Photonic Computer - a "supercharged GPU" with very low energy consumption

Yes, we all wish for Quantum Computers... but in the meantime we need something here and now! Could Photonic Computers fit that role?

Just about everyone has heard of fiber optics – using light for data transmission – but did you know that light can also be used for computing?

There's a new commercial product expected for early next year (2022).

I contacted the CEO, Nicholas Harris, of a 4-y.o. startup, Lightmatter, interviewed in April 2021 here.

Photonic computers, at least in their first commercial appearance, are essentially accelerator cards for Linear Algebra - and so of special interest for Machine Learning and some types of simulations.

Their claims are remarkable:

10X faster than some of the best GPUs
using 90% less energy
can be used with existing software stacks, such as TensorFlow
commercially available early next year (2022)
a lot of future growth, as additional wavelengths of light get used in parallel

My own interest is primarily for Machine Learning, Neurocomputing and Systems Biology. In particular, what will we use to power what may very well be the immense computing needed to advance Longevity Science?

Lightmatter, a Photonic Computer pioneer

According to this May 2021 article from TechCrunch, Lightmatter has raised a total of $113 million and has 70 employees.

On his LinkedIn, the company's CEO says:

During my time at MIT as a PhD student and post-doctoral fellow, I explored what may become the ultimate tool for understanding the exquisite details of our universe: quantum computers. Quantum computing is profoundly exciting, but supporting hardware technologies are not ready yet.
I've developed a level of expertise in designing components and systems that create, processes, and detect light and what I've been able to build leads me to believe that the next step in the evolution of computing is all about light.
I founded Lightmatter with a mission of creating photonic computers and new ways for chips to communicate.

Salient points from the more recent Aug. 2021 interview:

Lightmatter got a new round of funding, $70 million in E series; among the investors is Google
Commercial availability is expected for the coming year (2022)
It seems doable in the near future to use around 16-32 wavelength of light, for enhanced parallelism

Other Companies

As of Sep. 2021, no other company seems as close to the market as Lighmatter. But it may be good to keep an eye out for Lightelligence.ai, started by MIT postdoc Yichen Shen; here's a brief May 2021 interview of him.

Another company on my radar is a small Israel firm, CogniFiber, founded by Eyal Cohen. They have plans to roll out their first products in early 2023.

Background on Photonic Computers

If you want a simple intro to Photonic Computers, I recommend this Apr. 2021 article in Forbes.

For an in-depth intro to Photonic ICs and Programmable Photonics, there's a great set of 2021 workshops from Wim Bogaerts, a professor in a research group at Ghent University, Belgium.

I recommend starting with Photonic ICs, Silicon Photonics & Programmable Photonics (1/2021). Here are a couple of screen shots from it:

Encoding information in light beams

Photonic Integration

And here are a few screen shots from Programmable Photonics (4/2021), which I think is an excellent to-watch-next workshop:

Large-scale Silicon Photonic


General-Purpose Programmable PIC (photonic integrated circuit)

Matrix-vector product

Comments

Discussing Neuroscience with ChatGPT

UPDATED Apr. 2023 - I'm excited by ChatGPT 's possibilities in terms of facilitating advanced learning . For example, I got enlightening answers to questions that I had confronted when I first studied neuroscience. The examples below are taken from a very recent session I had with ChatGPT (mid Jan. 2023.) Source: https://neurosciencestuff.tumblr.com In case you're not familiar with ChatGPT, it's a very sophisticated "chatbot" - though, if you call it that way, it'll correct you! 'I am not a "chatbot", I am a language model, a sophisticated type of AI algorithm trained on vast amounts of text data to generate human-like text'. For a high-level explanation of how ChatGPT actually works - which also gives immense insight into its weaknesses, there's an excellent late Jan. 2023 talk by Stephen Wolfram, the brilliant author of the Mathematica software and of Wolfram Alpha , a product that could be combined with ChatGPT to imp...

Graph Databases (Neo4j) - a revolution in modeling the real world!

UPDATED Oct. 2023 - I was "married" to Relational Databases for many years... and it was a good "relationship" full of love and productivity - but SOMETHING WAS MISSING! Let me backtrack. In college, I got a hint of the "pre-relational database" days... Mercifully, that was largely before my time, but - primarily through a class - I got a taste of what the world was like before relational databases. It's an understatement to say: YUCK! Gratitude for the power and convenience of Relational Databases and SQL - and relief at having narrowly averted life before it! - made me an instant mega-fan of that technology. And for many years I held various jobs that, directly or indirectly, made use of MySQL and other relational databases - whether as a Database Administrator, Full-Stack Developer, Data Scientist, CTO or various other roles. UPDATE: This article is now part 1 of a growing, ongoing series on Graph Databases and Neo4j But ther...

Using Schema in Graph Databases such as Neo4j

UPDATED Feb. 2024 - Graph databases have an easygoing laissez-faire attitude: "express yourself (almost) however you want"... By contrast, relational databases come across with an attitude like a micro-manager: "my way or the highway"... Is there a way to take the best of both worlds and distance oneself from their respective excesses, as best suited for one's needs? A way to marry the flexibility of Graph Databases and the discipline of Relational Databases? This article is part 5 of a growing, ongoing series on Graph Databases and Neo4j Let's Get Concrete Consider a simple scenario with scientific data such as the Sample, Experiment, Study, Run Result , where Samples are used in Experiments, and where Experiments are part of Studies and produce Run Results. That’s all very easy and intuitive to represent and store in a Labeled Graph Database such as Neo4j . For example, a rough draft might go like this: The “labels” (b...

