4 ms·
> Turning back to nineties And 80s (consider Britton-Lee). And the opposite too: "mainframes are too complicated; we'll write an OS that does all IO in the ke
by gumby 6y ago
> Turning back to nineties
And 80s (consider Britton-Lee).
And the opposite too: "mainframes are too complicated; we'll write an OS that does all IO in the kernel (e.g. Unix) and run it on minis that don't have channel controllers". Come the 2000s and every disk drive and even keyboard has a processor it it.
And edge-vs-core computing in the network...
The best part is hardly anyone reads the literature much less history, so if you've lived through a couple of cycles you can see the pendulum starting to swing back and, to mix metaphors, "skate to where the puck will be"
- Ericson2314 6y agoAmen to that. I do welcome the return of "spread out" computing. I agree ease of use is important, but putting everything the CPU microcomputer style makes for a very imperative notion of ease of. Ergonomics without that would be much more elegant.
- 01100011 6y agoRandom thought: computing may very well be spread out one day if computational machinery is cheap to build but limited by heat. I could see computers becoming much, much larger, but with a smaller duty cycle for a particular compute element to assist with thermal management. Imagine building structures, like tables, walls, or skyscrapers, out of nothing but doped silicon.