The automotive industry is undergoing tremendous and rapid change on many fronts. OEMs are exploring new functions and features to add to their vehicles, including chiplets, electrification, autonomous capabilities, as well as new vehicle architectures that will determine how vehicles are designed from the ground up. All of this depends on the relationships between all the ecosystem actors involved. What follows are excerpts from a panel held recently at the AutoSens Conference in Detroit, which included David Fritz, Vice President, Hybrid and Virtual Systems, Siemens EDA; Kamal Khouri, Vice President and General Manager Automotive at GlobalFoundries; and Frank Schirrmeister, Vice President of Solutions and Business Development at Arteris IP.
SE: The impact of chiplets is being discussed regularly these days. What does this mean for relationships between ecosystem actors today?
Fritz: Think of chiplets as an individual die without the packaging and the ability to connect multiple chiplets together via a new communication protocol created by Intel called UCIe, which is very similar to PCIe. The idea is that instead of building a huge monolithic chip, you could go to Qualcomm or Samsung or NXP and get individual chiplets and then put them together. This is how you would create a much more domain-optimized solution. Imec created a chiplet consortium – which we’ve joined with many of the companies we’ve been working with for some time – whereby the question is what are chiplets really supposed to do? What is the functionality? How do you put realistic workloads on them? How do you do power and performance analysis and all that kind of stuff? We’ve been working with a lot of the companies that are in the imec consortium to help them decide these things, and imec is doing a great job of pulling it all together and trying to standardize this and get it the focus it really deserves.
Khouri: Chiplets are testament to the fact that Moore’s Law does not work in all applications. And as we’ve seen, you need specialized technology for the diversity of solutions that we’re trying to address. The single-digit nanometer, for example, might be great for AI, great for high-performance computing, but not so great for driving a pump or motor for the seat controller in the car. And it’s not so good for your radar. But how you want to bring all of the compute and sensing components as close as possible, optimize that solution, optimize power consumption, that’s a big testament to the types of technologies that we need to address and the diversity of problems that we’re trying to solve.
harness master: There are many challenges. If you look at the ecosystem as a whole, it’s a huge opportunity. I fully agree with all of this. It makes life very communicative from OEM to Tier 1s through Tier 2s as well as on the IP side, then on the tooling enablement side, until you get the technology you need to make it work. To take an example, the OEM will decide to use certain chips together. They need to talk to each other. UCIe is the beginning of the equation. That’s the streaming interface and it’s absolutely essential. On top of that, you have a whole protocol stack. The OEM will decide that certain components need to communicate and will ask which protocol to use. For IP providers that support most protocols, it’s great. There are interfaces like CHI from part of AMBA and there are other slightly competing bits. There is the UCIe. There’s also Bunch of Wires and so on. There are other mechanisms available to potentially connect, so the OEM will make that decision. So everyone in the design chain needs to align and make sure these things fit together. Then the tools can analyze everything, how it all fits together, and the IP vendors will provide the underlying building blocks, which will be implemented by the foundry.
Fritz: All of this is absolutely true, but what does it really mean? We’re all engineers and ex-engineers, so it’s really easy to dive into the details. Let’s back out of that for a minute. I made a statement about six years ago that in about five years, virtually every OEM on the planet that decides to succeed beyond this industry transition will at least be thinking about building their own SoCs. It wasn’t very well received at the time because I was wrong most of them were looking at it five years ago. They went further, but the learning was that the amount of expertise, the skills, the knowledge, the cost of making a very large monolithic SoC makes it extremely difficult to make. This means that if you have the ability to take these dies, these chiplets that already exist, and all you have to do is package them differently and have your own pinout, then you can have something that fits your needs and you you don’t have to invest $100 million dollars – or more like half a billion dollars – to make some of these chips. Now the question is who will produce the chiplets. We’re seeing companies like Qualcomm, Samsung and Intel, and even Tier 2 NXPs, Infineons – they’re thinking about it. IP providers [Arm, Arteris IP, Synopsys, and others] are also part of it. If you have the ability to boil that down and say, ‘This is specifically what we know we need and we have KPIs from our customers,’ I can create that exact chiplet. I just have to validate that chiplet. If you have multiple companies building these chiplets and all you have to do as an OEM or Tier 1 is put them together and decide what the pinout is, then you have an ideal solution – or as close to ideal as possible. can get without investing hundreds of billions of dollars to make it work.
