HN user

xt00

2,259 karma
Posts0
Comments530
View on HN
No posts found.

The amount of data they must have at this point running so many of those raptor engines has got to be insane... at least 300+ engine launches now -- wow.

Totally agree — if AI tools are already or nearly at the point where you can say “write a program to do X” and it does it, that’s like telling people they need to learn skills to order something at McDonald’s. The goal is for the barrier to entry to be basically zero. Oh sure today there are things like “I made a claude.md file that does this and I wrote a really clever prompt!!” But the goal is for that work to be deleted as well — where is the magic skill that is / will be needed?

Tons of CEOs right now keep saying “young people need to learn how to use AI to be successful” and also “we aren’t planning to hire any new college grads due to AI”.. so which one is it.. seems everybody understands the super pro AI CEOs want to lay off nearly the entire company and run it on skeleton crew with a ton of AI and get ultra rich. While “some other” companies should totally hire lots of young people but not them.. where does that end?

And if you want to divide up the world along a couple of axes into job usefulness and how much the worker cares, then the author of the article was in the “useless job and cared”, while you are in the quadrant of caring and have a useful job best I can tell — which is great. Some people have useless jobs and don’t care much about their job but are happy to get paid. If you for example are a teacher of small children who depend upon you and you basically don’t care then that’s problematic while having a BS job that you care about a lot about is probably mostly an organizational / bad management situation.

Do we know who is funding this? is this one of these things where Meta doesn't want the responsibility for this, so they are pushing to have the OS have the responsibility or something like that?

How long before there is a "we've detected your account has been used multiple times to re-setup a phone.. we've re-enabled the Google Nanny Safety mode.. also we've locked your google account just in case.. " I mean other than hackers, who has needed to factory reset their phone more than once in a year you must be doing something shady... right right?

When you say "this is NOT from Hastings" I had to double check my post again -- I guess you are saying that the Pade approximation is not from Hastings, but the polynomial approximation that the OP referenced from nvidia from A&S and ultimately from Hastings, definitely is in Hastings on page 159 -- I think you were referring to the Pade approximation not being in Hastings, which appears to be true yes. In the article it is interesting that the OP tried taylor expansion and pade approximation, but not the fairly standard "welp lets just fit a Nth order polynomial to the arcsin" which is what Hastings did back in the day.

To be accurate, this is originally from Hastings 1955, Princeton "APPROXIMATIONS FOR DIGITAL COMPUTERS BY CECIL HASTINGS", page 159-163, there are actually multiple versions of the approximation with different constants used. So the original work was done with the goal of being performant for computers of the 1950's. Then the famous Abramowitz and Stegun guys put that in formula 4.4.45 with permission, then the nvidia CG library wrote some code that was based upon the formula, likely with some optimizations.

I imagine there is a process in place to allow cities / states / communities to place the cameras on polls. If Vegas somehow got around public comment process to put these on poles, then what would stop any random company from requesting to put their own camera there? Like lets say a motivated individual went through some process to put a camera on a pole someplace near somebody that would definitely make the govt official / flock exec etc nervous, what is stopping them? It sounds an awful lot like Flock is basically going to town's and saying "we will put up a bunch of cameras in a bunch of places" probably based upon algorithm's etc. How do they decide where these get put, who gets to decide that? Why can't any random company request to put up a camera on a random power pole? After they give the map to the govt officials, do they get a chance to say "oh this one by my house, can you move that?"

So using Stefan-Boltzmann equation if you have a 1m^2 surface at 100C you can radiate about 1kW from that surface -- assuming both sides radiate that, then lets assume it is double. Assume each blackwell chip + support electronics etc needs about 2kW of power to run. So each 1sq meter of say a copper plate is needed to cool 1 blackwell chip. So if you have some way to make some massive radiators that are basically giant plates spanning thousands of square meters, then you should be good. the Stefan-Boltzmann equation is proportional to the 4th power of T (in kelvin), so if you can somehow manage to use a heat pump for the heat from the GPU's into your heat sink such that you could run your radiators at a much hotter temperature, then the blackbody radiation that they put out dramatically goes up. So cooling is quite challenging but not impossible. (I also neglected importantly that you would need to use the giant solar panels as a sun shade for these radiators otherwise they would be pulling in heat from the sun)

For power, you need to somehow manage to generate all of the power that you would need to cool. So the most logical would be some huge solar panels -- assuming you could use similar tech to the space station, you can get aroudnd 100kW from those solar panels -- assume you can do say 10X better somehow, then now you have 1MW of power.

Unclear what the goal here is -- if the idea was doing this for cost, it sounds super unlikely to pan out -- if they want to put a datacenter in space such that nobody can tell somebody what to do, it would seem just as easy to go hide a datacenter in some random far flung corner of the world in a bunker. Seems just like a great way to light some money on fire.

Yea in an ideal world there would be a legal construct around AI agents in the cloud doing something on your behalf that could not be blocked by various stakeholders deciding they don't like the thing you are doing even if totally legal. Things that would be considered fair use, or maybe annoying to certain companies should not be easy for companies to just wholesale block by leveraging business relationships. Barring that, then yea, a local AI setup is the way to go.

Inference / Agentic AI implies "running models performantly using CPU cores" most likely (maybe with some optimizations / special AVX512 stuff) -- so essentially "welp, no sense in trying to build GPU's, we are too far behind nvidia to catch up".

Yea it is tricky for them -- the old model of "search, see google text / link ad, scroll, click website, scroll, see some ads on that page as well, done" will be replaced with "search, see google text / link ad, read AI result, 'and here are some relevant websites'" -- where all of the incentives there will be to "go into more depth" on the websites that are linked there.

