@sun@shitposter.world You can I think buy a similar motherboard from a different ven-dore and install a 6502, yes?
I would like a 6502 or Z80 coupled to a 56001 DSP
(And yeah a 68356 would be cheating)
ral
@nom@mk.spook.social
1776 > 1976 > 1984 > 2020 > 2026
They say Openclaw is vulnerable to attack. Much ado about little.
The Internet is vulnerable to Openclaw.
I no longer make hard right turns on the 405 fwy doing 70mph at 3am in the morning
Favorite source of power: A cardboard box filled with violence.
Posts
Latest notes
No posts yet.
@sun@shitposter.world VGA is not magic. It's made of gates... I had maybe the first truecolor VGA board on my desk at Everex made from mostly 7400 chips/discrete logic once. It took up 3 ISA slots an I wasn't allowed to touch anything!
It's probably the FPGA that contains that logic that's the cost. Probably there's a cheap one now that can be programmed with VGA logic.
@sun@shitposter.world @SuperDicq@minidisc.tokyo Yes indeedie I think we all know that openai or claude on a great GPU are just dandy... but the local models are "good enough" to build horizontal product already.
Also, many robotic AI tasks can be done at home with fallback to the cloud. I mean it's pretty obvious.
A lot of systems are going to REQUIRE fallback to local models so please think about that when you're designing AI products in the near term for use at home.
@sun@shitposter.world @SuperDicq@minidisc.tokyo Not true, really, for those of us in the US who can extrapolate/have some modicum amount of imagination.
Definitely true if you've become a slave to the big ai modules / are trapped in lockstep greater Asia.
@sun@shitposter.world @SuperDicq@minidisc.tokyo Well I'm guessing you can install ollama on an Android/Termux + bypass OpenAI... can run the pieces wherever you like. So it can be "distributed" however it suits you.