Are Optical Computers the Future of Computing? - with Martijn Heck 

Martijn needs another round of applause here for his teaching technique. Layering pictures and stories of everyday objects directly on and around the highly technical schematics has made a photonic enthusiast of me. Great talk Sir Heck!

This was an absolutely excellent lecture. For someone like myself, who started out using a slide rule, the development of computing technology and especially the miniaturisation of electronics has been breathtaking in it’s speed of development and implementation. As a kid, photonics and A.I. was the stuff of far out on the rim sci-fi. Now they are reality. On both a technical and an intellectual level it has been a truly wonderful age to have been in existence.

As one who has spent forty years researching optical computing and neural networks, this is a very good introduction. I am a bit surprised that limited progress has been made in the exploitation of the massive parallelism capability of Optics (as illustrated in the talk by the use of the double Fourier Transform system). This was mainly due to the lack of stability of existing liquid crystal light valves.

In 1983 my dad used optical computation to study "Spectral consequences of photoreceptor sampling in the rhesus retina." With a laser, tinfoil, a pin and an optical table he accomplished computation that would have required major time on a large computer.

When it proves difficult to increase the power on a chip, programmers will have to start becoming creative with code. It has been too easy to just write billions of lines of code without thought to streamlining. The old Commodore 64. When it first came out the programs were unsophisticated. Without access to increased processing power, the programmers managed to find ways to overcome the hurdles of the day, power, memory, etc. By the end of its reign, the old 64 was displaying far more sophisticated and interesting programs. Mainly games of course. Then again, games are really what has driven PC development. Even the Amiga improved once the programmers got to grips with coding. Since the advent of Windows 95, there has been a steady reliance on chip speeds to push heavier and heavier code. I think this will have to change as access to faster chips falls off.

For someone who knows little or nothing about computer and chips it was a good overview. Even though he went through different computer fields like machine learning without adding nothing to the optics topic a part of making the presentation longer. When he finally started digging into implementation details the core content could have been made in 10 minutes presentation with a help of some animations. It wasn't bad as presentation but I'm a bit disappointed because I didn't learn anything new

Oh my! Just what I was looking in a long while to have a high-level view of the field! Amazing talk as always by the Royal Institution! Thank you for making all this content available!

Outstanding lecture understandable to the layman!

Amazing talk , thanks for breaking down this state of the art technology , and putting it in such a way , that a Mr Average like me can understand what you are saying Martin ,. Really appreciated .

Good analogies here by Martin makes ver easy to digest.

A Royal Institution lecture aims at a very broad audience. Viewers knowledgeable in the areas of computing and semiconductor technology should skip the first half hour, which recaps the last 70 years of electronics.

Awesome to see this. I thought of this possibility in high school about two decades ago and even built a few PoC prototypes.

Great content , always wanted to know more about this field

What a great talk. A million thanks 👍 the photonic integration reminded me of the movie Zardoz. The crystals that were used as a storage medium.

Exciting! In my PostDoc project I am now developing a new technique for synthesis of BaTiO3 for applications integrated photonics, so this topic is close to my heart 💖

All the good stuff begin around 50 minutes mark. The rest is history.

Oof the audio is quite something

At 7:09, computers don't use bipolar junction transistors (BJTs) like the ones in this explanation. They use MOSFETs, which have very different symbols. The transistors you have pictured have neither a gate, nor a source, nor a drain.

Thank you! I really enjoyed this. And to Dr. Heck. Very enjoyable presentation. I sincerely had no idea :)

So interesting. Thanks for sharing. Do we call every thing we don't understand a singularity? Haha

How did they know to suggest this

I watched this on the TV. My cat absolutely loved the red dot.

You definitely cannot get smaller with optical chips...

Thank you so much Prof. Heck. I learned so much from your talk. The future of computing truely looks amazing with Photonics.

