Last year, almost to the day, I had been working for Cadence for barely four months, and I used my first CDNLive as a platform to introduce myself in my very first blog post . I now have a year of writing and editing about EDA under my belt, including a mess of articles and white papers and about sixty Cadence on the Beat posts. I have written about everything from System Design Enablement , to the arts (particularly music) and AI in Is it Amper, or is it Music? , to the perils of Weapons of Math Destruction . I have written about storytelling in Science Fiction and Technology Reality , thought about Why Home Renovation Should Be Easy and even wrote all about Life, the Universe, and Everything . I went to DAC , the Embedded Vision Summit , #CES2018 in Las Vegas , and other conferences and summits; I attended several dinners, keynotes, and talks. In December, I even exercised my creative writing chops with writing a nine-installment A Cadence Carol (and I have no idea what I will do next year to top that!). All in all, it’s been a fun year, and I thank my co-workers and friends for being patient with my never-ending questions and probing for ideas. CDNLive 2018 So here we are at the beginning of the CDNLive year again, the first in seven world-wide CDNLives, and the keynote was an inspiring combination of words by Lip-Bu Tan, our CEO; Rodrigo Liang, co-founder and CEO of SambaNova Systems; Gopal Raghavan, CEO of Eta Compute; and Tom Beckley, VP of Cadence’s CPG group. Now, I won’t go into too much detail about the keynote; I know that Paul McLellan has written about it, and about the news about Virtuoso 2018 , which Lip-Bu announced in his talk. What I will say, though, is that this year’s keynote set my imagination spinning in a way that I hadn’t expected it to. Lip-Bu started talking about enabling a data-driven economy, reiterating five key waves in our industry: mobile, automotive, “machinelearningdeeplearning”, edge computing, and data centers. This wasn’t a surprise to me. What was new, however, was later in the keynote, where he talked about the emerging disruptive technologies as potential drivers of future growth. These technologies are shown in his slide: If you’re at all like me, you will have heard just a little bit about this stuff in the news or overheard some talking about some of these. But to have them pointed out by our CEO as drivers of future growth, I figured it was time to delve into what these things mean. Today I’ll write at a high level about what these things are; expect more in-depth posts about each of these in coming weeks! Silicon Photonics Photonics is transmitting data through fiber optic cables, as opposed to copper wiring, which is faster, lighter, and cheaper than going the copper route. As Paul McLellan said in a blog post from last June : Optical transmission is very attractive since it can support high speeds, the transmission medium is not heavy ("the transmission medium is light" is true in two different ways!) like copper wiring harnesses, and it is largely immune to external interference from electromagnetic and RF sources. Silicon photonics means implementing photonic circuits on regular CMOS chips, so the electronics are integrated on the same die as the photonic devices, with both electrical and optical input and output ports, resulting in amazingly fast computes, allowing Moore’s Law to continue on its merry way. For a three-minute Ted talk on the subject, check out Mario Paniccia talking about the possibilities. https://youtu.be/Ou9AGiMvdkQ Carbon Nanotubes Okay, so nano means very very small, and tube means… tube. What’s the big deal? Teeny tiny carbon structures in a tube shape? Here’s the thing—carbon nanotubes have some pretty nifty and unique properties. In theory, nanotubes can carry an electric current density of 4 × 109 A/cm 2 , which is more than 1,000 times greater than those of metals such as copper, where for copper interconnects current densities are limited by electromigration. Carbon nanotubes (CNTs) are being explored as interconnects, conductivity enhancing components in composite materials and many groups are attempting to commercialize highly conducting electrical wire assembled from individual carbon nanotubes. The electronic properties of individual CNT fibers ( i.e. bundles of individual CNTs) have been measured to have a resistivity one order of magnitude higher than metallic conductors at 300K. Large quantities of pure CNTs can also be made into a fiber, and individual fibers can be turned into a yarn ( knitting, anyone ?). Apart from its strength and flexibility, the main advantage is making an electrically conducting yarn (not just for the most shocking sweater at the Christmas party!). By further optimizing the CNTs and CNT fibers, CNT fibers with improved electrical properties could be developed. Carbon nanotubes also have some interesting optical properties, in that they have useful absorption, photoluminescence (fluorescence), and Raman spectroscopy properties—which, I expect, could be applied to silicon photonics above. A nice (though simplistic) overview is here, from Nova. www.youtube.com/watch Neuromorphic Computing This is not your run-of-the-mill AI. According to the Human Brain Project , neuromorphic computing implements: … aspects of biological neural networks as analog or digital copies on electronic circuits. The goal of this approach is twofold: Offering a tool for neuroscience to understand the dynamic processes of learning and development in the brain, and applying brain inspiration to generic cognitive computing. Key advantages of neuromorphic computing compared to traditional approaches are energy efficiency, execution speed, robustness against local failures and the ability to learn. I think the most relevant bit is that