> Now apply that to weapons systems in conflict against an enemy that DOES have modern production that you (no longer) have... it's a recipe for disaster/enslavement/death.

How do you maintain this production with a sudden influx of ballistic missiles at the production facility - or a complete naval blockade of all food calories to your country?

Redundancy combined with defensive systems... The same as every other war in the history of mankind.

> It irritates me to no end that Gowin won't open their bitstream format because they'd displace a bunch of the low end almost immediately.

All of their IDE/programmer/etc binaries are basically entirely unprotected, almost all of their chips are entirely implemented in https://github.com/YosysHQ/apicula - if other manufacturers cared to implement it, it wouldn't be hard.

Support is stuck at the old levels--none of the GW5 series are implemented. This is just like how the Lattice support is similarly stuck at the ice40/ECP5 level which is almost a decade old.

> EDA stuff was designed by software people.

No - EDA software is built by hardware experts moonlighting as software engineers, which is partly why it is so obtuse.

I said "was", not "is". Also, I know how that's working out. Working with Vivado makes me want to barf.

Yes. The key enabling feature is a lack of malloc()

Want to know the easiest way to spot someone who has never used a bump stock or an actual full auto firearm?

They'll cite bump stocks as a threat.

The whole concept of rights is to protect activity you don't agree with.

Who are you to really say otherwise?

The taxpayer: this is a drop in the bucket in the areas that need them.

I can produce a 1911, Glock, or an AR-15 in all of these places in under a day and no law can really prevent that. The tech isn't going anywhere, it's a century old.

You are underestimating the quality and scale of American gun production. You can make your own gun, sure, or get an old one, maybe it won’t even blow your hand off. This is the one manufacturing area that the Americans have a clear advantage over the Chinese in.

I use Chinese contract CNC providers for non-firearm parts all the time: you're sorely mistaken if you think any random Chinese millshop can't produce a 100-year-old design just as well as any American one.

They can't produce those designs or even gun parts without risking the death penalty, so no private business will go near them, and they definitely won't specialize in it.

> any random Chinese millshop can't produce a 100-year-old design just as well as any American one.

Once. They do it once, then there is a huge death penalty trial that is rehashed on CCTV for weeks. You definitely won't get any scale out of it, no one will invest in something that will get them executed, so definitely nothing near to what the Americans can produce as the world's most economical gun producers.

These days, it feels like America basically sucks at making anything besides guns. For guns, no one else really compares, and if it weren't for the Darien Gap the whole of the Americas would be overflowing with guns from the USA (also, Brazil also makes lots of guns, meaning the production is roughly divided between the USA in the north and Brazil in the south).

Digital logic design isn't software programming, and today's FPGAs are for most intents and purposes 'single-configuration-at-a-time' devices - you can't realistically time-slice them.

The placement and routing flow of these devices is an NP-Complete problem and is relatively non-deterministic* (the exact same HDL will typically produce identical results, but even slightly different HDL can produce radically different results.)

All of these use cases you've mentioned (AV1 decoders, NN layers, but especially a JS runtime) require phenomenal amounts of physical die area, even on modern processes. CPUs will run circles around the practical die area you can afford to spare - at massively higher clock speeds - for all but the most niche of problems.

Gowin and Efinix's tools are extremely spartan compared to Vivado or Quartus: they're pretty much straight HDL to bitstream compilers. There's also a FOSS implementation flow available for the Gowin chips (but I haven't used it.)

HDL isn't getting any easier, though, and that's where most of the complexity is.