Electronic Design Automation (EDA), Equity, and Expansion in the Cloud
|Aired:||December 14, 2022|
Deirdre Hanford, Chief Security Officer at Synopsys, joins Nitin in episode four to discuss her work in the semiconductor industry, and where she sees it growing in the future. In a wide-ranging discussion around her thoughts on the industry as a whole, Hanford touches on Synopsys’ work with electronic design automation (EDA), IP building blocks, and software security. These areas of focus pave the way for a robust discussion around innovations in cloud infrastructure, which Hanford believes holds a key to workforce development, enabling greater equity among designers around the globe. An essential piece of the semiconductor ecosystem puzzle, Hanford’s insights as CSO of Synopsys are essential listening.
Hi, and welcome. My name is Nitin Shah. And welcome back to Circuit Talk Full Stack innovation. So this is a series of conversations with leaders in the semiconductor industry. And today I have the pleasure of introducing Deirdre Hanford from Synopsys. So Deirdre welcome,
Nitin, thank you so much. It's a pleasure to be here today.
Great to have you. So first, let me start with asking you to tell us a little bit about yourself, your role at Synposys.
You bet, I am happy to let you know my quick background and really start with today. And I know we'll explore other parts of my career later on. I am Synopsys chief security officer. And I'm not a typical information security person. I was asked to take on this role about three years ago, because as as Synopsys grew to a larger and larger technology company, we realized that we needed to think about our own security, but also thought about the impact of security on the whole electronics industry. And we have a lot to contribute. So I will I have responsibility for our in house information security. I'm also charged to look at the security of in microelectronics overall.
Okay, got it. Thanks. And we will explore more about your background, which is fascinating over the years that you've been at Synopsys. So, Debbie, why don't you tell us a little bit about what Synopsys does, I mean, my understanding is that in order to design integrated circuits, chips, micro electronics, you need very sophisticated tools, which will allow you to do design, verification, and eventually have those designs fabricated in in either in the foundry, or a fabrication facility. But you play a very key role in the industry. So please tell us more about Synopsys, its products, its intention,
you bet. And I think you've got it, you've got it pretty well covered. But we are we talk about really silicon to software, because we cover in a very, very broad spectrum. And attend we have three product lines at Synopsys one is electronic design automation software. So that's a whole spectrum of capability. But anyone designing a chip, or designing a new technology process, or going into manufacturing needs to use design automation, or manufacturing automation to get that job done. So we you know, probably about least two thirds of our business is in this EDA software realm. So that involves like designers at AMD that are writing RTL and thinking about the next cool processor that that they're building, or an engineer at Qualcomm that's trying to verify a Snapdragon processor, before they go to market. These are just examples. I'm not implying that anybody is or isn't a user of our capability, but want to give you a sense of the scope of the EDA realm. The second domain that we have is a very large franchise for us. And that's IP building blocks. So I'm looking at our monitor right now. And you know, there's the there's a HDMI port on that monitor. And so somewhere, there's a TV chip on that monitor that needs to speak HDMI, well on the tin, if you're designing a chip that needs to speak HDMI, that's a requirement, that's not necessarily where you're going to differentiate. And so we sell a host of IP building blocks for chip designers, because they want to be able to work their magic, which is the guts of the chip, but they need to get on and off chip. So we are the leader in interface IP, and we're actually number two in IP behind arm who's obviously number one in the space. So leader in EDA, number two in IP building blocks. And then actually we have a business that's about seven years old. Now. That's all about software security. And you know, this is a different space for us, because obviously people doing chip design are also designing a lot of software. But people like Starbucks are designing software, people like Walmart are designing software, people like Costco are designing software. And they all need to make sure that that software is high quality and secure. And so there's a whole domain of products there as well. But I think today what what's most relevant to this conversation is really EDA and our IP business.
That's really helpful. So focusing on the EDA tools for the moment. So your tools not only work with on chips, which are silicon based, which is a vast majority of most of the production across the world, but you also have products which deal with wideband back bandgap semiconductors and even the emerging quantum devices. Is that correct?
EDA technology is really geared for you know, the sea moss realm. But you know, if you're doing custom design, you can be designing in all kinds of various domains. So for instance, one of our programs that we just wrapped up is with AI I ARPA, where we've been working on a whole EDA flow for superconducting electronics. And we were able to take that soup to nuts flow that we've done for kind of traditional devices and apply it to a whole new domain. So I think there's really, and then the third common I'd have there besides, you know, kind of traditional and more novel devices is, when people are designing new processes, they need to do a lot of device modeling. And so there's a whole lot of exotic in that realm. And people are using our technology, CAD capability or this technique called Design Technology, co optimization, to think about what new devices they should be bringing online, and what new technologies could really advance the state of the art of semiconductor.
