[ / / / / / / / / / / / / / ] [ dir / asmr / cafechan / cyoa / htg / kc / leftpol / vg / vore ][Options][ watchlist ]

/tech/ - Technology

You can now write text to your AI-generated image at https://aiproto.com It is currently free to use for Proto members.
Name
Email
Subject
Comment *
File
Select/drop/paste files here
Password (Randomized for file and post deletion; you may also set your own.)
* = required field[▶ Show post options & limits]
Confused? See the FAQ.
Expand all images

File (hide): d4a580f267fb68c⋯.jpg (112.04 KB, 619x316, 619:316, Skyrmion_619x316.jpg) (h) (u)

File (hide): f50c4937c565134⋯.jpg (52.26 KB, 600x354, 100:59, memristors-01.jpg) (h) (u)

[–]

 No.897566>>898040 >>898059 >>900081 >>900106 [Watch Thread][Show All Posts]

Let's talk about computers.

We're starting to hit hard physical limits with our plain old transistor-based electronic computing, electron tunneling is already beginning to interfere with our attempts to make transistors smaller than 8nm or so function properly. The current culture seems to be pushing quantum computers using photon polarity or electron spin as the way forwards, but that's not the only new way of computing open to us right now. Currently I know of three other methods for new computing: Skyrmion computing, optical (photonic) computing, and memristor based computing.

Skyrmion Computing

>Leverages solid state physics and quasi-particles technically a form of quantum computing.

>Proper name is "Cavity Spintronics," part of the second quantum revolution quantum devices.

>Utilizes magnetic spin currents instead of electric currents, therefore using less energy and producing less heat.

>The magnetic swirls skyrmions have patterns much smaller than most magnetic domains, raising data density limits.

>Skyrmions have already been used to store info on magnetic material

>Room temp skyrmions exist and can possibly be used https://phys.org/news/2018-02-magnetic-skyrmion-room-temperature.html

>That means quantum computing at room temp.

>Absolute bleeding edge, active research going on RIGHT NOW!

<Absolute bleeding edge.

<Goddamn fucking complicated, condensed matter physics is no joke.

<So new that practical demonstration of this for computation is limited.

I really love the physics behind these and I really want this. Thankfully it seems pretty viable.

https://phys.org/tags/skyrmion/ Articles and papers on this, really requires your own research to get it

And here's an extremely early article to help clarify the whole "skyrmion" thing.

https://spectrum.ieee.org/nanoclast/semiconductors/nanotechnology/skyrmions-communication-with-magnetic-swirls-instead-of-electrons

Optical Computing

>Basically replacing electric transistors with "optical transistors"

>We're already really good with information transfer and manipulation using light wireless communications and such.

>Photons move much faster than electrons in a circuit.

>Can also be leveraged in regular electronic computing to create hybrid computers.

>Like in using optical components to speed coordination of electronic ones e.g. coordinating processing cores.

>"Optical transistors" would have a higher bandwidth than electronic ones.

<"Optical transistors" are still theoretical, more research is needed before we can actually build one.

<Somewhat difficult to make optical to electrical conversion fast enough to prevent bottlenecks in computers

<Useful wavelengths of light are in infrared and way too big around 1000nm.

<Current workarounds bring heat problems that require extreme cooling or a new material.

I'd like laser computers, but this tech seems the least useful of the three for now.

http://archive.is/wyIzS extremetech article that actually covers this entire section

Memristors

>"Memory Resistors"

>The fourth fundamental circuit element.

>Variable resistance based on previous current that has flown through it.

>Using tiny "read" charges can access a history of applied voltage.

>Was theoretical until Bell Labs found memristive properties in a material in 2008.

>Can be used for greater than binary computing due to multiple possible resistive states Largest I've heard is base-10 or so.

>Allows for easy builds of 3D devices due to structure, greatly increasing possible computing power.

>Can be used as extremely fast non-volatile memory as well as computation.

>Can more easily model neurons due to similar properties in function.

<Some people say they don't actually exist and none of this would work.

<Argument over definition of memristor and what constitutes one.

<Memristive properties only show up at nanoscale.

<Very little news and few papers about memristor tech since 2015.

<All previously announced projects seem to be dead in the water.

The closest to commercial memristor tech seems to be RRAM (Resistive RAM), which not everyone believes count.

http://www.memristor.org/

So what do you think anon, are you wanting to switch from electronic transistor based computing?

Do you have a favorite future computer tech?

Is there anything else that might be the future instead, or do you think quantum computers are the way to go?

How long do you think we'll be using electronic transistors for if you do want the switch to happen?

 No.897572

Jamie bring up that video of the elk solving a rubiks hypercube


 No.897822>>898009

I like the idea of optical computing and plugging in crystals for the aesthetics.


 No.898009

File (hide): 9ef84e5134dbd7d⋯.jpg (80.87 KB, 712x374, 356:187, Crystals.jpg) (h) (u)


 No.898011>>898059

If it involves quantum anything then it's vaporware. You literally have to believe in meme magic to buy that crap.


 No.898040

>>897566 (OP)

>Variable resistance based on previous current that has flown through it.

I can't even begin to imagine how much of a cluster fuck this makes circuit analysis. It's already bad enough with higher order differentials all over the place without having to take into account previous fucking circuit states.


