I’m not sure where to put this. Maybe it’s off-topic, but I’ll post here anyway. In What can we learn from eVe V? a couple of hardware companies were mentioned, namely https://eve-tech.com/ and https://puri.sm/. I’m personally not really convinced by Eve Tech, but Purism is legit.
I’ve been dreaming about a completely fanless computer with a powerful GPU for some time. Compulab offers the Airtop. https://airtop-pc.com/
The reason I’m posting about it here is they are very Linux, i.e. open source friendly. Do you guys have any experience with or thoughts on the Airtop or similar.
Quite frankly I think we need to start getting some metal working and computer chip processing stuff going and start building open hardware from scratch. Better yet create an open hardware computer hardware processing kit so people can 3D print/produce/manufacture their own hardware/chips at home. The hard part really is getting the minerals and metals but it’s high time we created open processing instead of kept bitching about this or that company protocol.
I’m not a VLSI designer but I did look into it quite extensively a few years ago for some reason. You’re talking about one of the most exclusive industries in the world. The costs are so insanely high at most levels of the pipeline that there are only a handful of providers in the entire world for some of the steps.
One of the huge problems about hardware security is that it’s near impossible to verify that all components are safe. A CPU is about as complex as an entire city, and it’s made up of modules designed by dozens or more different companies. How will you ensure none of them had a rogue employee, were bribed by X government, or just honestly screwed up?
Well first thing that comes to mind is bring down the cost of production so one doesnt outsource production of those modules. Note I proposed open hardware production not open hardware outsourcing. Every time one outsources production one ups the cost. So I’d suggest automating the process rather than outsourcing to external labor. As for complexity: is the very existence of this project not evidence that we can solve such complex issues?
If one produces their own modules for their own CPU how can there be rogue employees?
You need a building where entire floors are literally dedicated to just keeping the air clean enough for other. The things you etch on the board are so narrow that you need to cut the mask in a way that you get the right pattern after diffraction. You need to apply numerous such masks with sub-nanometer precision. You need to model the components on the circuits accounting for all quantum effects or else nothing will work, and it takes a very advanced piece of software running on a very expensive piece of hardware to do it.
Much of the things that take building VLSI circuits is complicated, needs extreme precision, took who knows how many thousands of man-years to develop, and they are not just under copyright, but straight out corporate secrets.
“Bring down the cost” sounds simple, but so does “let’s reproduce the Eiffel Tower in my back yard at full scale” – it’s certainly a cute idea, but it will not happen.
Now, I’m all for designing them, and RISC-V is a cool thing in that direction. Please realize you’ll still have to get it taped in Taiwan or China.
You do realize computers used to be the size of rooms and cost several thousand dollars for a few megabytes. Now you can fit a cell phone in your pocket for a few hundred dollars, nevermind the cost of your standard desktop or laptop.
Now what you’re saying is it essentially takes a whole building to produce a computer chip. But if it used to take a whole room or building to house a computer and that has been brought down why does it seem so unrealistic to bring down the size of production for the manufacture of a computer chip?
VLSI is one of, if not the, the most complex area of engineering today, that does not only bring together numerous separate domains of technology, but it requires the best of each.
You’re basically talking about taking the best of the best we have, scale it down, and make it affordable.
Let’s move on to the point that you think decides it:
Because it really is. Just think about it. We’re at the level where we can probably replicate the first generation of integrated circuits at home. That’s many decades after they were first manufactured at scale.
I’m not arguing that producing today’s VLSI will not be available for the home user to manufacture in 2080, I’m arguing that at no time in history will the latest tech at the time will be available for the random guy to manufacture at home.
Manufacturing cutting edge tech is always prohibitively expensive for all but the largest players. I’ve already brought plenty of examples to illustrate why it is true for this particular field. If it’s not already blatantly obvious then I don’t know what else to say.
TL;DR Manufacturing integrated circuits at home is indeed possible. It’s up for debate whether people will abandon their 2018 gadgets for something at the level of the 60s.