Photonic Chips Will Change Computing Forever… If We Can Get Them Right
Articles,  Blog

Photonic Chips Will Change Computing Forever… If We Can Get Them Right


The electrons shuffling around inside our
computers move pretty fast, theoretically their top speed in silicon is up to 100,000
meters per second. Which is good because the faster they can
move, the faster chips can do calculations, the faster I can refresh twitter. But there will always be something faster. Light. Why don’t we use photons in computers instead
of electrons? We already use photons for communication;
fiber optic cables are what keep the globe’s computers interconnected as they hurtle data
around at light speed. But when it comes time to process that data,
we convert light to electrons and run them through transistors. The conversion uses energy and electrons are
comparatively slow, so researchers are exploring how to build light based transistors instead. Transistors are the fundamental building block
of your computer’s central processing unit. They’re basically nanoscale switches that
will either stop current or let it flow. It’s important that a transistor’s output
can act as the input for the next transistor. Actually it has to be able to send the current
on to at least two transistors for logic signals to propagate, and it has to preserve that
signal quality to keep errors to a minimum. We have made optical switches, but for one
reason or another they don’t work as a replacement for the silicon transistors we use in electronics,
like because the output signal is a different wavelength than the input, making controlling
the next switch in line difficult. Or the design for the switch is too big, in
the micron scale instead of the nanoscale, so gains in speed come at the expense of size. Or the design requires a cloud of supercooled
atoms, which isn’t exactly practical. Even though scientists have been trying since
the 60s, an optical transistor that can replace an electrical one has never materialized. So instead of using our current approach to
computing and substituting light in for electrons, we might have to try something entirely different. Like trying to build a photonic chip that
mimics how a human brain performs calculations. Scientists at MIT have built a chip that uses
beams of light sent through channels of silicon to mimic a neural network. The beams split and combine and travel at
different speeds as the silicon is heated and cooled. Brighter beams represent bigger numbers, and
as these beams split, combine, and interfere with each other, their brightness changes,
effectively performing several small multiplication problems lightning fast. They’ve applied this photonic chip to train
a neural network to understand vowel sounds and have had modest success. They think they can improve their results
dramatically by making a purpose built system, instead of the jury-rigged proof of concept
they tested. But given how optical computing has historically
failed to live up to expectations, skeptics aren’t holding their breath. Before we see an all light chip, we’ll probably
see hybrid optoelectronic systems that take advantage of light’s ability to transmit
data quickly across large distances. But until transistors reach their lower size
limit and can’t improve any more, it looks like we’ll be relying on slow old electrons
to do the mathematical heavy lifting. We haven’t found our way to the future of
computing — yet — so make sure you’re ready when we finally get there, by starting
your very own website with Domain dot com. Domain dot com is awesome, affordable, reliable,
and has all the tools you need to build a new website. They can fulfill all your website needs. They offer dot com and dot net domain names,
and intuitive website builders. They have over three hundred domain extensions
to fit your needs, from dot ninja to dot dog, to dot love! Take that first step in creating an identity
online and visit domain dot com. Why would transistors have a lower size limit? Blame quantum weirdness. Trace explains why we might one day ditch
silicon here.Fun fact: there’s a disagreement whether light based computers or quantum computers
are the future, but thanks to the nature of light it’s possible future computers could
be both!Thanks for watching

100 Comments

  • Atanu Das

    Dear seeker, can u plz make video on why some animals live so longer and some a shorter life and what determines that?

  • Joshua Stevens

    There is something faster, look up quantum computers, as far as I understand it works on the principle of quantum entanglement. Feel free to correct me if Im wrong. Im also pretty sure there are working quantum computers existing.

  • Nocturnus Schwartstein

    You'd only get marginally faster if you use light, if not worse performance since photons travel SLOWER through material compared to electrons through a conductor. Unless you can make some nano vacuums in the chip and use small wavelengths.

    Perhaps we should look into sonic, thermal, or magnetic computing?

    Signal density, speed, frequency, power consumption, I can see the advantages somewhat. Unfortunately this is doubtful. And considering the wavelengths of photons tend to be quite big compared to electrons, unless you were to use highly penetrating x-rays or gamma rays you might have trouble fitting those fat particles down your tubes.

  • Muhammad Osama

    Why not use Si-Ge or GaN (or even InP)? We could see quantum computing to take some time but that doesn't mean we have to rely on slower processors in the meantime!

  • Jorge Lima

    There is a misconception that light travels faster than electrical signals, and therefore our computers would be faster if we would use photos instead of electrons. It is true that photos propagate at the speed of light in the medium, whereas the speed at which electrons move in conductors is fairly low. However the propagation of electric signals is not related to the speed of electrons. Electric signals propagate exactly at the same speed as photons, the speed of light in the dielectric medium. It is true however that there are other factors that limit the speed at which we can process electric signals that can be overcome by using light.

