Can Silicon Photonics Solve the Bandwidth Bottleneck in AI Data Centers?

If you are interested in electronics, history, or simple amazing stories, you will enjoy this podcast interview with our Moore’s Lobby host, Daniel Bogdanoff. Mike Engelhardt has been developing simulation tools since 1975. He is best known for creating LTspice, which is believed to be the most widely distributed and used SPICE analog electronic circuit simulator in the world. Recently, he realized that he could take advantage of the massive improvements in both hardware and software to develop a radically improved circuit simulation engine; thus, QSPICE was born. Engelhardt explains how he created a better SPICE tool that could dramatically increase simulation speed while simultaneously improving functionality, accuracy, and reliability. As you will hear, Mike’s amazing career includes developing simulation tools for applications as diverse as scanning electron microscopes and locating underground petroleum deposits. Here are a few more teasers about what you will hear in this far-ranging discussion with Engelhardt: -The biggest early roadblock to simulation adoption in the EE world. -Why simulations are more important than hardware to an engineer. -The importance of time-domain simulation. -Insights into QSPICE for power simulation of wide bandgap semiconductors like GaN and SiC.

Ananth Avva is “a big believer that if you give human beings the right type of information and you contextualize it, they will make the right decision ultimately for the organization.” In his current role at Altium as the Senior Vice President and General Manager of Cloud Platform, he is leading efforts to improve and accelerate collaboration because “collaboration trumps everything else” as it drives speed to market, lower costs, and more efficient development. He has seen companies attempt to transition engineering teams to software-as-a-service (SaaS). In his experience, “that usually goes sideways pretty fast.” With Altium 365, Avva and his team are trying to seamlessly bring SaaS into engineering. Their goals are to enable what they call shifting left and shifting right. When shifting left, they aim to take decisions that were traditionally late in the product life cycle and move those requirements earlier in the design process. This can allow the procurement, operations, compliance, and manufacturing teams to provide vital context for better design decisions. Similarly, shifting right passes design information downstream more quickly to avoid costly delays or respins. If your organization is developing electronic products, you will benefit from this Moore’s Lobby podcast discussion hosted by Daniel Bogdanoff. Avva explains that companies consistently overestimate the cost of change and underestimate the benefits. They may also have the wrong “whys” behind why not to change. He goes on to explain that these can be “a bit of a phantom menace.” In his opinion, these are changes all companies will need to make to remain competitive in the market: Let's say you don't do it. The world is headed that way anyway, so what happens? Let's fast-forward five years, you're gonna have to do it.

Thomas Keller is the Director of Platforms and Core Technology at u-blox. He learned early in his career that technology development does not always progress in a perfectly straight line. During his PhD studies, he worked on Orthogonal Frequency-Division Multiplexing (OFDM) as a candidate technology for 3G cellular network deployment. OFDM lost that initial tech battle to Code-Division Multiple Access (CDMA) but later returned in LTE cellular systems. In this fascinating conversation, Keller discusses the challenges facing the industry today as low-power, low-bandwidth, low-cost IoT devices attempt to connect to a 5G network focused on high-power, high-bandwidth, low-latency applications. Interestingly, much of our current 5G networks are still relying upon an LTE backend. What happens when LTE goes away? In this Moore’s Lobby podcast with our host, Daniel Bogdanoff, Keller offers his thoughts on what should come next and guidance for developers who need to navigate the phase-out of LTE worldwide. This interesting topic deserves thoughtful consideration and impacts system designers and companies across a wide range of industries.

If you enjoy hearing passionate people talk about jobs they love and technology that fascinates them, you are in for a treat. Noah Gedrimas grew up with a love of vehicles—cars, tractors, construction equipment, and snowmobiles. In college, he built a one-fifth-scale car and then programmed it to self-park, which led him to a “newfound respect for mechatronics” and its ability to support really complex maneuvers. In this interview, Gedrimas reflects with enthusiasm on his time at Continental Automotive, working on everything from autonomous valet parking to electric shuttles and semi-trucks. Now, Gedrimas is applying that passion to ground-positioning radar technology at GPR. They are using ground-penetrating radar to capture unique, sub-surface data for positioning. Inclement weather compromised line of sight, roads with poor or no lane marking, weak GPS signals, and road terrain no longer affect the uptime and availability of autonomous navigation. You will want to join our host, Daniel Bogdanoff, as he chats with Gedrimas about: -Cadavers in cars. -The feature-rich information that lies below the ground. -The compute requirements when compared to cameras or lidar. -Early adopters of GPR technology. And much more that you will not want to miss.

After beginning his career with Hewlett-Packard, David Su met a friend for lunch to learn about a new startup called Atheros. He was so excited by the vision to develop world-changing products that Su “went for lunch and never left.” At Atheros, Su had a “front-row seat” developing wireless WAN technology that transitioned from novelty to necessity while Atheros grew from a startup to a billion-dollar behemoth. That little startup was eventually acquired by another company you may have heard of: Qualcomm. And what led Su to get involved with another startup? Well, lunch, of course. Su and his friends recognized that they could use their design experience to help fix a problem that was partially of their own making: battery consumption for wireless RF products. This was the genesis of Atmosic.  With his decades of design experience, Su admits that he stands on the shoulders of giants when creating new low-power products. The company has recently expanded its RF IC offerings from Bluetooth to Zigbee and Matter over Thread. When asked how an IC design team knows when they have become either too structured or too unstructured, Su thoughtfully answered that if you never create anything innovative or your chips don’t work, you have probably fallen into one of those two ditches.  You will want to join our host, Daniel Bogdanoff, as he discusses CMOS RF, energy harvesting, and the importance of teams with Su. They touch on many other interesting topics including:  -What he remembers most fondly from his days at Atheros. -A development failure for one technology that led to success for another. -The balance of circuits, system, and software design necessary to optimize product performance.  