at the end: improving energy efficiency, execution speed, and robustness against local failures. Gopal Raghavan, CEO of Eta Compute, talked a bit about neuromorphic computing in his portion of the CDNLive keynote, in the context of moving all processing to become embedded; that is, instead of being connected to the Big Datacenter in the Sky, the processing required by your phone can be done on your phone, and not involve a cloud-based intermediary. Think about your phone—the biggest energy drain are those applications that are connected to the cloud, where the energy is spent finding a Wi-Fi connection or connecting to some 3G or 4G (or soon, 5G) network. If we can get the energy usage way down, with the addition of improving processing speeds, we are getting closer to the power usage used by our own brains. Gopal compared the neurological systems of a honeybee versus a datacenter, in the following slide. We clearly have a long way to go. Quantum Computing We’re getting to the magic realm with this one. So first, let me take a stab at explaining quantum entanglement. As I understand it, an entangled system (that is, an atom) is defined as one whose quantum state cannot be factored as a product of states of its local constituents; they are individual particles but are also part of an inseparable whole. In entanglement, one constituent cannot be fully described without considering the other(s). In other words, say you have an atom that decays into a pair of other new atoms. These are entangled particles. You can then separate these two particles—by a test tube or by an ocean or by a light year—and if you measure a characteristic of one of these particles (for example, spin), you know that the other particle complements the measurement. If you manipulate a particle in one location, the entangled particle will react in the exact same way. (It would be as if you had a set of twins and one were to get a haircut, the other’s hair would also get shorter!) Okay, so remember that a quantum state is a probability distribution for the value of each variable that can be measured in a particle. Quantum superposition is any two or more quantum states can be added together (“superposed”) and the result will be another valid quantum state; conversely, every quantum state can be represented as two or more other distinct states. So— quantum computing . It is different than computing using binary digital electronic computers based on transistors. Digital computing requires that the data be encoded into binary digits (either 0 or 1). Quantum computation uses quantum bits, or qubits , which is a unit of quantum information, and it can include superpositions of states. The Bloch sphere is a representation of a qubit , the fundamental building block of quantum computers. A quantum computer maintains a sequence of qubits instead of 1s and 0s. A single qubit can represent a one, a zero, or any quantum superposition of two qubit states; a pair of qubits can be in any quantum superposition of four states, and three qubits in any superposition of eight states. In general, a quantum computer with n qubits can be in an arbitrary superposition of up to 2 n different states simultaneously. (In comparison, a digital computer can only be in one of these 2 n states at any one time.) All this to say, instead of dealing with only 1s and 0s, a quantum computing system can process… infinity? Confused yet? Me too. This is the best I can do using the resources I have at hand. Stay tuned for more on this in blogs to come. Blockchain We hear about blockchain technology mostly in the context of bitcoin. As I understand it, blockchain is a security protocol, originally developed to trade in bitcoin, consisting of a continuously growing list of records (“blocks”), which are linked and secured using cryptography. It is managed by a peer-to-peer network that adheres to the same protocol. Once a block is recorded, the data cannot be altered without altering all subsequent blocks, which requires agreement of all members of the network. Everything I have read about blockchain technology is related to security. The article “ The Truth About Blockchain ” in the Harvard Business Review put the implications of blockchain thusly: With blockchain, we can imagine a world in which contracts are embedded in digital code and stored in transparent, shared databases, where they are protected from deletion, tampering, and revision. In this world every agreement, every process, every task, and every payment would have a digital record and signature that could be identified, validated, stored, and shared. Intermediaries like lawyers, brokers, and bankers might no longer be necessary. Individuals, organizations, machines, and algorithms would freely transact and interact with one another with little friction. This is the immense potential of blockchain. Another good article about the implications of blockchain was published in The Economist, called, “ The Great Chain of Being Sure About Things ”. Very readable and interesting article. I’m curious to know why exactly Lip-Bu listed this technology as a major source of disruption and growth driver for the EDA industry—unless he meant it as a disruptor to all technology in the coming years. I think I’m missing a piece of the puzzle. As Paul McLellan pointed out, Looking further out, Lip-Bu has an advantage over most of us. … no semiconductor startup is created without it crossing his desk, so just the "deal flow" gives a good perspective on where innovation is going. I trust Lip-Bu, and am excited to see which of these new technology disruptors will become part of our technological world—and how. What do you think? Is he right in his predictions? I’m curious to know what you think. Drop me a line at mcollier@cadence.com . —Meera![]()
↧