  • Hoodnetwork ktg

    I've been a subscriber since i was in 6/7th grade tht was about 5 years ago,and still loving the videos .👍🏻

  • Layla Azure

    If data/information can be transfer at the speed of light using things like the photonic chip and etc: Artificial intelligence development will be boosted… If that's possible… Would we see a Termintator in the future years… 😕😕😕

  • Asrat Mengesha

    What do you mean by photogenic? – do particles called photons exist or you are just giving a general name for your fast chips?
    Thanks.

  • jklw10

    doesn't graphite move electrons just as fast as light? the inertia of electrons in there should be effectively 0 because of dirac fermions and other fun stuff

  • Paul West

    My very first computer had a processing speed of about 2kbs and 40kbs of ram? lol.
    Laughable yes by today's high tech standards but it was the state of the art back then. They keep pushing the limit on size and power like it's a freaking contest thus ignoring the enemy of ever laptop via "HEAT!"
    It's fried my last 5 high tech wonders including a few PCs too!
    Lastly, i would love to see laptops with 22" screens for building it bigger would solve many problems….

  • Erudite

    Why is a cloud of supercooled atoms impractical? Yeah it requires much more redesign of our entire architecture… But isn't that the point? We're going to need to be really good at gettijg and keeping things supercooled for quantum computing anyways… May as well start building new designs for it now.

    Imagine how cool it would be (pun not totally intended) to crack open a device and see a cloud of supercooled atoms suspended in diamond encasing or something… So dope hahaha
    But seriously, technology is all about change and progress, high team we change it up these chip designs that are almost half a century old now.

    We could then use this new design to also look into building biotech artificially organic "brains" for AGI… Think Ex Machina.. Imagine being able to build nanochips like that such that we will better be able to use implants that grow with us!
    Transhumanism here we come!!

  • Chris de Vries

    @0:15 you said "but there will always be something faster" and then you say "light." The erroneous fact is that when you talk about the speed of light then there really isn't anything faster than that 🙂 So no, there isn't always something faster.
    I'm talking about the physical speed, not the computation speed.

  • Senshi Atsuki

    What about the new discovery scientist made by accident using graphene, which confine light to the one atom limit.

  • Xylos144

    This channels feels like the kind that would just report click-baiting half-truths about science and technology. The format all matches.

    But it's actually to the point, properly covers correct details in broad strokes, and doesn't overstate anything. It's very much appreciated.

    Please keep it up. The more popular you are, the more you push out the click-baity competitors.

  • He Man

    open
    me
    NOW!!

    Well hello there, thank you for opening and letting me out today, I am going to show my undying gratitude by sending positive vibes your way all day by having a pokey bum wank and thinking of you while I do it, have a great day and pass on the positivity.

    ( ͡ ° ͜ʖ ͡° )

  • EyeInTheSky

    Intel had these light chip at least 7 years ago. Don't think for a second these gargantuan corporations will release anything like this until they have too. They will exhaust the current elector/silicon physical limitations and itterations first. Just think abou it for a second. With light chips, there is no "Core" L1, L2, L3 iterations(L being light) once you have their first Chip it will be as fast as anything ever will be. There will only be one product and it does not matter who makes it as it has reached the fastest speed possible forever. There is no money to be made here in terms of small incremental steps and stretching the profit to R&R ratio for as long as montitarliy possible.

  • erling cantik

    China Copied from Alien in their space base.
    Chinese researchers develop new photonic quantum chip
    https://youtu.be/YfNMCSZ_7GM

  • Corrine Tsang

    .The proton chip was deveveloped by professor Jin Xiaomin using a 3_D printer..The photon chip together with  a fast hack proof quantum computer will give China the lead Proffesor Pan Jianwei ,whom developed the China to Europe quantum communication network is developing his first love ,quantum computer  in a 10 billion dollars research center at Hefei,China.The photon chip can be used in I-phone.

  • BluesCreation09

    Breakthrough! Chinese researchers develop new photonic quantum chip.

    Published on May 16, 2018
    Chinese researchers have developed the world's largest-scaled 3D photonic quantum chip, which marks a big leap in the development of quantum computing.

  • Roger didit

    Two days after we have so ent billions in research, the Chinese which are seeded throughout our nation's research labs and universities, will have the details and be ready to ship them to us at dumping prices to control the market.

  • hendrix4151

    im looking foward to a future where computers are used to control people and get people not to support evil behavior like abortion,polluting, immodesty, stealing, gambling, bad performance, gay marraige, sexual assault, poisoned and horrible food, drugs, porn, prostitution and violence for entertainment..satanic bipolar music, or bullying in schools. those are abominations. who ever controls the currency controls the world. you can not buy or sell without money and your survival depends on the government ruled by the traditional catholic church. i think it's been shown that alot of people abuse their freedom a quarter of Americans have a sexually transmitted disease. the 2008 economic collapse was caused by horrible behavior. liberty allows bad people to persecute good people throw them all in jail.

