09:08 <sb0> diginet, if you're paying, yes, definitely

09:11 <sb0> it's more a marketing/economic problem than a technical one

09:13 <sb0> who would buy it?

09:14 <mumptai> i would buy one of 10k

09:15 <mumptai> but not 1 of 25

09:33 <larsc> one for 10k?

09:34 <mumptai> s/for/of/

09:34 <mumptai> I guess in a run of 10k units they would be affordable to me

09:35 <mumptai> talking about it, i'd probably go for two, just to have a second unit for bring-up and the usual h/w-related pains

09:37 <sb0> at 10k they'll cost several hundred € for performance levels inferior to rpi (plus the sw toolchain issues compared to ARM), and everybody will ignore them.

09:44 <sb0> the only reasonable thing you can do now is sell expensive OSHW chips for niche markets - Gaisler has been doing that for years, with a certain success...

09:48 <sb0> or you can become the master of scientific dumpster-diving and build your own fab for cheap. but that will take long (and you'll need to eat), and you'll need space (which often isn't cheap)...

09:52 <larsc> can't you just move the fabrication to the cloud? ;)

09:58 <sb0> I seriously believe #2 is possible. you can even dumpster-dive He-3 these days :)

10:15 <mumptai> what about something like europractice?

10:17 <sb0> multi-project wafers? still won't be cheap... if you do something for the masses, you are competing with rpi, and so your chip cannot cost more than a few dollars and needs to be fast.

10:18 <sb0> but if you have a couple dozen thousand euros to blow on one chip, and time to do the layout yourself from the misoc rtl, then go for it...

10:30 <sb0> I wonder what will happen to lowrisc - will they manage to make working chips, and will they sell?

10:31 <sb0> both points look quite difficult...

10:43 <mumptai> as always the market is basically against such projects, as they don't have the mass for mass production

10:47 <sb0> yeah, so you can say 'fuck you' to the market and go dumpster-diving. I'm sure there's a lot to of cool stuff to learn on the way :)

10:47 <mumptai> or create a market that appreciates open-source approach to soc design

10:49 <mumptai> besides millionaire hackers with a love for hardware, i could only thing of obsolescence burdunded industries ... but those are usually already stuck with something that is already a few decades old

10:56 <larsc> well, you'd have to provide a feature that other products can't provide

10:56 <sb0> mumptai, yeah, expensive chips for niche markets is viable, cf. gaisler.com

10:57 <mumptai> but see that gaisler did a decade of campaigning in the european space market

10:58 <mumptai> and SEE hardness and radaiation hardness were/are a bit tricky (at that time)

10:59 <sb0> I didn't say it was trivial, I said it was viable :)

10:59 <mumptai> oh, it is

11:00 <mumptai> but it contains more than the core technical aspects