Steve Klinger, Vice President of Product, joins the Moore’s Lobby podcast to discuss how LightMatter is using silicon photonics to improve speed and reduce power consumption in AI data centers. With two previous $1B+ startups under his belt, Klinger knows a thing or two about identifying successful technology solutions to current industry challenges. While compute performance continues to grow rapidly, interconnect has not been able to keep pace. In this episode, Klinger explains how LightMatter’s flagship product, Passage, creates a programmable optical fabric for the efficient interconnect of chiplets and other silicon ICs. Klinger explains that they are trying to solve the problem of efficiently accessing all of the bandwidth on one chip and sharing it with another chip. If they can improve the interconnect bandwidth density, it will allow performance scaling to continue increasing at the workload level. Klinger emphasizes, “There are data centers with hundreds of millions of dollars of GPUs sitting idle, waiting for the network topology or the interconnects to catch up.” So, join our host, Daniel Bogdanoff, in this deep dive into silicon photonics with Klinger. In this discussion, they address many fascinating topics, including: - What makes silicon photonics unique from traditional photonics? - The common traits shared by Klinger’s previous $1B+ startups. - The many job openings available at LightMatter.

While leading the system design of new head-word displays for fighter pilots, Tomide Adesanmi was most excited when he got a chance to innovate. However, like most electronics design engineers, he found that the majority of his time was spent searching distributor sites for components, drawing symbols, and working with spreadsheets.  Engineers can also relate to the anxiety he felt, worrying that he might have made a silly mistake by missing a tiny detail on “page 243 of the datasheet” for a microcontroller. So, he quit his job to try to tackle the mundane problems of design using “good software, good algorithms, and electronics.” A few months later, Circuit Mind was founded. Adesanmi and the Circuit Mind team aim to allow designers to quickly optimize designs with variable weighting on size, power, cost, parts availability, and more. Our Moore’s Lobby audience of electronics engineers will definitely want to listen in as our host, Daniel Bogdanoff, and Adesanmi chat about the possible future of circuit design. You will hear: -Why it is important that Circuit Mind uses deterministic algorithms as opposed to machine learning. -Defining the team and roles necessary to build these new tools. -What company makes the best datasheets? Contact Circuit Mind to schedule an appointment to learn more, get a demo, receive a quote, or even run a trial of your design on the ACE platform.   

Over the course of his fascinating career, Mark Himelstein has worked on several significant computing technologies at historic companies like MIPS and Sun Microsystems. He has also worked as a consultant in various roles that include architect, VP of engineering, and advisor. However, RISC-V may have a greater impact on the computing field and our world than any of those previous efforts.  Himelstein gives us an insiders view on the open standard process that is often “cooperation and competition, simultaneously.” “The thing that keeps us as a community is the effort in the software ecosystem. Nobody wants to really go off and go on their own. They just don't want to do it. It's just too costly. I don't care if you're the biggest company in the world or the tiniest.”  Listen in on this episode of the Moore’s Lobby podcast as Daniel Bogdanoff and Himelstein chat about a wide range of interesting topics that include: -How Himelstein encourages people to get involved with RISC-V by telling them to not just complain about something, but join in to help make it better. -The significance of vector operations for computing and how those advantages may soon be extended to matrix operations within the ISA. -Three major things RISC-V is currently working on for future release.  

Daniel Cooley started his career in RF chip design at Silicon Labs and now leads technology and product development at “the number one wireless supplier” for the Internet of Things. In this thoughtful interview, Cooley explains why adding wireless connectivity is only the first step to completely rethinking product designs and features. He noted that “the home run cases aren't where wireless is the feature; it's where wireless made that product better.” Cooley explains that Silicon Labs’ primary goal is to help companies get started with their preferred wireless protocol quickly and efficiently so they can focus on their applications instead of trying to debug the wireless links. Silicon Labs’ experience was forged on having “cut our teeth selling more than a billion wireless chips over many, many years.” Our Moore’s Lobby host, Daniel Bogdanoff, chats with Cooley about a range of fascinating topics that include: -Building wireless products for “metal benders.”  -The surprising economic benefits of adding wireless connectivity to shelf labels. -An explanation of the Matter protocol and why it is important. -The Silicon Labs partnership with Arduino. -Why the sky is falling…or at least the cloud is coming down. -Reflection on a major project and when Cooley “realized none of it was going to work.” -Why Cooley believes we will see a rethinking of the Internet infrastructure around non-human electronic devices.  

While Moore’s Law scaling has driven incredible advancements in computing, AI, and smartphones, many applications don’t need or benefit from the most advanced semiconductor nodes. From its inception, Pragmatic Semiconductor’s goal has been to take a…well, pragmatic…approach to develop an ultra low-cost, fast cycle time alternative to traditional silicon processing. Oh, and did we mention that the resulting chips and wafers are also flexible? You will definitely want to check out this Moore’s Lobby conversation between White and our host, Daniel Bogdanoff, as they dive into: -The technology and manufacturing of thin-film silicon -Europe’s largest-ever VC funding for a semiconductor company -The potential advantages of flexible silicon for building a more robust supply chain -White’s top priorities for improving the flexible silicon ecosystem