It has been a long time which there is a question in my mind. There are cheap s around. For huge computings on CPU, would many based single board computers like or replace a single ? If this is possible and considering that softwares are equally optimized on and ARM, the obvious advantage of a bunch of ARM tiny computers as a whole cluster over a single x86 would be that they consume much less power, thus they generate much less heat and hopefully longer product life. Also if any of them went down, your whole computing resource is still available but with a little less computing power. I have a NanoPi NEO and a 3rd gen Ci5 M series. But a piece of software is needed which could benchmark both my x86 and my ARM and on the compiler-level, is optimized equally for both.



(click to see original post this answer to)

Purely on the merits of these things, what you said works for me...

✔️ - reusing the past ARM hardware (or even newer ARM)

✔️ - funding raspberry pi more than anything else

✔️ - not tying into new tech / using old or friendly people reduces back doors

❓ ... and technical specifics I don't know of right now, but I might even accept less performance for 'longevity' or 'encouragement' of those things 'good'/more stable rather than everything else cheaper or brute-force as heat makers or costly-to-earth- complicated components subsidised by public money / taxed by banks all-the-while.

❗ ... and their owners who turn out to be quite political (very aligned with the state / gov's that enable and fund them!! Oh dear STEM what happened to you - were you always badly funded?)

- it's a small is beautiful kind of theory perhaps (doesn't create as many massive fires perhaps)

- I don't think scaling up anything works to the extent of involving / creating more problems that it solves.
(short answer: It's a human thing here, small makes people do it themselves, scaling up that encourages ignorance and short-cuts then end up making things too short somewhere else in the world! i.e. scaling up encourages more ignorance than people having to deal with their own stuff / wants / needs )

Other reasons / notes why it might work:

- 'resilience' might be better as a cluster

- Might make adding / subtracting power from number of boards more measurable in some way or useful


- Without people learning to do better decision making themselves, super-computers fail us all or encourage a lot more mismanagement by people we never see controlling / designing super-computers.

- Cheap is often manipulation to introduce things [Every thing is expensive in many meanings of the word]. Because of that (even to a weak degree) doing things only because it's cheap is often the wrong way :) in many ways 🤑 🐟 🎣
(that's industry----^ corralling fish using 'cheap' as supermarkets do.)
Imagine the picture:
"Oh look how cheap it is now! What a coincidence, I'm sure it's not political or anyone chose to push for that evolution branch"



Sign in to participate in the conversation
Qoto Mastodon

QOTO: Question Others to Teach Ourselves
An inclusive, Academic Freedom, instance
All cultures welcome.
Hate speech and harassment strictly forbidden.