Sauron

The hardware is kind of beside the point since we're comparing software processes on the same device. Obviously more efficient hardware will draw comparatively less battery power, but it will still be proportionally constrained by the capacity of the battery itself. The question here is whether running an LLM like this won't draw significantly more power (on the same device of course) compared to a more constrained, but arguably more than adequate, assistant chatbot. And my impression is that no amount of "engineering" will significantly change that ratio, because there just isn't much you can do in terms of software optimization to reduce that workload. Maybe with specialized hardware acceleration the difference won't be that large, but it remains to be seen and it's certainly not a given as you seem to imply. Here, have a graph of battery life by generation: https://www.statista.com/statistics/1308110/iphone-battery-life-comparison-by-model/ you'll notice that while overall the trend has been increasing (mainly due to the batteries just getting larger, once apple got over the obsession with hyperthin phones and tiny screens) it varies year by year, with the 12 for example having a significantly shorter battery life than the 11, and notably abysmal performances from SE models and some top end ones even long after the example I gave - in fact, the 8, X and Xs had even worse battery lives than the 6 and 6s, which spawned the camel back battery cover.

doesn't work for interactive tasks afaik that's not really a thing you can do... if there's retraining involved it can be done asynchronously but the meat of the operation would still be the inference of a response, which you can't get from just cutting down the model (if we knew which specific parts of the network are needed to answer specific types of question this would be a whole lot easier...). To be fair I'm not specialized in neural networks so maybe it's possible and I just don't know about it.

iOS was not the problem, the insufficient battery size was. Apple hardware and software is not magic, if the battery is too small you'll get bad battery life and, if you overuse the hardware by running heavy LLMs, that battery life will be shortened further.

patents run out eventually

there was a patent a while back about a tv that would require you to say the company name out loud to skip advertisements. that never materialized.

Uhhh... not really. Battery life has been largely hit-or-miss on iDevices, as exemplified by this utter monstrosity:

This is likely still not desirable since you'd still be working with a limited battery life. It would certainly save a lot of money for the service provider though.

It's probably fine, but you should probably have complained upon receiving the delivery in this state. Just put it in a pc?

Just because they filed a patent doesn't mean they plan on doing it. A lot of patents are filed just to prevent competitors from using the idea or "just in case" they want to use it in the future.

...which leads into the point that there's generally a reason these things are they way they are. As you said, off-the-shelf components are intended to work with a variety of other parts and manufacturers would (understandably) rather stick to safe defaults than deal with a flood of "incompetent" customers complaining that their device didn't work out of the box, despite being advertised as compatible with the rest of their setup. I would wager most of these users also prefer this arrangement. It's a bit like cars - most people use their car to commute and don't care that they could theoretically get 10% more hp by tweaking the fuel injector settings, with a slight chance of being left with a non functioning car in the process...

Not necessarily, there are people whose job it is to do this sort of research as well as bug bounties by Apple and others to incentivize them. Not that I think apple silicon macs are unpopular, just that this isn't really an indication of either. Not to mention the ability to expand your memory.

Let alone prove the owner of said ID card is actually the one signing up for the service.

I disagree, packages go through validation and testing before they're thrown into the repos and this didn't just randomly occur because of a single malicious commit by a random contributor. Clearly there was long standing malicious intent behind this and collaboration from the upstream developers; hypothetically this could have been done with a "bugfix" of an older release, which could have made its way into any distribution. Moreover Arch was not affected, so it's possible that if the backdoor had been active in the Arch build it would have been caught by the Arch maintainers - we just can't know for sure. Debian Sid and Fedora Rawhide are considered testing distributions so they don't have the same safeguards as mainline Arch - if you want that experience on Arch you need to look at the testing repo.

Microsoft has a lot of people working on various Linux related projects (and Linux itself) so this isn't that weird.

Yes I can, who says it needs to be law? Does the law prohibit you from showing age inappropriate movies to kids? There are warnings and labels, the rest falls under parenting. If you don't know your kids are on tiktok or whatever, you're a shit parent... but that's it. Except kids could just ask their parents, or their friends, for their code. If it's not tied to your online identity it just wouldn't work to do what this is trying to achieve. Just think of systems like PGP... you can just give away your key if you want to. What we have now is simply not giving your 10 yo a smartphone if you don't think they can use it responsibly. Much better success rate than anything involving ID. Either way the system you propose is not the one being implemented, so that's neither here nor there. Specifically for their children, no... but having it around the house where the kids can reach it, or having their older siblings sneak it in? All the time. In fact, forget alcohol, this is a known phenomenon with guns. I wonder what mr. DeSantis would say if someone proposed restrictions on gun ownership for families with children... Not to mention a lot of people don't, and still wouldn't if it were made illegal, see social media as comparably dangerous to vodka (for kids).