SE: A key part of that is manufacturing. What is the role of foundry in chiplets and within the ecosystem to bring everyone together?
Khouri: The individual chiplets will be manufactured on some process node or some process technology, and that will be an important factor. So, let’s start looking at the packaging technology and the chiplet design packaging concept. What do you need to do at the silicon level to facilitate validation and bug fixing? Let’s not pretend that the problem is solved. This is a great concept. There’s a lot of work to be done, from the foundry to the EDA tools, how do we validate this, how do we debug it in the field. There’s a lot of work to be done, but I think we can all agree that the concept is a good one.
Fritz: Work has just started.
harness master: It is also the only way out. We better figure this out because, as Kamal pointed out, Moore’s Law forces us to do this, because it’s a ‘more than Moore’ kind of thing. There are two reasons for chiplets. When you have existing chiplets that you now want to integrate and they need to work in tandem with existing chips, let’s not forget the other side of the equation, where the crosshair boundary forces you to not just be more complex. If they could, they probably would. There are still SoCs out there, but let’s not make the mistake that it’s the only way out other than Moore’s Law. Manufacturing is extremely important. Integration and tooling are important. What’s also fascinating are the timelines. So, there is a power dynamic that must be observed. At a recent McKinsey/GSA event, they were talking about the emergence of Tier 0.5, which is a company where the OEM and Tier 1 are discussing and trying to figure it out, but then it all comes down to the tools, the IP, and the foundry underneath for it to be implemented. So there’s a whole new set of communication challenges that we need to work through, but it’s the only way out. We better find out.
SE: This is also finding and generating value. What kind of advice can you give about the value of relationships and value creation?
Khouri: This discussion is stirring memories of a few years ago when we were sitting around debating the next generation of vehicle architecture. People would get onstage and ask, “Is it centered? Is it core computing? Is it distributed? Is it zonal? Is it domain control? First of all, I’d like to ask that we don’t try to have a solution that’s looking for a problem. Specifically, what are we trying to solve with the chiplet approach that we couldn’t solve any other way? I think this is important. It’s cool technology, don’t get me wrong. We are all engineers and we love to solve things. But I recommend caution there. And that brings us back to the question about this participation in all domains, even OEMs asking what is the problem that we need to solve for you that we cannot solve efficiently with what we have proven today. Understand the problem statement from the OEM to whoever is Tier 1, to the IP providers, to the EDA tools. This is critical and is an example of why open channels of communication need to happen.
Fritz: A slightly different approach might be helpful. I like to say that the entire automotive ecosystem is being turned inside out. By that I mean it used to be that Tier 2s would talk to their customers and they would design some silicon, and that would go to Tier 1. Tier 1 would do some integration, talk to OEMs, and then sell to OEMs. The OEM would say, ‘That’s more or less what I want. This is the closest. So this is the one I’m going to buy.’ OEMs have learned that the process is not working when you’re getting into big core computing, when you’re getting into high complexity computing. You’re throwing piles of insight, you’re throwing all kinds of artificial intelligence at this because vendors aren’t necessarily understanding the depth of what you really need to accomplish. So by upside down I mean why do OEMs really care about the size of their caches? Or how many CPU cores? Why don’t they just say, ‘Here’s my KPI’, because they’ve been doing this for the last 50 years and it’s not working anymore. Now Tier 1s are feeling pressured because there are outside companies coming in and talking directly to their customers, and they don’t have control over that conversation. Tier 1’s across the planet are wondering how they stay relevant. They stay relevant by talking to these semiconductor companies that are doing new things, trying to figure out how this all works and how it all comes together so they can provide something to the OEM. Now, Tier 1, which I believe is likely to survive this transition well, is teaming up with OEMs. That’s 0.5 and sometimes these companies are being broken up. An example is Foxconn, which is approaching 0.5. There are many developments that are the result of this merging of the need to put things together.