  • Miki Cerise

    Electrons already move plenty fast enough. It's not their speed that is a problem, it's the heat. We can run those switches much faster if want, it's just that they'll melt. 😉 That's why we've been stuck at about 4.5 GHz for awhile now. The big hope for photonics from what I understand is that it will take less energy to manipulate wavelengths of light, and so you can run the transistors many more times per second before they got as hot as our modern processors do. That is where the speed gain would come from.

    As for opto-electronic systems where information is sent using light and electronics are used to perform calculations… eh, that was weird… I had to double check the date on this video when you said it. We already have them: fibre optics. That's what I'm using to communicate with you. Well, I've been told the US is still mostly on copper and DSL. 😉 But in Europe and Asia at least most information is already transmitted optically over long distances between electronic terminals. In any case, the main advantage of fibre optics over electronics from what I understand isn't so much to do with the speed of electrons or photons, it's bandwidth. We see that as speed because bandwidth is parceled out among customers, so the more customers the less 'parcels' you get for a given time.

    But really… electron speed over wires is more than fast enough for practically any application on Earth. Even if we found some sort of particle much slower but that somehow require much less energy to manipulate it would still make calculations much faster. 😉

  • Jeffrey Photonboy

    "we'll probably take advantage of lights ability to transmit data across large distances"

    Wow, that is a fantastic idea! Wireless transmission of data… if only there was some practical use case like wi-fi, lasers or mice, IR remotes etc… sure, you meant "large" to mean between local chips likely but guess what… that is already MUCH SLOWER than just running a copper wire since instead of dealing with electron drift speeds (which is fast) you have to:
    1) use electrons to create photons
    2) transmit photons
    3) receive photons
    4) turn photons back into electrons

    Maybe I'm wrong, but is there really a use-case for SHORT optical transmission like that?

  • Andrew Kelley

    You guys I have to tell you this the best way to create way to make a optical transistor is by using three sources of light and switching between two of them then useing the interference pattern. So the guys working on this are pretty close. I mean really close all they have to do now is add that and a data line memory and they will have a working model.

  • TinFoilVeteran

    It's not the speed of the electrons that is the problem, it's the heating and complicated data management. With light computing, data can be computed in real time without needing to go through the cache, v-ram, ram, hard drives. And all without generating a lot of heat.

  • Krzysztof Filip Maksymilian Stypuła

    So what is the main problem for building photo-transistor coz I couldn't get it from this video..

  • Kris G.

    Optalysys is about to offer for sale their optical co-processor that can be plugged to a PCI-E port https://www.youtube.com/watch?v=55JK5bDLoAg

  • Ohm State

    I like when Julian explains things I seem to understand him better. It may be his tone. I dont get distracted by how he says something and can instead focus on what he is actually saying.

  • Ohm State

    I think the age of sythetic organic computers is when robotics and inherently artificial intelligence, will be on par with human capabilities; with regard to regeneration and even autonomy.

  • Leslie Porter

    Brilliant. I think we're close to a breakthrough. All that's needed is time, focus, and funding. Great commentary.

  • 磊赵

    The “black light” digital photon computer.
    Stronger crossing laser reduces the excited atom population, and stops the laser device working, then a non-gate is realized. The crossing laser is vividly called "black light".
    The electronic computer is reaching its speed limit, most of the quantum computers can only do simulation operation, most of photon computers are expensive and slow and also simulational. I invented a photon computer that is cheap faster and digital.

    1 A review to the laser principle
    When an atom of the laser crystal is irradiated by the light from the pump source, it absorbs a photon and is excited. When a photon passes by an atom, the atom emits a photon witch has the same attributes(direction frequency and phase) as the passer-by photon. The atom backs to the ground state,and the new photon has a different polarization direction, witch is random.

    2 The“black-light” non-gate

    The digital light circuit is binary. Assume that that light exists is 1,while that light non-exists is 0. To realize a non-gate, light existence must be changed to light non-existence, and also the inverse process. The phenomenon from light to no light, is vividly called “black-light”.

    At the beginning, the laser works in the left and right direction normally. To do a non operation, strong light iirradiates from up and down to the crystal; then the the laser works in the up and down direction. The reason is that: after the light from up irradiation, the count of the excited atoms in the laser crystal decreases, that leads to the stop of laser working in left and right direction. By the way the light from up must be strong enough.

