utopiah @ utopiah @lemmy.ml

Posts

1

Comments

624

Joined

3 yr. ago

Again... I didn't even read the article but "[redacted to remove bias] University researchers have developed [better] than leading [whatever]." is definitely interesting yet also pointless. Of course research is important, even fundamental, to the production process... but it's not a fair comparison because production, at scale, and economically reliable requires a LOT more constraints!

So the research, regardless of the source, is welcomed but comparing to production rather than comparing to other research labs pushing limits on the same dimensions is not useful.

PS: for my starting "Again" see my post history.

Edit : AFAICT "outperforms the most advanced commercial chips from [...] Belgium’s Interuniversity Microelectronics Centre." IMEC doesn't do commercial chips, just research.

with 9100 being 6nm

For historical context TSMC announced 7nm in 2018 https://en.wikipedia.org/wiki/7_nm_process and 5nm in 2020, so 5 years ago.

China is now making their own chips domestically that are only a generation or two behind the bleeding edge.

Maybe I'm missing something here, which chips are you talking about? Are you talking about something other than Kirin 9000S and if so which ones please?

I'd add :

0. back up your data

I let you read the comments from their source since you didn't actually bother reading mine.

Edit: people can check my Lemmy history on the topic, I ask the same thing here every few months. Anyway also the moment to suggest Chips War (even though, as always, outdated) as a good book IMHO on the geopolitics of chips manufacturing.

Feels like we have news like that every quarter but not a lot of actual change. Does any foundry outside of China, e.g. TSMC, buying or even getting any partnership to test them? Without subsidies? What's the yield relative to alternatives?

It does beg for a DeepSeek moment for hardware, namely actual competition stemmed from necessity, but again so far that race has been a lot of claims.

To clarify, I've been using Linux for decades... and I still take notes! For example https://fabien.benetou.fr/Tools/Shell or https://fabien.benetou.fr/Tools/Ffmpeg so please, pretty please, do NOT keep it all in your head! There are myriads of way to record your learning so don't be shy about it.

First and foremost, welcome to Linux!

Few pointers to hopefully help the process :

• "Not knowing where to find anything." indeed, it's disorienting but it will come. You can find actual "maps" but honestly, just as you would do in other operating system, use the search function. If it's not obvious this way, search online. The first few times it will be weird then each time it does become easier until it actually makes sense!
• "The lingering feeling of instability." have a /home directory (not "folder", that's funnily enough a Windows term as they tried to be different, going from the unanimously used / to their own C:\ things) so that you can actually go "nuts" with your installation, actually messing things up but without the fear of losing your precious data! Each new install is an occasion to learn. That being said, Linux is very VERY stable. I've been running the same installation for years, on desktop and servers alike. If something goes wrong it can usually be fixed and it's, again, an occasion to learn. That being said, having a dedicated /home directory on its own partition or even disk gives you the opportunity for a low effort low risk blank slate.
• "The capslock works differently" ... well this one is quick, you're looking for the SHIFT key if you only want to type few characters in uppercase ;)
• "Every once in a while, my desktop icons get rearranged." yet another occasion to learn. What's the bug from? Is there an issue open? Is it being worked on? By whom? How? Why? You might even be able to fix it!
• "It seems impossible to get Firefox to not restore sessions after shutting down the computer with it still open." it's in the Firefox preference : Settings -> Startup -> untick "Open previous windows and tabs", literally the first option.
• "The above all add to a bit of a general ‘stuck together with adhesive tape and love’ feeling." nice, and that's just the surface, it's now YOUR system so you can do whatever you want, even if everybody else disagree.
• "Not knowing how to install programs." well that loops back to all the learning opportunities above and the last remark, it's YOUR system so you can use whatever you prefer, both in terms of apps, settings or even how to install (or not! Check e.g. Nix) apps. There are even "weirder" things like https://github.com/ivan-hc/AM but the point is, you decide, again, always!

scrot if you want to use the CLI, (KDE) Spectacle if you need a UI.

FWIW each new install is faster, especially if you write down the "weird" steps.

On a broader and more philosophical perspective, cheating or IMHO more appropriately hacking, is in the eye of the beholder.

Is it really cheating if you respect all the rules? Aren't the rules actually poorly defined in the first place?

What matters more I'd argue is the social contract, namely is what you are doing detrimental to yourself and or others. For example I lock picked a door just months ago, and it wasn't my door, and I'm not even a certified locksmith! Well, it's because my neighbors asked me to as their key was jammed from the other side. So... at least according to them, who owns the house, it was helpful.