I think the growth in energy use is more related to a growth in features, redundancies and security. Also industry in an effort to remain competitive rarely seeks the most efficient algorithms for their customers, rather the simpler algorithm is the fastest way to market. The comprehensible algorithm in the mind of the programmer isn’t necessarily the best. If reducing energy use is a goal...time is needed in the production process

PIC totally can be applied to computing. Electrospinning new types of thermal resistant fiber optic wire conductors along with new thermal resistant transistor tech and new transistor composition tech. Imagine the time when we create zircon base comp vs silicon base comp at commercial scale

I'm sad I cannot live till 2100

great talk, well spoken

Heckin' ace, I have been saying this for literally ... two or three years!

Nice talk, very informative. I'm wondering about the way you state the fundamental problem. Is it worth it for us to continue trying to maintain our exponentially growing information technology? Should we curb our enthusiasm for this type of progress? Is the progress worth it and is it ultimately making life better? I don't have the answer, but it seems important consider both sides, why it is worth it and why it might not be worth it.

Just add in more colors into the laser to boost the efficiency of the signal. Quite remarkable and probably often taken for granted as time passes. Just let it sink in for a moment.

Martijn. are you cooperating with ASML to produce photonic chips? Looking forward to practical stuff, long time listening to great promises. Hope to see things coming up...

There's no option for english subtitle, please add it.

Moore's Law is a Theory. Assign some bytecode to each color of white light's spectrum wavelength, then use the color of the light, to be assigned as bits. Use some photonic detectors, to detect and transmit the lights color data byte code to the memory, to be further processed. These memory chips would also be light based. When you want to add two color's together, you get a new color wavelength. That new wavelength is the delta which would give you the computed new byte as a new color wavelength. Possibly could use a combiner prism for addition, a diffusor prism for division. Overlay for addition. And Separation for subtraction. If we want to utilize light's full transmission of data, we would want the full spectrum of light's color wavelengths as representations for bytes.

At 24 minutes he makes a small mistake, saying that a 100tbit connection is a hundred times faster than the fastest domestic connection. Well, afaik those connections are usually 1gbit (I have one at home) so that means it is 100.000x faster. Thats quite insane over 1 single fiber!

Very interesting lecture. Photonics = future computing

Start is here: 19:50

Optical computing is a complicated creature.but when you start to see the simplicity clear screening layers to digital switching.

Very very good!!!!

I used to work for a large computer manufacturing company. In the business, we called them photonic computers.

excellent, thank you

Well done, thank you Dr. Heck. I was expecting some discussion on holographic data storage and program execution though, and you got real close to it with the Fourier transform lenses, but it didn't happen. I wonder, is that because you didn't want to go off on the topic, or could it possibly be that you don't know about it (?), which I would find hard to believe as smart as you are.

Nice presentation.

Actually, I think we can have an idea of what a more intelligent being could do. It's based on the study of organisms - what is called a super-organism. What I mean by that is that we can look at how groups of people do things. Using a statistical method, you can basically calculate the combined intelligence of such a group. Of course we can't predict unpredictable things, like whether something that intelligent could get psychic powers or exist in multiple extra dimensions, but it could give a functional ballpark estimate...

That’s really amazing stuff

goed bezig, bedankt!

engineer- we still don't have enough'a compyoutin' powr cap'n! science officer- you will find that the lens assembly always had your answer Mr. Scott. 🤣

What exactly is holding this back from being public technology... especially for smartphones?

Can we build transistors from Quarks?

The dutch 'heh' as opposed to the canadian 'eh' i love it

I am just a layman, but an overclocker and know how pc's work. Question: Since he saying photonics would be suited for communication on the computer, wouldn't this be great for connecting the CPU to PCIe(specially the graphics card🙂)CPU to RAM and CPU and the chipset? Just think of the PCIe speeds we could get, and how much lower latency from the peripherals.

Brilliant talk and thank you for the explanation for a layman. Thank you from a armchair scientist.

Thanks for that.

A vr headset like the oculus 2 pro or meta headset, and games could make many sales, or some kind of talking AI that can learn how to talk , and be bluetooth worn on the body and have a app as well. Could pay for more technology advances. Or a box at home that connects to the internet-based and connects to the app of the smartphone. All or any could pay for technology advances. Would be the best to detect what emotions are in someone's voice when the app is on when talking to someone and the Ai learns to talk as well. I also would like to know when anyone or anything is moving towards me and at what speeds with some kind of outside glasses I can connect to my smartphone with any earphones as well.