That's really helpful. So another area, which I believe is, is very prominent right now, in terms of evolution of designs, particularly in silicon, but maybe even in some of these other technologies, is to do with hybrid, sorry, heterogeneous integration, meaning not only designing the chips themselves, but the packaging and the assembly around that. So is that also an area that you're focused on in terms of heterogeneous integration and the technologies and therefore the design tools around that?
Absolutely. And attend I got to share with you two weeks ago, I was at the TSMC oyp event here in San Jose. And I, Jim Keller, who is you know, like the guy designing chips on this earth, he's done some of the most amazing chips throughout his career, gave a keynote at oyp. And, you know, he designed massive chips at Apple massive chips at Tesla, and now he's at 10, store it, I fell off my chair, because he said, The world is going to heterogeneous integration, I thought, oh, my gosh, you heard it right here at the TSMC event. And that's pretty profound. If folks like Jim, who've made their career in big bad chips, are thinking about the leverage that can get in heterogeneous integration. So I think we're at a sea change right now. And I know there's a lot of government investment, we'll get to chips later. And I know, DARPA has got initiatives in this space. But absolutely, this is this is part of the Synopsys, portfolio. And part of our roadmap as well, as an industry we have a lot of challenges to cover, when it comes to heterogeneous integration. You know, folks talk about, you know, this, that thermal envelope, once you put all those devices together, folks talk about, you know, partitioning and planning. And, you know, how do you decide, you know, what, how to break up the design to be most effective. So, this is an area where we have offerings in the market today, and we absolutely will be collaborating as researchers to meet the needs of this evolving market.
So this is quite transformative, because one thinks of chip design in terms of electrical circuitry and layout. But now you're in this realm of, as you're alluding to, multi physics, you're dealing with thermal, you're dealing with layout, architectural building blocks, which have to be split up interfaces, between triplets, as they call them, and so on. So does this represent new challenges for your company in your growth areas?
I think so I think, you know, there's going to be a lot of new challenges that need to get addressed new business models, I think, as well. You know, I think companies are deciding, you know, how big chip companies are thinking about how they want to get into this chiplet space, you know, who's going to own if there's a problem? Right, you know, we see the foundries getting into this Chipless space and deciding, you know, are they going to create kind of the standards where all these pieces come together, for instance, TSMC, and others have announced kind of their approach to this, this challenge. And, you know, I think that we're going to see the aggregation of these devices also becoming, you know, a whole new a whole new offering, so will the foundries lead there, where the packaging houses lead? There are, well, new entrants lead there. I don't I don't think we know yet where this is going. But I think it's going to be an exciting time in our space.
Another aspect of your industry, and in the software industry, in general, is the cloud. So my impression is that many industries, especially in the enterprise business, have moved a lot of the software and capabilities into the cloud, and created new business models, better access to tools and capabilities, and have transformed industries as a result. How do you see the EDA industry not only at Synopsys, but in your partners and so on, moving in that direction.
