What do you think about the idea to carefully choose the hardware components more compatible with FreeBSD ?
This you have to do anyway. Picking hardware not compatible with FreeBSD was missing the point of the idea completely.
I explained you already, that's putting the cart before the horse.
Picking the hardware is roughly estimated 1% of such a project. For app. 99% it's a software project:
See what it all needs to get Steam games running under FreeBSD at all
satisfyingly. For that you don't need to bother about hardware at all, at least not now. Just use what you already got, maybe spend your machine a better GPU - but first: get steam games running on your machine, see what it needs to get as many games running as possible, fully automatically, since
nobody will install, configure and tinker with a gaming console's system, not even manually update anything, except a few nerds, not interested in using it for its original purpose.
Once you've figured out that - what software is needed (wine?), what modifications, and configurations have to be done,.... having your gaming console realized without the gaming console's hardware, but on a common desktop PC -
then you can start looking for suitable hardware, look for building a new, compact computer. First several prototypes, then miniaturization, then pre-series models, then series models. One does not start with the last one.
You already present some nice looking enclosures, while you have no idea yet how many RAM, which CPU, which GPU, how much storage capacity you will need for this thing, what PSU is enough to power it, how many watt thermal dissipation all this delivers, if that enclosure you presented is capable to deal with that heat - not even if this all fits into it. Unless you know all that, it's pointless to look for any hardware at all.
If you want to create something for to be sold in series production, you have to match the target as close to 100.00% as possible. Any variance from the exact match will either provide an unsatisfying performance or increase the production cost so the price, which leads in either case to unsatisfied customers, so losing money - lots of money.
For to hit a target you first need to know the target. You cannot hit a target you don't know.
You don't start building a ship by chosing a nice looking propeller, and then see how to fit the rest to it.
If you do so, it's pure coincidence if you get a swimming boat at all, but it will be for sure not a satisfying one, since there are always compromises to make. And when you start with the least part, your compromises are doomed to fail. So always start with the important ones, then see the lesser ones fit to it - not vice versa.
This is one of the battery protection circuit boards I designed some years ago for mass production. It contains of two MOSFETS, one protection IC, 2 capacitors and 3 resistors soldered to a PCB. So very, very simple.
5k pcs per lot charge can be produced for app. 0.90 € per piece within 6 weeks including shipping, tax etc.
This may understate the fact its design took me almost a week, based on more than three years experience building such circuits: many prototypes, evaluations,
tests, similar circuits I built before, experience in electronics mass production etc. etc. For each part I evaluated hundreds of parts from dozens of manufacturers.
That's common development engineering.
Back in those days there have been no ChatGPT, but there already were circuit simulation. But those were of no use to me, since for the parts I'm using there have been no models. And I need to chose the parts to fit the application, not asking the customer to fit his application the models I have in my computer. To create models of my own had been possible, but had been more effort than doing it without simulation at all - the board is simple. Anyway I had to do real world tests. One cannot trust a computer's output when it comes to the real thing.
Any simulation - and if it's some AI - can only be as good as the models its based on.
And no simulation can respect every situation.
So real world tests are always mandatory anyway.
Plus for any product to be sold, several market approval tests have to made and fulfilled, like at least
CE and for consumer products also better
GS for the european market,
UL for USA, and several others for other markets (China, Japan, Canada,... I don't stress them all here. Take a look on the bottom side of any consumer product to see all the certification logos to get an idea.)
Not having those - no sell. Or prison.
And don't you think, you get those certificates for free.
(WIP FreeBSD notes)
) who likes configuring stuff before playing the game, and Windows has full-compatibility with games, and Linux a large amount. Why be limited to about 100 of em?
)
