Comments on: Sandia Pushes The Neuromorphic AI Envelope With Hala Point “Supercomputer” https://www.nextplatform.com/2024/04/24/sandia-pushes-the-neuromorphic-ai-envelope-with-hala-point-supercomputer/ In-depth coverage of high-end computing at large enterprises, supercomputing centers, hyperscale data centers, and public clouds. Mon, 06 May 2024 17:10:02 +0000 hourly 1 https://wordpress.org/?v=6.7.1 By: UK https://www.nextplatform.com/2024/04/24/sandia-pushes-the-neuromorphic-ai-envelope-with-hala-point-supercomputer/#comment-223715 Sat, 27 Apr 2024 22:36:11 +0000 https://www.nextplatform.com/?p=144041#comment-223715 In reply to Timothy Prickett Morgan.

Yes, it is. And the end of the story some thousand years ago was, that nobody was understanding each other anymore and the great project ended.
It is already starting and happening: My trust in everything written, especially on the net, is already decreasing faster than in the past, as it is known that text parts are generated by AI, which hallucinates and has deep failures…same for pictures and videos…trust will decrease…mistrust increase…
The next step for burying it are over-enthusiastic people like Huang or Keller. As already commented here somewhere, about what will happen, when (made by erroneous people) erroneous AI will make erroneous code…the ending of this is clear.
Same, as already commented here somewhere, what happens to people’s knowledge and thinking, when more and more outsourcing the brain to machines…it will decrease…by this, the next developers of hard- and software will make even more mistakes as being more erroneous itself any relying more on AI being also more erroneous…it is a perpetuum mobile to destruction…it drives itself downwards…sawing the branch you are sitting on…

God has time and gives time to turn, as he decides what we know and what not.
Psalms 94:10 …even he who gives knowledge to man?
The bad thing is, that really some people believe in AI, IT,…they really believe what they say about AI…
Psalms 2:4 Then he whose seat is in the heavens will be laughing: the Lord will make sport of them.

And even worse: It is God himself who does this, for don’t believing the truth (his word and his son, the only saviour who paid for our sins):
2 Thessalonians 2:11 And for this cause God shall send them strong delusion, that they should believe a lie:

That’s the worst thing one can happen: Being on the wrong way, but really thinking being on the right one.

]]>
By: mr4nders0n https://www.nextplatform.com/2024/04/24/sandia-pushes-the-neuromorphic-ai-envelope-with-hala-point-supercomputer/#comment-223685 Fri, 26 Apr 2024 19:21:30 +0000 https://www.nextplatform.com/?p=144041#comment-223685 Rain AI are to supply Open AI (creators of chatgpt) with $51 million of neuromorphic chips. Their near turn goal is to develop a neural network of such density and complexity to match the same number of “neural”-components and connections as the human brain within the next 5 years.

]]>
By: Timothy Prickett Morgan https://www.nextplatform.com/2024/04/24/sandia-pushes-the-neuromorphic-ai-envelope-with-hala-point-supercomputer/#comment-223672 Fri, 26 Apr 2024 12:56:35 +0000 https://www.nextplatform.com/?p=144041#comment-223672 In reply to UK.

It’s a Tower of Babel, alright.

]]>
By: UK https://www.nextplatform.com/2024/04/24/sandia-pushes-the-neuromorphic-ai-envelope-with-hala-point-supercomputer/#comment-223664 Fri, 26 Apr 2024 09:17:06 +0000 https://www.nextplatform.com/?p=144041#comment-223664 “We have a more interesting question: What happens if you could make a simulated human brain with 1 trillion simulated neurons, or 10 trillion simulated neurons, instead of real human brain with 100 billion real neurons? Would it be “smarter,” whatever that means?”

->Will not be smarter nor better in any way, as the human brain is designed by God, but chips by humans with errors in their thinking, as all the patches, updates and CVEs prove (…and hallucinating answers from current AIs).
Let’s see if God lets this happen or end it like with the babel tower.

]]>
By: Timothy Prickett Morgan https://www.nextplatform.com/2024/04/24/sandia-pushes-the-neuromorphic-ai-envelope-with-hala-point-supercomputer/#comment-223647 Fri, 26 Apr 2024 01:22:09 +0000 https://www.nextplatform.com/?p=144041#comment-223647 In reply to dharmOS.

I searched for cubic inches and didn’t not the answer had different units. So yes, a metric to imperial error. But more dyslexia than anything else. Anyway, it is all fixed. Thanks.

]]>
By: Hubert https://www.nextplatform.com/2024/04/24/sandia-pushes-the-neuromorphic-ai-envelope-with-hala-point-supercomputer/#comment-223637 Thu, 25 Apr 2024 22:04:02 +0000 https://www.nextplatform.com/?p=144041#comment-223637 Interesting article! Spiking Neuron Networks (SNN) are quite fascinating in their analog neuromorphic nature with in- or near-memory dataflow-like positioning. If I understand it, the basic units implement Leaky or Exponential Integrate and Fire (LIF or EIF) neuronal models (with some correspondance to Hodgkin-Huxley) in hardware, along with some possible reward plasticity (for learning, in feed-forward mode), in some asynchronous connectionist arrangement (synapses, with no master clock).

I’m glad Sandia is testing this tech with Hala Point’s Owl-brain-sized 12U system, which Intel bills as the “World’s Largest”. It should help identify strengths and weaknesses, as needed before scaling it 87x to reach “human size” IMHO. Also, although memristors may not quite yet have panned out as viable silicon fo such neuromorphs, it seems that sub-threshold transistors might be just the ticket (eg. Figs 2 and 3 in this 2023 paper: https://www.nature.com/articles/s41699-023-00422-z ). Fascinating!

]]>
By: dharmOS https://www.nextplatform.com/2024/04/24/sandia-pushes-the-neuromorphic-ai-envelope-with-hala-point-supercomputer/#comment-223626 Thu, 25 Apr 2024 14:13:14 +0000 https://www.nextplatform.com/?p=144041#comment-223626 >> 1,200 cubic inches of the average human brain

A minor correction but the human brain is about 1.2L or 1200cm^3 (not 1200 cubic inches). Another example of metric to imperial errors.

https://en.wikipedia.org/wiki/Brain_size

]]>