So we've been we've been, there's there's numerous terms in cloud, but there's this concept of lift and shift where you kind of take your current workloads and then you move it to cloud. And I would say all the ADA vendors have demonstrated that their tools work in the cloud and that, you know, customers if they want to, like for instance, burst to the cloud, you know, that you can, you can provide that kind of baseline capability. That's really table stakes. And so a great example of someone wanting to burst into the cloud. I would be if you're trying to do library characterization of a new library, and you want to run a gazillion corners, and you want need to run tons of simulations, and you just want to have, you know, as much CPU capability available as possible to reduce that very compute intensive job. Another great example is if you're doing, you know, sign off of a big chip, suddenly, you're doing a gazillion timing, sign off runs. And once again, you want to try and you know, they always say, nine women having a baby in a month, you know, you want to try and get as much work done as possible as you can, and leverage the cloud in that regard. So this is really, I think, a requirement on the EDA industry. But we've got a challenge, because a lot of companies have very, very, very sophisticated in house or, you know, kind of managed cloud infrastructure themselves, right. And so I've got, you know, one of my one of my customers, they've got CPUs that are kind of bespoke servers for verification workloads, right, they've tuned the heck out of that they've optimized the cost out of it, and that IT department feels very proud of that accomplishment. So you know, when you've got a lot of investment in the very complex workloads around chip design, you know, it's hard to displace that and say, hey, just move all that to the cloud. But what we're seeing is a really interesting move is that small companies are born on the cloud, they don't want to invest in all that IT infrastructure, they don't want to have, you know, a large IT departments that that are charged to manage all that infrastructure. So we see big companies looking to burst in the cloud, we see small companies that want to be born on the cloud. And you know, now we're really getting to the point as an industry where we're looking to truly optimize our workloads for the cloud. And that is a journey that we're on, and I know others in the industry, Iran, but the big breakthrough in in, you know, I had been on Cloud panels for years and always said, Well, when are you going to offer cloud, you know, EDA tools by the minute, when can I get it by the drink, I don't necessarily want to own it for multiple years at a time. And so we've recently introduced a SaaS model, and you know, an ability for folks to access certain parts of our portfolio on, you know, cloud based units of time. And I think that's going to be another step in this transformation of project teams moving, moving to the cloud. Well, we hear a lot from our cloud partners, and I believe this is, you know, we all have amazing compute infrastructure in our companies. But that's a moment in time. And the cloud is constantly refreshing their hardware, and constantly updating, you know, their infrastructure, so that you can always get the best and the greatest if you're willing to kind of unleash yourself from your on prem infrastructure and move to the cloud. So we're on a multi year journey. I think there's various reasons why people are moving aggressively, or maybe more slowly. But it's a fact that people are moving to the cloud.
That's great. So maybe if I can summarize what I heard and see if you want to comment on it, which is moving to the cloud performance, compute workloads, that sort of, as you said, table stakes, one can do that. But it also I think, improves accessibility, and then allows you to transform to different business models. So again, you're improving access was somebody who does not want to invest in huge amount of infrastructure, but still can access your tools. So the business model innovation, and presumably, there's another benefit, which is that if you're operating the cloud data, sharing the ability to store data to be able to analyze it, apply, some of the advanced things that you're working on in machine learning, and so on, can also be late. Put on top of that, and all of that I think all of those things lead to enormous improvements in productivity, which is really the goal for any new design or architecture, which you need to put into production. So is that roughly right, your thoughts?
Yeah, I think very, very much. So let me make a couple of comments there. One is, we're involved in a DoD program called ramp actually, we're performer under team Microsoft. And that that program is all about creating a secure design environment, and then leveraging that, for quantifiable assurance, right gathering design artifacts, it so that's, that's a whole nother discussion. But what's what's really interesting about the these types of programs like the ramp program is you do always have multiple parties that want to gain access to a design that have a particular reason, you know, maybe you've got a third party that's designing a part of a chip. Maybe you've got another group that's performing verification, maybe you've got a support team coming in to debug an issue. And can you create this collaborative environment on the cloud where all parties can go in really a supervised way? And I think that this is unleashing, you know, really exciting possibilities because chip design is a team sport and not all those team members are necessarily in the same logo, and I think here you have a great a great point to be made there. Another thing I want to mention in the cloud is, as we look at look toward workforce, workforce development challenges in this country, and I know I'm jumping to another topic, but we really want to, we really want to focus on equity and inclusion, for you know, all all universities to be able to participate in workforce development programs, you know, all the way from community colleges to, you know, top tier engineering institutions. When you have the cloud, you can imagine, you know, standing infrastructure, so that all kinds of parties can just easily log in through a thin client, and be able to do, you know, remarkable work, you know, smaller universities may not have the scale of the compute infrastructure, and they can certainly get access to EDA tools, but maybe they can't get access to the foundry, collateral, etc. So if you can stand things like that up in the cloud, imagine the equity and inclusion that you can deliver to designers around the country and around the globe. I think that's really transformative.
Now, that's fantastic. I'm going to ask you about workforce development in a couple of minutes. But wanted to get your perspective on the chips act. There's a lot of discussion, a lot of potential. And I would say there's a lot of discussion about fabrication facilities and technologies at the physical level. But your thoughts more on from the EDA perspective, because as a new technology is introduced, and there are innovations, breakthrough challenges within the scope of the chipset, surely the EDA tools and capabilities have to be in sync with that. So that then those technologies can be designed by this vast community that you're building.