 No.898059>>898432 >>898625

>>897566 (OP)

Very exciting, but you left off one other thing I think is deserving of greater exploitation today: Analog computing. Obviously not for general purpose applications, but as coprocessors for specific applications, the speedup could be phenomenal.

>>898011

It's true that "quantum speedup" hasn't been unambiguously measured yet for annealing systems like D-Wave's, but at this very moment we're on the cusp of gate-based quantum computers very probably breaking the barrier to supremacy over classical systems:

https://www.sciencedaily.com/releases/2018/04/180412154446.htm


 No.898261>>898298 >>900091

I for one hope we don't pass the silicone barrier, then software will have to stop sucking dick instead of being a bloated shitpile that can keep functioning thanks to expanding hardware capability.


 No.898298

File (hide): d58728509b89536⋯.jpg (1.02 MB, 1750x1313, 1750:1313, FREEDOM.jpg) (h) (u)

>>898261

>silicone barrier


 No.898432

>>898059

These people don't even understand how Quantum computers work, and not even the ones developing it understand the principles.

You're falling for another huge marketing scheming, the Computational Tesla, the Mecha-Jew.

https://www.youtube.com/watch?v=IiLplTc8rQY


 No.898454

While you might not consider it a new form of computing, 3D integrated circuits expanding on the current form of computing allows for more condensed circuity which reduces power, heat and increases performance (these three things are obviously all related to each other).


 No.898466

> not just stacking silicone chips with tiny water channels in between them


 No.898488

Something like https://en.wikipedia.org/wiki/Synaptic_transistor sounds pretty cool. Trying to emulate the human brain seems like a good direction, since nature already did the work.


 No.898625

File (hide): 9fd46ee4f647fc0⋯.jpg (30.09 KB, 298x403, 298:403, any_day_now.jpg) (h) (u)

>>898059

>on the cusp

We've been on the "cusp" of all the vaporware modern, untestable, theoretical physics have been promising us for almost 50 years now.


 No.900081


 No.900091

>>898261

You mean you don't want to buy a cutting edge quantum computer only to have it bogged down by programs written in an ultra high level programming language under a million different layers abstraction that can be used by soccer moms and tumblr tards?

freak


 No.900106>>900658

File (hide): 0257d6cce902ad5⋯.jpg (121.19 KB, 400x275, 16:11, cores.jpg) (h) (u)

>>897566 (OP)

Have to admit I'm really new to all of this, so I read a bit of the cavity spintronics articles. I'll work my way through more. I think it is interesting that computers might be going back in the direction of magnets. I also always heard that binary was not the way quantum computers would express things, but this field (no pun intended) seems to suggest otherwise.

Of course this is all neat stuff. I just hope it winds up viable/inexpensive/not bogged down by red tape.


 No.900658>>900757

>>900106

Many facets of QM.

"Quantum Computing" in the popsci vernacular refers to using the superposition of states to allow qubits to represent a combination of 0/1. In the 90s a mathematician named Peter Shor created an algorithm for factoring integers that utilizes these entities. It is much more efficient than the Euclidean algorithm. If implemented in hardware then most of our public key crypto (which relies on the difficulty of factoring the product of two large prime numbers) has been broken.

The spin current idea in the OP doesn't reference superposition--it suggests thermal efficiency can be gained by replacing electronic current (the movement of charged particles) with "spin current" (presumably the flip/flopping of the spins of a network of interconnected magnetic domains). Spin is a quantum property hence "quantum computing" of a different flavor. I haven't studied any of this for years but there's no theoretical reason that these two types of quantum phenomena can't be combined. Spin states can be superposed just as well as any others.


 No.900757

>>900658

Well, that would be a hellova thing.


 No.900767>>900794

Saw this in local news, now found source in English https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5454531/ http://www.ceskatelevize.cz/ct24/veda/2452516-tisickrat-rychlejsi-objev-ceskych-fyziku-promeni-ukladani-dat

>Antiferromagnets offer a unique combination of properties including the radiation and magnetic field hardness, the absence of stray magnetic fields, and the spin-dynamics frequency scale in terahertz. Recent experiments have demonstrated that relativistic spin-orbit torques can provide the means for an efficient electric control of antiferromagnetic moments. Here we show that elementary-shape memory cells fabricated from a single-layer antiferromagnet CuMnAs deposited on a III--V or Si substrate have deterministic multi-level switching characteristics. They allow for counting and recording thousands of input pulses and responding to pulses of lengths downscaled to hundreds of picoseconds. To demonstrate the compatibility with common microelectronic circuitry, we implemented the antiferromagnetic bit cell in a standard printed circuit board managed and powered at ambient conditions by a computer via a USB interface. Our results open a path towards specialized embedded memory-logic applications and ultra-fast components based on antiferromagnets.

This seems pretty big.


 No.900794>>900795

>>900767

I want to learn about this shit more and more. I was up all night reading about it the other night when this thread was posted.


 No.900795

>>900794

' clearly not the pcb implementation bit though. That is amazing




[Return][Go to top][Catalog][Screencap][Nerve Center][Cancer][Update] ( Scroll to new posts) ( Auto) 5
21 replies | 4 images | Page ?
[Post a Reply]
[ / / / / / / / / / / / / / ] [ dir / asmr / cafechan / cyoa / htg / kc / leftpol / vg / vore ][ watchlist ]