Thank you for posting this. I was putting up fiber cable for an internet company and that's when I realized you can make an optical chip. If the internet is ones and zeros with light going on and off with different color light to decipher all that data then we can literally break speeds into the thousands of gz

The prof gave me idea: hybrid CPU’s, electronic and photonics! With electrical applied to speed up,or slow down light, thereby making,ichting waves add up,or substraat… light v could even be used to have analog meaning, so making a signal so that it can have a wide range of values. That of course exceeds the possibilities of qua tum o outing by an infinite amount, as quantum computing i essence is digital, digital like at least… a light signal can interact with another light signal,i a very varied number of ways… two signals interacting can have many different outcomes, resulting signal(s!)… that wouod increase the computing possibilities enoirmously, and the computing power of a single switch could be 100’s times what it is digitally… a single signal could eg have values ranging from 0..1024. The switch that signals acts upon is a lot more complex, but as it is optical, it can be made to behave in very different ways, depending on the i out value; duplicate, add, substraat, becomezero,become minimal, become maximal, skip step, skip x steps/switches (like a jump to next switches), send back to sender, send back to sender’s sender of x steps back, erc etc. All that should be possible within a single switch… thereby making the power,versatility of that swotch a lot bigger… requiring a very different way of progra,,omg those, using those Possibilities in such a way that they do the same as current CPU’s/switches… not just and, or xor, etc… t he first challenge however is to design, kind of invent switches that use fotonics, What we need and want switches to do, what we need and want them to be capable of, that is important then… All.ogical operation, at the very minimum of course… Do we want them to be able to take on values other then 0 and 1? Then first thing is that we make switches that can assume all byte values… and also have different behaviors depending on those values. If we wouod only have THaT, we would already have a computing tech that exceeds the power of current silicon based switches by a great lot, 2^256(???)….

Does that mean we can have photonic GPUs in which there would be no moving parts, therefore... even no need for a cooling system... while crossing the LED waves of light will do all calculations?

But don't you still need electronic circuits to control the optical components? For example you need electricity to run the lasers? It seems to me that optical integration would still be plagued by the limitations of electronic components.

Are Optical Computers the Future of Computing? after 10 years of research, i can say, YES!

The pictures are really outdated sometimes

Why use light when u can use faster quantum architecture and processing that can do things that seem to violate the laws of physics and supply work arounds to the whole e=mc2 speed limit to light.

Goede lezing, en de beste man kan het lekker niks schelen dattie zo''n mooi Nederlands accent heeft. Vind ik juist heel goed. 1

Imagine if you take a cluster of photonic/quantum computers and have them "evolve" replicators like cells or DNA or self-replicating 3d printers, right now the quine programs take exponential amount of time on our CPUs to find, but imagine a replicator based on real physics. You could then use the replicator to exponentially replicate itself and build quantum computers so the entire Earth is a quantum computer, and perhaps even put one replicators to exponentially replicate itself around the SUN O_o.

Is there any working full optical computer exist?

very interesting

optics will lead into quantum fields to improve speed of propagation of information. as it is the latency of electron flow and the propagation delay of signal is limiting computer performance. i did not find this discussion useful, but more of a speculation with no direct answers. truly optical gates with amplifiers may be helpful into the future as well as gravity wave gates that are quantum fields that do not depend on signal propagation. there is still so much for us to discover at the atomic level and so much that we know that is not yet mass manufacturable. most of what we use for computers today was invented over 150years ago. other than fine tuning processes, and reducing capacitance and improved current flow and better lithography. what do we really know now? other than quantum effects that limit the size of parts, using quantum tunneling to force current thru devices, and a few other modern tricks most of the science can be considered 'old school'

With 3 base colors, binary is surpassed. How can trinary programmation make sense?

how does an optical Computer store information?