Wow, that's, that's a long, long involved question. But let's let's go break it down one by one, look at the chips act. And chips act is really a proxy for similar acts around the globe. But let's focus on the chipset for a moment. And so you know, of the of the of some odd $52 billion of funding, the vast majority of it, or 39 billion is going to be geared toward building new fab capabilities. Now, every time you build a new fab that's geared to be a foundry, you need to make sure that if you're going to stand up a new process technology, you need to have a whole set of EDA tools that are that work with that particular technology. And you need to have those IP building blocks, because if you know if I'm Nitin Shah, the new startup CEO, and I want to design a chip for a particular market segment, I'm interested in that new fab that was built in Texas, or Arkansas, or Ohio, or what have you. But if it doesn't have the EDA tools, qualified for that node. And if it doesn't have the IP building blocks that I'm looking for, well, I'm going to look elsewhere. So if you think about all those fabs that are going to get built in the country, any one of them that are going to stand up and be a foundry, which will take you know designs from third parties, they're going to need EDA enablement, they're going to need IP enablement to be successful. So we're going to play is an EDA industry and as an IP industry a strong role in the manufacturing component of chips. A really exciting area, of course, is the research and development side of it. And I'm super, super honored to have been appointed to the chips industry advisory board. And our charge is to really advise Secretary rimando on the r&d aspect of chips and make sure that a were the you know, the all of the research and development initiatives are addressing the needs of you know, what, whatever gaps we have in our in, in r&d for microelectronics. So where does EDA fit in there? First of all, you're gonna have a host of researchers that are either researching new materials or thinking about new device structures, or perhaps researching new you know, Chip topologies they're going to need access to clean rooms, of course, but they're also going to need access to design tools and prototype and prototyping facilities. So we're, I view and in that way that you know, we need to provide them their their miners out there and we need to give them shovels and pickaxes, right, which is really EDA tools in order to get their job done. So I think it's really in and that's going to be enabling that amazing work. But what's interesting is every time there's an advancement in in semiconductor, the whole EDA industry needs to respond and the example I like to use because it's fairly current is this notion of backside power delivery. And you know, backside power delivery has got an amazing potential for you know, for addressing thermal issues number one for providing more compaction on the on the top side of the chip. But if you imagine how the topology of backside power delivery, there is a lot of work that's going to need to be done by the EDA tools to make that work. And so we need to provide tooling for the researchers. But a lot of the research that actually the EDA industry is going to need to participate in directly, because our whole tool suite will need to evolve and match the needs that come out of this great research. So I think that, you know, the, you know, we're going to be foundational, in both the grant side as well as the r&d side as an EDA industry.
That's great. And I think there are a lot of sort of very profound things that you're touching upon, and particularly as you said, as technology evolves to the vaccinated power technologies, and then beyond that, whether it's heterogeneous integration, or whatever your ability as an industry to be, not only on top of that, but ahead of that is exciting. And by the way, congratulations on being appointed to the industry advisory committee, I think it's, it's very, very important contribution to the industry overall. So
I won't, you know, I can't get into any details, obviously, but you know, the Secretary, in a kickoff meeting said, do not show horses, your work horses, we need you all to contribute. And I think that's really awesome that you know, this committee is, is really a sincere effort by the government to make sure that industry is informing the proper spending of this very large industrial policy called chips.
That's fantastic. Thank you. So if I may disagree, I want to do pivot to something you touched upon earlier, which is workforce development. Actually, I'd like to start a little bit with your own journey. You've been with Synopsys, right from your very start of your career. So can you just share very openly, I mean, your perspectives your journey overall, and really sort of speaking to hope folks who have a chance to see this, this discussion, which is to really learn from what you've experienced in your career, maybe some challenges, but also the achievements which are very evident. So
that's like a 22 point question. Let me let me say, Yeah, I'll make it I'll try to make it as relevant as possible. So I did my undergrad and engineering, and then decided rather, came out kind of when the market was a little soft. So we decided rather than go straight to industry that we get a master's degree. So I went off to UC Berkeley and had the the honor of of getting my masters and under Richard Newton, for many of you on this cast may know him he was he passed away. But when he did, he was Dean of Engineering at UC Berkeley, but one of the leaders in the Berkeley CAD group, so he was my advisor. And after my very first job in the industry, he called me and said, there's this company. They're a startup, but they're going to change the world. And I will say, Yeah, sure, I've heard this story before. But I listened. And I had the opportunity to meet RTGS, who was the CEO at the time of Synopsys, and still is almost 36 years later, and got really excited about the potential of this new concept called logic synthesis. And so joined on back in 1987, and had been with the company ever since now people say, Well, that sounds really boring. But I have to tell you, I've