Claude Code is so _so_ much better at MCP than Claude Web, which has issues with managing DCR client state, is comparatively terrible at surfacing debug information up, doesn't let regular users see under the hood at how tools are described or called, etc.
Using Claude Code or your IDE of choice to book a hotel is a fun unintended side effect of this.
It is walking away from her user base. (I’m not complaining — I’m an open source dev too and recognize I have no right to put demands on her time. But what is the future for Asahi Linux after this? I don’t see one.)