(Around the 24minute mark) Wait..wait. Can the different colors of light transport various signals, side by side, through the same optical cable? Without the (four) signals getting mixed up..along the way

I read a fiction book some 20-30 years ago about an AI global defence system. It was written by an engineer who had worked on weapons development and the like. He set the AI up so that it would take over and keep humans from destroying the world. One of the concepts included as "3D memory" which used optics internally. Anyone know what the book might be called?

So…Star Trek, but four centuries sooner than that. I like.

Optical computing+ neural network= ?

There was talk about ten years ago of using Lichen as a biological computer

what if Nvidai made a photonic Quantum chip that has a Optical circuit with a GPU beside the Quantum chip for Ai

This professor is good! Might he be Ben Hecks Dutch cousin?

Couldn't we make the chip bigger or make it in layers so we can fit more transistors on the chip

I was hoping for actual physical reasons why optical computing is faster. This presentation is quite elaborate, however, in the end all that was done was to list the components of an electrical computer and say that each of them would be replaced by miniaturized optical components. This will make computing faster. Better Fourier transforms and image recognition. Well, how? Where is the physics?

As presentations have moved to the on-line space, they have become longer and filled with more rambling. This is probably due to removing the pressures of keeping a live audience entertained and engaged. The lack of this feedback loop opens the presenter to keep talking until they are exhausted to their satisfaction, without considering the satisfaction of the audience. Looks like a good physiology PhD dissertation topic to be researched.

1 GB per second would have been unthinkable just 10 yrs ago at least for home networks. 10 yrs from now we will be introducing 100TB per second networks to the home. Partly due to homes having so many home features that use the internet and partly due to the density of 8k video, games that are connected to thousands of others playing the same extremely Data dance games, and programs that take advantage of the higher speeds.

photonics is the future, if is able to invent the "photonics transistor" = "phansistor" an make it in the same nano-milimeters as current chips.

TIL that the Morse law can be illustrated by Leego blocks. ;-)

I wonder if Quantum Photonics will be used for cryptocurrency mining.

WHOAAAA!! now you talking~ Recently I debating some PC lover about facts that their computer units are actually 'a fridge' in case of power consumption. I use laptop and feels bit "righteous" after those PC users. But now WE ARE, ALL OF US just same pricks bc this facts. Now we really need solution for efficient and low energy consumption types of devices/computers or those talks about climate change goals & solutions are just an empty cans.

Yes, they are. Why? Because a group of people want it to work... This makes logical sense...even on a quantum level.

een munt heeft ook een randkant, dus dat klopt wel...

What if we can mix binary with others. Maybe use "e" like plank constant make a "e"naray. Then mix in quantum process for specialized process like what "JiuZhang" did. Optical has the capacity to merge both sides together.

22:11 fiber vs. Cu data speed? & (color) "What u see ..." xD "Why There Are No Bluish-Yellow Crayons: The Forbidden Color Experiment" color Light in eye: yellow+blue=white & red+green=white

Huygens Optics "Making Optical Logic Gates using Interference " 16.01.2021

Mrs Richards: "I paid for a room with a view!" Basil: (pointing to the lovely view) "That is Torquay, Madam " Mrs Richards: "It's not good enough !" Basil: "May I ask what you were expecting to see out of a Torquay hotel bedroom window? Sydney Opera House, perhaps? the Hanging Gardens of Babylon? Herds of wildebeest sweeping majestically past?..." Mrs Richards: "Don't be silly! I expect to be able to see the sea!" Basil: "You can see the sea, it's over there between the land and the sky." Mrs Richards: "I'm not satisfied. But I shall stay. But I expect a reduction." Basil: "Why?! Because Krakatoa's not erupting at the moment?"

A transistor is essentially a valve, which allows a voltage to pass. Simplifying TTL below (0.8) 0.4v =0 and above 2.2 (2.7) up to 5v = 1 THIS IS VERY SIMPLE But how to create valves for light? We can make a computer even with valves with water, but this is not practical.

This is interesting

Wow!