done probably a 1213 1415 very different roles at the company. And and I think, you know, we all in our career, want to learn new things and have new experiences in the valleys. Oftentimes, that manifests itself as people hopping from company to company. But I'm here to tell many people that you can also build a really exciting career in a single company. And certainly, I've had a lot of opportunities and feel like I've earned more earn my keep along the way. But, you know, a lot of my experiences were, because I was really willing, at many, many stages of my career to take lateral moves. So I started in, in the very early days, art had a prototype, and it was going to take a year before he had a product. So he said, Why don't you go take this out and work with some early customers? So we shipped me off to Sun Microsystems, and I work with folks like Andy Bechtolsheim and Chris mela kowski, before he went off to found Nvidia, and you know, my job was just and Jon Rubinstein when he was running it another startup called Start. And, you know, a lot of folks that became icons in the industry, we're trying to do chip design the old fashioned way. And my job is to get in there and kind of get them excited about a new potential way of doing things. And then but but over the course of my career, I took on many different roles, some of which I never imagined I was even qualified for. So I think my my advice to folks throughout their careers is be willing to take risks and don't fall into the trap of oh my gosh, I'm not qualified for that. And, you know, I'll give you an example. I will was asked to run corporate marketing at Synopsys and that had a very wide remit. And I sat there and thought, I don't know, Jeepers, I don't know an ad agency from a PR agency. Why on earth do they want me to run corporate marketing, I can barely run PowerPoint. And as it turned out, they wanted someone who understood the technology that could help position that technology to the broader market, because we were scaling as a company. And I thought that that was a great opportunity to bring my knowledge to a new team, and then be really willing and open to learn from that team throughout that period. So but much of my career has been focused on supporting our customers around the globe, I've had the privilege of leading very large global organizations that technically support chip design all over the universe. And that's caused me to have to, you know, visit many wonderful places and learn about many complex chip designs around the globe. And as I mentioned previously, you know, art and cheap food at the time, cheap food has since left the company, but asked me to take on this cheap security role, because they saw that we needed to scale and once again, it was important to have a business person understand the implications of security, on on a company of our scale. So I've taken a million lateral moves. I think a lot of people obsessed about moving up and getting their boss's job or getting that next fancy title. I actually am realize that it's those lateral moves, that I believe strongly position you for more exciting roles down the lane. And you know, having moved into the field, I spent a year in sales, because I really wanted to understand the business context. And you know, having grown up on the technology side of the house, I've spent time as a general manager, and you know, now a lot of my time is actually spent in Washington DC, is a whole other learning experience. So be willing to take lateral moves, be very willing to have an open mind and meet new parties and be honest about here's what I know. And here's what I'd like to learn. And can we work together to that to the end or the benefit of our customers? So that's hopefully a decent enough Journey to 10 No, no.
degree, I think the my quick two takeaways from what you said, is this theme of learning, which is, which is extraordinary, you've done so many different things. But at each stage, it looks like you're willing to take a risk, whether it was a lateral, whether it was a promotion, but really take a risk, have confidence in yourself, obviously, but be willing to learn just new things. And I think that's, that's very remarkable, very commendable. But it's something that I think I'm happy that you've been able to share with that. And the other component that I heard when you were talking was customers. And in some ways, I think many companies don't give people an opportunity to deal with and work directly with customers. And it sounds like that early customer interface that you had, it has been a driver as well. Right, which is that you get to know what the problems are, you get to know, even the satisfaction of satisfying a customer, which is also good. So is that an accurate read of the things that the learning the risks, and particularly customers that came up?
Absolutely. I think having having that customer perspective is so important. You know, in our industry in particular, you know, if we're not listening to our foundry partners, as they're pushing the state of the art, if we're not listening to our key customers, when they're trying to do more and more every year and Dr. Moore's law or the more than more access, you know, we're going to be out of business and, you know, think customer facing is always a joy I saw once one of my colleagues today is customer just taped out a giant chip and he's been so involved in that journey through the whole mission. I'm like, you know, it he feels as involved in that chip tape out as as, as our customer does. It you know, that's the that's the euphoria of our job. But you know, there's hard times as well. And I always view of your customers, if even if they're angry at you, and they're giving you a lot of you know, feedback, that means they're still vested and you're still working together, and you're getting the job done. So the highs and the lows are what makes it relevant and and really leads you to more success down the line.
Well, on that note, Deirdre, thank you so much for making time for us today. It was a real pleasure talking to you. Again, Deirdre Hanford from Synopsys. Thank you very much for joining us on Circuit Talk.
You bet. Thank you Nitin. Thank you