frezik

Does Windows booting take up that much space because of code, or because of data that code is loading?

Even Nvidia's gains are based on selling that hardware to outlets buying GPUs in shipping container quantities to run huge LLMs. If those companies dry up, so will Nvidia's market position. At the other side of things, their traditional market is gaming, but that industry is contracting and may not be making games that push hardware limits the way it used to. Their best long term bet might be a rumored Steam Deck/Switch like handheld (one of their own making; they do supply the chips for the Switch and upcoming Switch 2), but that's not going to justify a $3T market cap.

I have. Still small compared to the images and such that are used in a user facing application.

Edit: just to bring in real numbers, I have an old TypeScript project that results in a 109M node_modules dir. Which I agree is absurd. I also have an old anime video, 21 minutes long, at only 560x432 resolution, 24fps, which takes 171M. And that's my point: even in really bad cases, code size tends to be swamped out by everything else in user-facing applications. If there's any kind of images, music, or video, the code size will be a small part of the complete picture.

If you're old enough, then the first word processing program you ever used was probably on a screen 640x480 pixels or smaller, didn't support internationalization, couldn't provide true WYSIWYG to match output between the screen and a printer, and couldn't render fonts with anti-aliasing. Which of these features would you like to drop to reduce the size?

Everyone loves "tight" programs until they realize what they have to give up to make it work.

It's almost always compressed in some way. Still takes up a lot of space. You can fit a lot of compiled code in the space of a 1 minute, 128kbps mp3.

And there's almost always a reason. Code size tends to be modest compared to supporting data around it.

Are you willing to give up 1080p screens and 16-bit/44.1kHz sampled music? Or how about languages that can't be represented in ASCII, much less Latin-1? Because handling those take up way more space than code.

Code is rarely the biggest thing in these programs. You want textures that don't look stretched and pixelated at 4K? That's going to cost you.

Look in any game directory. There's probably a one big file--sometimes a few big ones--in there that you can rename to .zip and unpack it as one. It will dump all the textures/sound effects/etc. in the game, but have zero code. It will be something like 70-90% of the game's entire space.

That's what the industry wants to believe. Except that US regulators have shown a willingness to sign off on new nuclear power plants as long as you do all the paperwork right and show that you're not some moron who will dump a pile of plutonium in the desert and run water over it to make steam.

Nuclear takes 5 years to build according to initial plans. That's a joke, and everyone knows it. It's going to take 10 years, and the budget will double over initial estimate, as well. That means it will take 10 years before you see a dime back on your investment, and it could all be for nothing if the funding shortfall can't be made up. Some of this is regulations--you know, the kind that keeps another Chernobyl from happening--but a lot of it has been the fact that every plant takes boutique engineering and specialized labor.

The Westinghouse AP1000 design (what they used in Vogtle) was supposed to fix that boutique engineering. It did not. SMRs are also supposed to fix that boutique engineering, but their projects are also failing.

Meanwhile, you could invest your money into a solar or wind farm. It'll start generating power in 6-12 months and start putting money back in your pocket. Nothing about the construction is particularly boutique; it's almost all mass produced stuff. You don't need specialists to put them together, either. There is a track record of solar and wind farms meeting construction deadlines and budget forecasts. Given all that, who the hell would invest money into nuclear?

Scrapping the NuScale project had nothing to do with lawsuits. Governments pulled their financial support because projected costs were exceeding what was contractually promised, mostly due to pandemic-related supply chain and inflation issues.

This is typical of nuclear. The industry wants to believe its problem is regulation. It's not, at least not if you want to have better safety guarantees than the Soviet Union did. Its problem is that to be safe, nuclear is expensive, and there doesn't appear to be a way out of that.

When these artificial caps are removed and the plant is allowed to operate as intended and no kneecapped to allow coal and oil plants to operate at their peak effeciency rates, nuclear drops below .10USD.

Wholesale or retail cost? Either way, that's not especially cheap compared to renewables.

Nuclear may be cost competitive with putting solar panels in space at this point. Granted, that's back of the envelope costs for a hypothetical space based solar system compared to nuclear plants that already exist. But the fact that they're close is not a good sign for nuclear.

Plants will take 10 years to build, at least. If every permit was signed today, there wouldn't be a single GW of this new nuclear going on the grid until 2034. We're aiming for major reduction in CO2 by 2030. Oh, and the huge amount of concrete needed would create a massive spike in CO2 by itself. Timeline issues alone kill nuclear before it starts.

Edit: fixing autocorrect's corrections

It's limited in the geography where it could be useful, such as near techtonic plate boundaries. Iceland gets about a quarter of its electricity that way. Some advancements in drilling techniques have made it more viable in more locations.

https://www.jonesday.com/en/insights/2023/05/is-geothermal-energy-making-a-comeback