[infurl]

https://www.techradar.com/news/should-we-fear-the-next-generation-of-supercomputers

This is an interesting article which considers the technical and economic factors shaping the development of supercomputers. The recent commissioning of the Japanese Fugaku supercomputer almost tripled the performance of the previous most powerful supercomputer known as Summit, but building more powerful computers isn't just a matter of spending more money and buying more hardware to stuff in a  bigger building; to get to the next level requires qualitative advances in hardware and software each time.

While experts predict that the first exascale supercomputer could be built within a matter of years (it would only need to triple Fugaku's performance) at this stage nobody could even predict how we might one day build a zettascale system which would be a thousand times more powerful again, but if the technology continues to advance at the current rate it might only be a matter of decades.

The most powerful supercomputers are currently capable of simulating about one percent of one human brain so it's conceivable that an entire human brain could be simulated sometime in the next decade or two. It would be of limited use though because while a brain only needs about 25 watts of power, these supercomputers take around 40 megawatts.

[LOCKSUIT]

The common household desktop computer uses 80-150 watts. And can run a (already trained) GPT-2. Summit uses 15 megawatts and you say it is 1% of the power we need, so 100 * 15 megawatts is 150 megawatts. Do you really believe we need 150,000,000 watts if 115 watts can run GPT-2?



BTW I thought this was rather interesting.....i can't seem to find intel 7 5 and 3 nm chips on Wikipedia anymore:
https://www.tomshardware.com/news/intel-announces-delay-to-7nm-processors-now-one-year-behind-expectations

Really cool looking at the shrinkage:
https://en.wikipedia.org/wiki/7_nm_process

AMD hehe
https://www.tweaktown.com/news/71047/intel-promises-process-node-leadership-with-5nm-but-in-2022/index.html

And this:
https://www.quora.com/How-much-better-is-AMDs-7nm-process-than-Intels-14nm

[ivan.moony]

Microsoft Quantum Technology Docs

"Introductory" and "Key concepts" sections are written pretty simple and well, you can learn something about underlying hardware and operating software. Very short read, it takes about 20 minutes or less, while you learn a bunch of interesting stuff.

In the same docs, later, there are instructions for open source Q# language used to write and test quantum algorithms on classical processors if you can't wait for quantum computers.

[frankinstien]

One just might be able to build a quantum computer at home! 

[frankinstien]

Here's a video of another DIY QM computer:

[frankinstien]

Here's a very interesting video about lightning:



And here's an article where quantum ball lighting was created in the lab.

So what if, and I realize this is very far fetched, we could turn storm clouds into a Bose-Einstein condensate, literally turning the sky into a giant quantum computer.

Brings a whole new meaning to "Skynet" and "Cloud computing". 

[frankinstien]

This is a very interesting video about modeling with computers. It takes me back since just after high school I ended up getting a job with Xerox Computer Services in their data communications department. But what really sent me back was the computer described in this video used vacuum tubes and a drum storage device. I left Xerox two years later and worked for some electronics parts distribution company that used a Honeywell mainframe and a network of Honeywell minicomputers, but from there I went to work for Computer Science Corporation. Because of my electronics background from the correspondence courses from the "National Radio Institute" (NRI) while I was in high school it actually qualified me to be the project lead of implementing a new specially contracted device, drum roll please, wait for it...The new device was a solid-state peripheral storage system, what we might call an SSD, but this was using volatile ram and replaced existing magnetic drum storage units. Now, and here's a second drum roll, wait for it, the mainframe these SSDs were for was the Univac 1108 and 1180! OK maybe that doesn't ring a bell but does the Apollo moon mission ring a bell, the Univac 1108 was used to land astronauts on the moon. And for the third and last drum roll, wait for it, the company that actually manufactured the solid state storage devices was Intel! It was pretty amazing to see a huge room of drums replaced by a refergerator sized cabinet, it really was errie has to how much room was saved.

Oh, memories are poping into my head,  in my last semester at high school I was short 3 elective credits, elective credits! I asked the school to consider my NRI courses to fulfill the defecit and they refused, so I never graduated from high school but I did end up getting an equivalency certificate and then got a job with Xerox. Eventually I did get a 4 year degree and later a Masters degree but much later in my life...

[MikeB]

So what if, and I realize this is very far fetched, we could turn storm clouds into a Bose-Einstein condensate, literally turning the sky into a giant quantum computer.

You can't mess with clouds it can drive them away.... the ionisation of small particles in a low pressure system causes it to rain... Can't mess with the Ionosphere either.