Sounds like a great idea
What can go wrong?
A team lead by US Army researchers has been granted a patent on a new chip design that offers radically faster processing speeds and could form the basis for quantum computing systems in the future. Ron Meyers, a quantum physicist with the US Army Research Laboratory, told The Register that the new chip design is capable of …
What can go wrong?
We want The Turk back, NOW!
My pacemaker has fallen in love with my sandwichmaker!
I had a neural chip design idea once. Effectively a single powerful chip at the top layer which splits tasks between smaller chips which then further split the tasks between even smaller chips.
Say for example, a single core 5ghz processor sitting atop a quad core 2.4ghz processor, with each core sat atop another quad core running at 1.2ghz. Top layer processor allocates tasks to the mid layer cores which then split the task further among the bottom layer cores.
Rather than a constant stop / start of processes each one gets it's own core effectively. (or a much larger share of each core running at a lower speed)
I know realistically the design is retarded but it was an idea from back in college, feel free to patent it.
No.... it is a great idea and expect that you will see my rather delayed post below. There is a great simularity between ideas...
I sold computerised medical equip in 1970 and the PROMS were hand sown graphite rings.
I sold a 3 meg disc in 76 and the disc was 15cm and you had to park the heads by hand. cost 5000 bucks
My point is that anything can happen.
MY pacemaker heard me say, "Be still my beating heart"
Tell my wife I love her.
That was a contentless technobabble article, with the obligatory "quantum" thrown in. Also "non-Lipschitz function" because it sounds good. That might designate a Lipschitz-Continuous function, which is basically a function that doesn't do arbitrarily strong jumps or decays into disconnected points.
Is this the bridge of the Enterprise?
Agreed. Took a quick skim through the patent, and it appears to just be a chip design that implements some neural-network algorithm elements in analog hardware. No doubt I'm missing the good stuff in the details - I haven't worked with neural-network algorithms to any great extent - but I didn't see anything more than mildly interesting. The "analog computer on a chip" idea has been a favorite bit of water-cooler talk as long as I can remember, so that's not it.
This is coming out of a serious research organization, so I think there must be something here, but the article certainly doesn't help us see what it is, or whether it's of more than passing interest.
Hi, my name's Eddie, Eddie Lipschitz
(from The Whoopee Boys)
One in which a million monkeys will reproduce the works of Shakespeare? More likely it will aim for Dr. Seuss and still fall short.
Methinks they have seen this page ...http://www.tridi.com/quantumunionkindle/miasbrain.html...
The idea was developed twenty years ago when a young man was tortured by the idea that a transistor could exist in three states. He didn't know about cubits. The question is whether it will work.
The brain described was the nearest the author could get to the young mans ideas after many years of thinking.
Just a bit of quantum tunneling and a few super micro magnetometers and the babe will fly.
One early version of a memristor was an electrolytic device similar to a tunnel diode, which reversibly switched between high and low conductance and several states between.
Yes, magnetic memristors also exist and have been categorised.
The awkward part is making the sensors; I did wonder about making a hard disk like system with the magnetic memristors on the platter and inductive power, with an array of read heads to allow updating of the states.
Biting the hand that feeds IT © 1998–2017