by Anonymous Coward writes:
on Sunday March 24, 2019 @10:59AM (#58324772)
We already imagined that as kids in the 80s. I'm certain, our grandparents did too, on their first mainframe/supercomputer.
And back then, for most parts, they were! Backplane, bus, cards for anything, peripherals for anything else.
The thing is, that integrating ALL the things and disabling what you don't need just became cheaper. But it quickly became a trap too, since the buses and modularity were done away with too! So you could not even do it if you wanted!
I still imagine a backplane-like system with cards emerging out of the Raspberry Pi headers and stacking shields. And a similar back-"spine" system for mobile phone sized devices, where the bus is in the center and it's like a ribcage where you snap in the modules, and close the lid/battery/display/keyboard back side.
My point is: Ee dreamt about this for a long time, but until now, for-profit lock-in and cheap mass-manufacturing always prevented it. So unless you have a novel was to counter those underlying problems... no dice.
You need to catch up. Cheap mass-manufacturing hasn't prevented us from having easy to configure circuitry. In fact, the development of FPGAs [wikipedia.org] has made it possible to create and test extremely complex circuits easily and quickly.
Well, that's basically what the compute module is. All that is needed is to pull all the redundant "system" pieces out of the SOC and put them on the I/O board and a decent system bus for intercom. Not really a hobbyist grade project, but shouldn't be too hard to do.
The dream is bad though. Already I see hardware designers who treat things like legos, and it fails. You fall into a trap of thinking that everything is commutative; part A is good and part B is good therefore I assume that A+B is also good. There's pressure to speed up testing and validation and that encourages the quick and dirty approach so that the final products is.. well.. quick and dirty. Often things are fixed by by other quick hacks, if the fundamental problem lies with the board layout there's p
Luck, that's when preparation and opportunity meet.
-- P.E. Trudeau
We used to be closer to this dream. (Score:1)
We already imagined that as kids in the 80s.
I'm certain, our grandparents did too, on their first mainframe/supercomputer.
And back then, for most parts, they were! Backplane, bus, cards for anything, peripherals for anything else.
The thing is, that integrating ALL the things and disabling what you don't need just became cheaper.
But it quickly became a trap too, since the buses and modularity were done away with too! So you could not even do it if you wanted!
I still imagine a backplane-like system with cards emerging out of the Raspberry Pi headers and stacking shields.
And a similar back-"spine" system for mobile phone sized devices, where the bus is in the center and it's like a ribcage where you snap in the modules, and close the lid/battery/display/keyboard back side.
My point is: Ee dreamt about this for a long time, but until now, for-profit lock-in and cheap mass-manufacturing always prevented it.
So unless you have a novel was to counter those underlying problems... no dice.
Re: (Score:0)
You need to catch up. Cheap mass-manufacturing hasn't prevented us from having easy to configure circuitry. In fact, the development of FPGAs [wikipedia.org] has made it possible to create and test extremely complex circuits easily and quickly.
Re: (Score:0)
Well, that's basically what the compute module is. All that is needed is to pull all the redundant "system" pieces out of the SOC and put them on the I/O board and a decent system bus for intercom. Not really a hobbyist grade project, but shouldn't be too hard to do.
Re: (Score:2)
The dream is bad though. Already I see hardware designers who treat things like legos, and it fails. You fall into a trap of thinking that everything is commutative; part A is good and part B is good therefore I assume that A+B is also good. There's pressure to speed up testing and validation and that encourages the quick and dirty approach so that the final products is.. well.. quick and dirty. Often things are fixed by by other quick hacks, if the fundamental problem lies with the board layout there's p