I can share my age, but maybe I'd rather not share other identifying information like, you know, my name and address. Like on this forum, for example. Just because you don't care about privacy (or maybe you do? I doubt BrandonTech is your full legal name...) doesn't mean others don't. Even if I AM willing to use my real name on a social media site I may not want to have to prove it's actually me. In some places (namely florida...) having certain statements tied back to you with absolute certainty may even have you retroactively prosecuted by the state, for things you did that weren't crimes at the time or place you did them. Websites don't ignore it, they put the onus on you not to lie. If that's not enough for you, maybe you should put in some effort and actually parent your spawn... and if you are willing to accept them having a social media presence, why should the state of florida come in and stop them? Kids don't want to be anonymous on social media anyway, which I'd argue is part of the problem with modern online interactions, certainly not the solution. Anonymity protects you, especially as a child. It means you can just log off and stop engaging if you get harassed or bullied, without lasting real life consequences. It means you can't be as easily targeted by predators. The result of laws like this (and we have a real life example in the UK trying to enforce personal identification to access porn sites) is that 18 year olds will just create and sell accounts to minors. It doesn't work and it's a detriment to people's privacy and security. This is demonstrably not true. https://theconversation.com/online-anonymity-study-found-stable-pseudonyms-created-a-more-civil-environment-than-real-user-names-171374 Not that you couldn't just see that by the sheer number of buffoons spewing the most inane bullshit and dogpiling anything that moves, under their full name.

Most if not all social media websites already have age limits around 13, so this law's only effect is adding the invasive age verification - meaning it's only bad.

I don't especially care, ultraspecific emojis like this are pretty useless anyway. I never really got why skin color specifically was considered a problem with emojis though - surely bright yellow isn't anyone's skin color? Maybe we could have made them green or blue just to drive the point home? Gender I can see I guess (although the original emoji design, i.e. , isn't really gendered in any way)... but differentiating them further seems like the wrong solution. Heck I even question the idea of differentiating the number of children, like what if I have 3? 4? 5?... and so on... Them being ugly can be fixed, I'd rather have just a handful of generic emojis than have to sift through 20 of them just to find the one that matches the exact family I'm referring to best... leaving aside the problem with children being limited to 1 or 2, or if we really wanted to be pedantic the existence of extended families where the parents might be more than two. < removed by moderation >

No, you can't. I assure you polling ACPI information even once per second is not going to significantly stress a modern cpu. You can check this yourself with htop or any other program that reports live cpu usage. something like this? https://wiki.archlinux.org/title/Acpid you'll have to check yourself whether your laptop generates an event for the percentage changing... I wouldn't count on it, considering 30s or even 1 minute is a perfectly adequate refresh rate for battery level... I mean, if you're worried about your percentage changing much faster than that then you should replace your battery.

If you read the man page for cpufreq-info you'll see that it too can read straight from hardware with the -w flag. As far as I can tell, the difference is that "software" polling will read the /proc/cpuinfo file (populated by the kernel) while "hardware" polling will get its reading from different sensors on the CPU (typically by reading values in architecture-specific registers); sometimes this makes absolutely no difference. On modern CPUs with boost and other power scaling functionality, the "hardware" reading is more likely to be the correct one.

Rack sized GPUs are nothing new for nvidia, the DGX line has existed since 2016...

Admittedly it would be pretty funny if it turned out that this post was AI generated...

Could have been useful as a built-in development tool if done differently, as for actually using android apps though... probably not that useful. I've used bluestacks for a couple of things but realistically I could just as easily have used a native program, it's been more useful to test some open source apps before sideloading them straight into my device.

in the roughly 2010-2018 time frame consumer cpus barely got any better, especially in terms of game performance (partially due to the console generation of the time relying on relatively slow CPUs) so it really didn't make much sense to upgrade your CPU regularly. Nowadays if you have a cpu from those times it's probably holding your system back.

As I said, experimental support. It also only works on wayland afaik. I'm sure it can be made to work if you're willing to spend the time on it but it's quite restrictive and certainly not something you can expect of someone who just wants to use their computer out of the box.