I always "saw" tennis balls in Yellow

There were only few pyramids. Like that, there are fewer and fewer people who know how to make these things. Unless you disseminate this knowledge rapid enough, it will end in peril. Best example is the current chip shortage.

Exponential grote Himaliars 😆

Using optics for data transport makes sense. However, that is NOT new. Development to use optics for computing or storage is just NOT practical. If you want to make a computer x2 faster, all it takes is adding another processor and an interface. There is no new development required. As far as optical transport across a single chip, I have serious doubts that this would be widely adopted.

Biological computers ?

A full added but sensory overload you can't tell what's going on it's all up to them it's good huh be invisible's good huh

wish i7's were old in my reality 🤣

****steenkolen engels*** VERY THICK DUTCH ACCENT*** And of course, optical is the future, bery much ore so then quantum, which, imho, is a hype of which the potential advantages are very ,icy exaggerated. When you however look at constraints of hardware solutions, and which tech has most potential,it is very obvious that optical is the future; very low signal strengths needed, so little energy needed, thereby avoiding heath problems Cq cooling problems, connections can be very much downsized, much more then electronic switches can,,, With 5nm we very much approach the limit of how much smaller we can make silicon based processors. Perhaps other element that behaves same or better then pure silicon. But problem, in essence, remains, not so with optical, we’re higher light frequencies lead to more compact switching…. All that said, your presentation, I’d suggest you work on your pronounciation of English. “Steenkolen Engels” we wouod call your English i Dutch. Your accent is clearly a Dutch one! Not so bad, by far, as eg that of Su.toch mi kater president Rutte and many another dignitary,political Dutch representative. YOU explain working of cpu, instead of going into tthe ch incal hardware constraints… the explaining is th same for optical and siLocon, while it is the differences we are i interested in! Speed, and capacity not only determined by scale, also by the speed i which materials can react to changes. The reactions are not instant, and the more power, the speedier the reaction, often, at least to a certain level, with/in silicon. Light is always instant, so always speedier reactions… The quality, size and apex of light detectors is an important constraint with optical. But with interference manipulation, we should be able to make optical based CPU’s that are if the same size, as the current size is in the larger part of cases not a problem, but 10’s of times faster ANd with more transistor like connections/switches… aso, with optical, we should be able to extend moore’s law by yet another few decades. My own, Imho estimate is 30years… and with doubling each year, ^30 is a lot… a single cpu rivals a whole supercomputer of the current t year,in just 10,years time … in other words: an optical based computer with SOC tech,sized iPhone 13 mini, will have the computing power of a $3.000.000 supercomputer of current time…. in a single decade, CPU’s will be sooo good, small and fast, that software will become the ,ain clallemge; how to use unlimited power of computing and storage,when processing and storing becomes virtually instant, no matter the size of data or amount of computational steps, the main question becomes: how to use that power, what to use that power for? One thing fir sure: we will simulate a lot more, often and more and ore complex things, running thereby against our levels of understanding, and of system behavior. Feedback and forward loops for example will become common language, as everyone that is a bit computer savvy will use simulations. And the ease of which that can be done wilol increase a lot further, so human interface will know several revolutions. Voice controlled computers that become more and more intelligent, ie good In understanding what user wants, will be one of the first major challenges resulting from computer power becoming much larger and cheaper… matchbook sized computers, doing 1000Tbps computing and being connected with lots of other super small supercomputer like devices at speeds of 100Tbps… with the number of devices becoming 1e9 fold the current number, them knowing virtually everything, also what is in your fridge for example, all of them communicating, that would take a few AI applications to g ing that rise to a coins ious super intelligent,far exceeding our own! It alas, lacking human motivations, values, etc etc, and even without self interest, unless, of course, that is programmed in! With a well,programmed AI application, for example, financial systems will recolutionise towards high speed trading, making latency times the most important feature of any computer dedicated to making trading decisions.l. WORK ON YOUR ENGLISH, ESP PRONOUNCIATION, ****Steenkolen engels**** Fhe professor answers his own question: When we want to see co continuation of moore’s law, we will have to revert to photonics…