What are Graph Databases - and Why Should I Care?? : "Graph Databases for Poets"

This is a very gentle introduction to the subject. The subtitle is inspired by university courses such as "Physics for Poets"! (if you're technically inclined, there's an alternate article for you.) It has been said that "The language of physics (or of God) is math". On a similar note, it could be said that: The language of the biological world - or of any subject or endeavor involving complexity - is networks ('meshes') What is a network? Think of it as the familiar 'friends of friends' diagram from social media. Everywhere one turns in biology, there's a network – at the cellular level, tissue level, organ level, ecosystem level. The weather and other earth systems are networks. Human societal organization is a network. Electrical circuits, the Internet, our own brains... Networks are everywhere! What can we do with networks, to better understand the world around us, or to create something that we need? Broadly ...

Using Neo4j with Python : the Open-Source Library "NeoAccess"

So, you want to build a python app or Jupyter notebook to utilize Neo4j, but aren't too keen on coding a lot of string manipulation to programmatic create ad-hoc Cypher queries? You're in the right place: the NeoAccess library can do take care of all that, sparing you from lengthy, error-prone development that requires substantial graph-database and software-development expertise! This article is part 4 of a growing, ongoing series on Graph Databases and Neo4j "NeoAccess" is the bottom layer of the technology stack provided by the BrainAnnex open-source project . All layers are very modular, and the NeoAccess library may also be used by itself , entirely separately from the rest of the technology stack. (A diagram of the full stack is shown later in this article.) NeoAccess interacts with the Neo4j Python driver , which is provided by the Neo4j company, to access the database from Python; the API to access that driver is very p...

Interactomics + Super (or Quantum) Computers + Machine Learning : the Future of Medicine?

[Updated Mar. 2022] Interactomics today bears a certain resemblance to genomics in the 1990s... Big gaps in knowledge, but an explosively-growing field of great promise. If you're unfamiliar with the terms, genomics is about deciphering the gene sequence of an organism, while interactomics is about describing all the relevant bio-molecules and their web of interactions. A Detective Story Think of a good police-detective story; typically there is a multitude of characters, and an impossible-to-remember number of relationships: A hates B, who loves C, who had a crush on D, who always steers clear of E, who was best friends with A until D arrived... Yes, just like those detective stories, things get very complex with our biological story! Examples of webs of interactions, familiar to many who took intro biology, are the Krebs cycle for metabolism or the Calvin cycle to fix carbon into sugars in plant photosynthesis. Now, imagine vastly expanding those cyc...

Visualization of Graph Databases Using Cytoscape.js

(UPDATED APR. 2024) I have ample evidence from multiple sources that there are strong unmet needs in the area of visualization of graph databases. And whenever there's a vacuum, vendors circle like vultures - with incomplete, non-customizable, and at times ridiculously expensive, closed-box proprietary solutions. Fortunately, coming to the rescue is the awesome open-source cytoscape.js library , an offshoot of the "Cytoscape" project of the Institute for Systems Biology , a project with a long history that goes back to 2002. One can do amazing custom solutions, relatively easily, when one combines this Cytoscape library with: 1) a front-end framework such as Vue.js 2) backend libraries (for example in python) to prepare and serve the data For example, a while back I created a visualizer for networks of chemical reactions, for another open-source project I lead ( life123.science ) This visualizer will look and feel generally ...

Life123 : Quantitative Dynamical Modeling of Biological Systems

(UPDATED 8/2022) - Are we ready to embark on a next-generation detailed quantitative modeling of complex biological systems , including whole-cell simulations? An anticipated up-jump in computing power may be imminent from Photonics computers (which I discuss here ), and GPU's are rapidly gaining power as well... Are we in ready state to put existing - and upcoming - power to good use? This is a manifest, and a call to action What's Life123? It's about detailed quantitative modeling of biological systems in 1-D, 2-D and full 3-D, as well as a multi-faceted software platform for doing so. What's (pseudo-)1D? For now, let's say it's like the inside of a long, thin tube - with no interactions with the tube. Likewise, (pseudo-)2D can be thought of as a Petri dish, with no interactions with the lid or the bottom. Website : https://life123.science A purposeful decision to also utilize 1D and 2D But why? Yes, it's in part about "walk before you run...

D3 Visualization with Vue.js : a powerful alliance (when done right!)

[UPDATED MAY 2022] D3.js is a very powerful visualization tool, especially for specialized/custom needs... On the flip side, it's rather hard to use - with a steep learning curve. Even worse if one also wants interactivity ! But why is D3 so hard/clunky to use? And what can be done about it? Spoiler alert: Vue.js (or other modern front-end framework) to the rescue - if done right... All code in the examples is available in this GitHub repository . The Root of the Problem In a nutshell, what makes D3 awkward to use is that, for historical reasons, it tries to do too much : most painfully, it uses an old way to do direct DOM manipulation (i.e. restructuring the page layout) - an operation that nowadays is superbly handled in a far more friendly way by modern front-end frameworks, such as Vue.js Document Object Model ( DOM ) is a programming interface for web documents. In simple terms, it's the structure of the elements on a web page (text, images, etc.) Let ...

Julian's Polymath Explorations

Search This Blog