The device is probably 256 of those, or similar, chipsets on a board with integrated SIM slots and connectors for SMA antennas. You'd have a hard time producing something like this cheaper than China.

There's only 64 antennas but 256 SIMs, so there are either 64 separate quad-SIM modems (AFAIK only Mediatek made those, and in 2G only) or it's just one big FPGA, managed by a small CPU and connected to a bunch of RF ICs, like the inverse of a base station.

Or there are 64 modems that can switch between 4 SIM slots.

Better in what way? MobileX may have some properties that are attractive to criminals.

Mobile-X is cheap and is "pay for what you use"

I don't think there's some other seedy reason - Mobile-X is just the least expensive option I know of right now in the US that can be purchased at retail, so that is probably the main reason

More data etc, lower price.

I checked less than 90 days ago, and Mobile-X was cheaper than any other MVNO with retail availability in the US.

Do you have links to services that cost less, per line/sim? I don't think they exist, especially at retail.

I might then be completely wrong.

Common carriers are usually absolved of the malfeasance of their customers.

Probably a lot of places to buy MVNO sim cards from with cash as well.

Each one of those units is probably ~$6k for the device and sim cards. I don't think there were that many of them in the pictures to add up to $900k.

The article describes 300 servers and 100,000 SIMs across a handul of locations.

In some countries you can find entire office blocks filled with people who do nothing all day but participate in scam enterprises. I don't think the scale of this phone bank, if its as described, is that surprising really.

They made cold calling illegal in my country. Also you cannot just sell customer data. It made an entire industry disappear and nobody mourned.

But I'm sure some American lawyer would call that a breach of the constitution.

How does your country protect against callers and data sales outside of its jurisdiction?

Sounds pretty click-baity but if a company put the name of my software on an opaque list that ships with the OS that makes my software behave differently than I expect it to and didn't even TELL ME about it I'd be way WAY more pissed than this guy is.

Drivers do this to every single videogame. And it is good, otherwise those game would not run in every consumer 3d card.

I'm sure the driver devs have at least some interaction with the game devs when they do this.

Man I gotta give this person props, that is a LOT of effort to get your game to run native ARM64 on Windows. How many people out there are trying to game on ARM64 Windows devices?

We get a lot of people wanting to run the game on ChromeOS school laptops, which is also ARM-based.

I don't think that would contribute to much demand to run it on Windows ARM64.

I played Silksong recently on my Surface Pro 11, it worked fine. 2D game with minimal graphics requirements so it's a different case, though.

Space Station 14 is the same though but it might use some 3d stuff for rendering based on the comments in the blog about issues with OpenGL (or might just use OpenGL's 2D stuff) I know it has some 3d-ish things for visibility shadows but those might just be done in 2D, and from what I've tried SS14 (and SS13) is a 2D game.

OpenGL doesn't really have "2D stuff" (at least not in any version from the last 20 years). It's all a programmable 3D pipeline, which you can do 2D in by just not using the Z coordinate in the shader. However, doing 2D graphics like this tends to be simple enough code that you're less likely to trigger any driver or performance issues.

I like where their head is at here, especially the "superset of JSON" part. Some of the things I'm not _in love_ with like the %% ending blocks or how maybe a bit of significant whitespace might make things a bit less misleading with indentation as others have said, but overall I think I like this better than YAML.

US bought a big stake in Intel; the US is quickly attempting to replicate China's system of State Capitalism. If you can't beat em, join em, I guess.

Having done some work with building graph representation of compute resources (by importing tfstate & syncing from aws) and its a very useful representation for building things like visualizations and detecting dependencies. Moving tfstate to a real distributed graph system would help solve a LOT of the nasty hacks they mention in this article that I have run into HARD when working with very large numbers of resources and multiple teams/team sizes.