The classic "we are taking something away in order to help you -- don't worry!"

In reality, nobody who works on bringing android to other hardware is somehow impeded by having extra device tree info in the source tree.

The most logical thing happening here is like Google is saying "hey man its hard to get drivers working, and we don't want a bunch of freeloaders leveraging our work!!"

The fact that Sam says he would open his laptop to ask Chatgpt something is almost like he is trying to avoid saying he would use his phone.

The reality is that we likely don't need any new devices as much as people want to keep saying that. If you have airpods and a phone you could talk to chatgpt and say "show me how to fix my kid's bike with a simple video on my phone" -- it buzzes your phone and boom the video is there. Sure it is missing the ability to take a picture of the world / video -- so in that case, a pair of the meta rayban glasses would do that -- again just use your phone / cloud, it all works. Or skip the special glasses and hold your phone up to the thing you want to take a picture of. No need for magical new devices.

Having a camera staring at me while I talk to somebody -- yea I'm gonna pass on that.

"will find, analyze, and synthesize hundreds of online sources"

Synthesize? Seems like the wrong word -- I think they would want to say something like, "analyze, and synthesize useful outputs from hundreds of online sources"..

His attention span won't be enough to stick with physics for longer than 4 weeks.. he should just apply his skills in building companies to hire a couple of people to work on cool stuff that he is excited about. A good approach is to find some smart grad students at a local university who are in electrical engineering probably since you have to know lots of physics to get there and say "hey i have some money, want to build some cool prototypes?" and the best part is you can just ask those grad students how all this physics stuff works and they will happily tell you whatever you want to know.

Was one of the goals with EdTech to sort of gamify learning to get kids more engaged or something? I could imagine a scenario where some kids learn more by a paper / pencil / book modality and would do worse learning from an iPad— while other kids who are totally disengaged might learn something at all from an iPad so that’s better than nothing?

Great work by the authors -- great to see this type of effort detailed. Video codecs are complicated so great to see that somebody did this. It would be cool if there ends up being some presentation about this saying something like "here are the bad patterns you should avoid if you want to be able to port your C code to Rust at some point" (things that may be outside of what you actually had to do on this project as well as what you mentioned in the post). Also I'm glad you mentioned how long this took -- nice to get a scale of how hard these ports are.

Yea it definitely felt like at the time Intel was in a tough spot.. while the article is stating that the narrative about the margins being narrow for arm chips so Intel was nervous about the area, one thing I can attest to is that even if that narrative was false when it comes to Apple, competitors of Apple at the time definitely would be assuming that if Intel went into that space that they gave Apple a good price, so just the perception of them selling a good chip to Apple could have hurt Intel's fat profit margins. Like Apple was selling phones for the same price Intel was selling chips for -- so maybe the higher cost could be justified by Apple who knows, but the perception of Intel's profit margin they are willing to live with impacts their negotiations with other customers dramatically as well.

If the companies that are holding land are able to maintain steady profits with minimal competition, then the logical approach is to say we should incentivize them to build on the land, make improvements etc now to ideally decrease the value of the existing land to make it more affordable to people. So could you essentially say "if you own more than 20 properties that are not owner occupied, you have to pay a tax.. but if you sell the house, you get some large chunk of the tax money back".. something that should encourage the big builders to build something and to sell it -- not to build and potentially rent it out -- which is in many cases the most expensive type of house to rent.

Gradle is a very strange build system.. it actively downloads things if you are missing things, but then still will only work if you have the right combination of things on your system (on linux at least), and fails in cryptic ways. Not sure I know anybody actively likes Gradle and that entire build system -- there was something previously like ant something, etc.. if you have an older project good luck having it actually build even when trying to use Android Studio.. it seems to only work if you actively use Android Studio quite often and slowly adapt to each random change to keep your project working properly.

This is a typical "move fast and break things" mentality.. except that mentality betrays Sam's statements about doing a bunch of this stuff "carefully" etc.. its all a smokescreen.. nobody is going to realistically stop working on AGI in order to be careful.. basically AGI is being pursued like the race to get the atom bomb.. so yea history tells us its full speed ahead with no brakes. Scarjo is just the latest person getting stomped on along the way.. eventually it will be a whole ton of people getting stomped on.. whoops!

I think we all can understand the situation here unless people are really dense.. the Artemis program was setup at a time when the private space companies were still very new. SpaceX will soon be quite close to technically doing the entire mission themselves without Artemis at all. SpaceX took the money from NASA to help fund their Starship development and probably for other reasons as well. Net result is that by the time Starship can land on the Moon, they can basically do the entire mission without Artemis. So Artemis would be pointless.

"What’s weirdest about Apple’s antitrust and PR problems related to the App Store is that the App Store is a side hustle for Apple. " -- that seems to be a feature not a bug -- Apple incorporating more and more capability into the base iPhone OS / software stack is what keeps people coming back. There was a mentality years ago that the iPhone was the good hardware plus a bunch of apps -- but over time many of the most important apps have been built in apps / services from Apple directly. So in many ways, if Apple allowed the app store to be more open, then it would allow 3rd parties to actually build a bunch of value there without much control from Apple -- so it would make more money for Apple at the expense of creating a more vibrant set of 3rd party apps that almost assuredly work on both the iPhone and Android phones. So in many ways, Apple probably would be happy with close to zero revenue in their App store as long as people were 100% locked into buying new iPhones and airpods and ipads next year.