My overall point is that this is quite sensationalist, as most of AI "reporting" is (I put quotes around because truly it's just marketing or PR for AI corporations at this point) it actually is an expected behavior.

PS: reminds me of this streamers few months ago (sorry, no link) who was "shocked" that it's local AI exited its container to "hack" his computer. Well, lo and behold when you check his actual prompt, he does explicitly request the AI to do so.

Always has been... there is no reasoning, it's literally just spitting back the most likely answer based on previously seen answers. A 5 years old can do better.

Edit: "AI systems may develop deceptive or manipulative strategies without explicit instruction." ... right, well, guess what, the Web (which is most likely the training dataset for most LLMs) is full of "cheating" strategies. Don't be surprise if you find a "creative" answer to a problem... when it's literally part of what you train the model on.

My gosh people... this... this might be the year of Linux on the desktop!

(not even being sarcastic here, the reach streamers have is huge. I bet a lot of their audience is thinking now "Wow... if he did it, maybe I could!")

redundant

redundant AND prone to failure! What if the robot slip? What if it's out of battery. Totally non sensical!

Discussion on other lemmy instance https://lemmy.world/post/25708444

Well I (a developer) collaborated with an artist (3D modeler) recently and... I did not ask them to install anything.

Instead what I did is a develop a Web drag&drop page. They'd visit it, drag&drop their model and... see if it worked (e.g. visually or running animations) as they expected. That was it.

IMHO finding the boundaries that are important, and thus how to collaborate, is more important than a unique reproducible environment when roles are quite different.

TL;DR: IMHO no, you don't, instead find how to actually collaborate.

set up by non-programmers (such as artists) [...] requires users to learn i3wm and possibly use the command line

Yes, and I'd argue to also reflect back on (long form) content, e.g. write down notes on books, critical ones, and if you watch a movie or documentary with friends, chat about it with them, reflect on your understanding of the topic, what was good, what wasn't, develop a critical sense rather than just "consume" content.

Bonus : f*ck noise! Protect yourself when you are learning from distractions. There are myriads of things begging for your attention. Brush them off then in turn shape your environment so that you actually have a chance to learn. Learning is challenging, by definition, so you MUST make sure nothing and nobody gets in the way! Because plenty of people here have a technical side, here is a tool I built as an example https://git.benetou.fr/utopiah/online-hygiene/src/branch/master/index.js which gives me daily quota of Websites I can visit, or not, and when.

When during your life where you at peak learning rate?

Was it as school? Uni? If so, what did you do differently then? Can you still do it now?

I'll give few examples that honestly in retrospect are absolutely obvious and yet, few people seem to still do it :

• have a trusted teacher/mentor who can pinpoint your flaw
• do exercises that test your knowledge rather than read and assume you know
• repeat said exercises in with varying context and in increasing difficulty
• take notes (IMHO the biggest) that you gradually structure and index
• use said notes when exercises (which are safe spaces to challenge your understanding) gets tough
• have structured goals, namely you don't learn about a topic, move on randomly, but rather have 6 months over a topic
• learn regularly, e.g. weekly occurrence on a very specific topic, again and again for months on end
• last but not least, do it as a group, build, grow and sustain a network of helpful peers whom you are learning from but also helping

So... yeah, none of that is secret nor even complex yet most adults seems to leave THE place to learn and somehow forget EVERYTHING they actually learned. It's nuts.

Also most of that is free. Getting a notepad or a wiki or using documents in a directory on your computer is practically cheaper than a coffee in most places. There is no excuse to note take notes then organize them. Same for regularity and exercises, get a calendar then drill, again.

FWIW that works for pretty much everything, from an abstract field of knowledge, e.g. math, to a physical skill, e.g. welding or ice skating.

you probably don’t want the instance admins handling your payment info

Unfortunately centralized (but I don't think it's possible to do that with the traditional banking system due to gateways) but it's actually very easy to integrate with Stripe.com , as you to integrate yourself might be a day of work at most (source : I did few integration with WebXR as prototypes) but if you, as a self-hosted admin of the platform instance, just want to make it work, you probably only need a working account (to receive the actual money) and you'd fill up only IDs and voila, working.

Now there is also the KYC challenge, not sure how that work as a platform intermediary. Honestly at that point, as someone else suggested with Belgium 2ememain.be better letter user handle that themselves and in Europe with IBAN it's pretty trivial to pay online, no fee.

Might also be worth exploring https://developer.mozilla.org/en-US/docs/Web/API/Payment_Request_API