@csepp My proposition would be for the high performance to be related to the software part - where you do high performance software on the old hardware so it beats the inefficient software on modern hardware, or rewriting firmware to make it better.

You meant like retired supercomputers, right?

@csepp I consider creating performant code to be inherently a part of permacomputing. Less wasted work, though this can depend on the overall context.

I've used a bit more traditional HPC via message passing systems and the like, in some cases those are overblown complexity wise and resource wise. If it's really needed, it could lead to more efficient and less wasteful systems, assuming the machines are performing useful tasks 24/7 and not just idling.

Still, context dependent I guess

@csepp Do they require centralized infrastructure? Remember Seti @ home and folding @ home?

https://en.wikipedia.org/wiki/SETI%40home
https://en.wikipedia.org/wiki/Folding%40home

@csepp Maybe this is where permacomputing differs from frugal computing.

Considering we need weather forecasts, we had better run them efficiently, and that means centralised as performance is limited by communication overhead. But we can and shoould run them on "old" hardware, and use renewables to power them.

Also, when we oppose "centralisation", we should be clear what it is that we are opposing. I would say it is the centralisation of control and power that should be opposed. Physically putting compute resources in a central place does not imply this. It also depends on the scale of the compute resource, and the degree of centralisation.

@csepp What would a computer industry driven by efficiency and ecology rather than profit be like?

I figure we already have enough compute power for almost any conceivable peaceful purpose.