  • Harley Me

    so its that simple.. you change nothing in the current way we compute things and simply swap transistor for a simple led and light sensor. why do these company's think we have to make everything so small..
    personally. I dont care if my pc case has to be 4 foot sqaure.. jesus we already own a dec alpha mainframe lol

  • Mahamad Alkarim

    If we are to use light we must abandon semiconductors & start using crystals, like a multiple layered crystal boards & chips with printed optical pathways & simply the calculating instructions can take the form of these optical pathways

  • aoeu256

    How difficult is it to etch optical transistors into glass or ice using the suns power to burn glass or ice? If you had self-replicating 3D printers out of ice (terraforming mars) and sand-glass (could also terraform Mars, but its good to build houses and things in the Sahara dessert) then via exponential growth (1 2 4 8 16 32 64 128 256 512 1k 2k 4k 8k… 1 trillion in 4 generations) you can build all kinds of legos and use these legos to bulid things. They can also give you infinite energy via using there ice/glass lenses to power heat-pumps or chemical reactions. I can see the wheels and other moving parts all being created of glass/ice, as well as the extruding "arms", but how to give each 3D printer something that can compete or follow directions to build itself seems like the hardest part. Creating gears that translate etches in ice/sand (in ice/glass) into printer commands is possible.

  • truman show

    New main technologies like AI may require the advancement of conventional main technologies

    like quantum computing, optical computing or neurocomputing, giving returns to these existing technologies.

    I hope new techs centered AI and policies to utilize them will achieve the sustainability of human civilization 💖.

    https://ameblo.jp/lovelucifer/entry-12266020322.html
    / https://ameblo.jp/lovelucifer/entry-12429842470.html

  • Jim Murray

    Above reads, “According to Einstein’s theory of relativity, this should not have been possible.”

    According to the scriptures of life, “Stop simply believing everything you see and hear without first conducting some form of personal, hands-on research and/or experimentation before blindly stumbling down the path of least resistance , only to find more of the same old, same old information. Often more times than not, this data proves to be pointless if the information isn’t understood. Question everything , analyze, research, learn, share and teach your findings supporting approval or disapproval.

    We need to make informed choices, learn and teach all we can and spread love, constructively criticize all things while embracing personal shortcomings and missed achievements as improvement opportunities.

    Academia has long enough been passing off vast amounts of toxic, unfounded information, which has infected a plethora of intelligent minds. The very foundation on which fundamental education was built, was designed gathering poor quality materials, left incomplete, deteriorating, stealing away architectural integrity from an already unstable base flawed by design.

  • იაკოპო თავდამსხმელი

    Why you guys added "domain.com" commercial which tells about making sites without code after talking about optical chips?

  • Christian B

    As others have noted, light does not really increase the "speed" at which the computing functions actually occur. In fact, in silicon photonic modulators for example, the electrical waves move FASTER than the light. We have to remember that electrical waves operate by the same fundamental electromagnetic principles as light (just at different frequencies). Copper slows electromagnetic waves less than silicon. The benefit of photonics is bandwidth, rather than switching speed, for example. Photonics is good for moving data inter-chip rather than intra-chip. This is being implemented by companies like Ayar Labs.

  • Alkeryn

    The best with optical computer isn't that photons are faster (and they aren't by much)
    The good part is that you could increase the frequency to THZ due to the fact that it dosen't heat as much
    plus you could stack thousands of layers because you don't need to cool them really
    so you could have computers that are about 1 million times more powerful for even less energy consumption
    plus you can split the signal and merge it with optics so you could have great parralel computing extensions.

  • mike jones

    thats crazy, this means you could just take a crystal, and the way it splits light could be a certain calculation, so you'd have algorithms processed instantly, you'd just have to make them phsyically first? wierd

  • Richard Deese

    Thanks! It seems we have a lot of options when it comes to computer chips. I've been hearing things (including on this channel) about Gallium instead of Silicon, magnetic waves instead of (or in cahoots with) electrons, photons instead of electrons, 3D chips single atom memory, DNA computing, etc. I hope that someone, somewhere is working on technologies to intercommunicate between these various forms. I acknowledge that some of these may represent niche technologies, which may only be used alone for special purposes. But we also need to make sure there's some kind of interrelation methodology. Otherwise, someone's going to have to suddenly come up with answers really quickly when they turn out to be incompatible. I don't exactly expect some neat, simple, universal translating switch – no, that would be entirely too easy. But we do need to think about how all these different architectures might communicate. And that's not even considering how quantum computers might communicate with conventional ones. Thanks again. 𝓡𝓲𝓴𝓴𝓲 𝓣𝓲𝓴𝓴𝓲.

  • sneha priya

    Global Photonic sensors market is valued at CAGR of xx% within the forecast of 2018-2024.

    Request free sample @ https://www.envisioninteligence.com/industry-report/global-photonic-sensor-market/?utm_source=yt-chitti

Leave a Reply

Your email address will not be published. Required fields are marked *