Imagine your dream car of the future. Perhaps it’s completely autonomous and can drive you wherever you need to go, while allowing you to get some work done and catch up on your favorite TV show on the road. And the cherry on top? The vehicle drops you off while it finds a parking spot where it can charge itself while it waits for you to recall it to your current location.
This description of the futuristic car points to a new automotive trend where vehicles are increasingly becoming data centers on wheels that need to be able to monitor their “health” in order to function effectively, efficiently, securely, reliably, and safely. Silicon lifecycle management (SLM) is the answer to the many questions that arise as we move toward self-driving, electric vehicles (EVs) with more sophisticated infotainment systems.
The automotive industry is witnessing compute consolidation and expansion of compute power in the vehicle, as increasingly complicated features are being powered through an EV charging system in a wide range of environments and temperatures across the globe. The question becomes not only how we do this, but how do we know the more advanced silicon being put into cars is working well and will work well for years to come. With a vehicle’s average lifecycle expectancy increasing to 15+ years, such parameters become critical to scale future updates.
SLM provides a way to monitor the many stages of automotive systems-on-chips (SoCs) from testing and manufacturing to their function in the vehicle. This data is critical to OEMs as they deploy over-the-air (OTA) updates that proactively solve issues for today’s vehicles. SLM is also relevant to the next generation of software-defined vehicles as OEMs gather insight and visibility into key challenges and determine how they’ll need to shift their production to address those.
Read on to find out what the largest technology challenges are for automotive chips, the corresponding OEM difficulties, and how SLM can help address both categories of challenges to help make your next software-defined vehicle run for longer, provide more convenient features, and become more resilient against security and safety threats.
Automotive SoC challenges
New, custom automotive-grade SoCs are necessary to handle the increase in centralized compute required for the software-defined vehicle. As these automotive chips become smaller and more complex, the physics of these new form factors will increase the need to understand their performance.
While the entire industry is facing new challenges posed by the accelerated scaling of device and system complexity, the challenge becomes even more complicated with automotive silicon due to increased safety, reliability, and security needs. Below are the four main buckets of challenges automotive chip designers are faced with
· Accelerated Adoption of Advanced Process Nodes: For each new technology node, transistor densities continue to increase. While this density provides a great opportunity for adding technology capabilities, it also creates new challenges such as significant variability in the manufacturing process. This broadens the design envelope unless process variability across the die can be measured using sensors and monitoring structures. Automotive companies are now exploring multi-die systems as a solution to overcome this scale complexity challenge.
· Multi-Die System Adoption: As packaging becomes more advanced for multi-die systems, combinations of dies are ‘married’ together in various configurations, from stacked dies to 2.5 and 3D packaging and beyond. Given this, it’s important to be able to trace where each die sits on the process spread.
· System Complexity: Data aggregation as it relates to security, aging, and degradation, as well as power and computational throughput are all concerns stemming from system complexity. Additionally, the in-field systems of tomorrow will have multiple software updates throughout their lifetimes. If not managed carefully, updated software may cause vehicles to consume more power, shorten their lifetimes, and have a negative impact on user experiences.
· Workload Increase: Finally, workloads for automotive chips can be unpredictable, requiring real-time optimization and application diversity which can add further challenges and extra considerations.
Automotive OEM challenges
Combining the technical challenges of automotive silicon with the broader landscape of OEM challenges is where the rubber really meets the road for SLM solutions. There are many different obstacles and considerations for OEMs when designing their vehicles and making decisions about how to resolve issues that arise during a vehicle’s lifetime on the road.
· Warranty Costs and Recalls: The more complicated a system, the greater the opportunity for potential failures and the harder it is to resolve them in a timely manner. Recalls can become a huge cost consideration for OEMs. This also leads to a greater impact on supply chain disruptions and, as we mentioned above, potentially more chip shortage events.
· Increasing Security Challenges: As more cars receive OTA software updates, new vulnerabilities arise and can become especially worrisome if they were to affect a self-driving car. OEMs are increasingly focusing on reliability, security, and safety given these new factors.
· Overhaul of Electrical/Electronic Architectures: Zonal architectures are changing with new features such as electrified powertrains, advanced infotainment systems, the ADAS / L3+ autonomous driving level, and overall faster release cycles.
· Accelerating Time to Market: New global entrants into the automotive field have put pressure on existing OEMs to speed up their own more traditional design processes. This pressure also has implications for supply levels and costs of SoCs, global availability, and more.
Moving forward, electronics will comprise the most important components of vehicles and will affect the above factors and more. OEMs can simply no longer afford to be blind to what is happening inside chips as that eats into profits, directly affects the safety of drivers, and can lead to missed opportunities to make their vehicles the most advanced on the market. SLM is the key that leads to more awareness about what is going on inside a vehicle as well as enabling the vehicle to proactively fix issues itself to become “self-healing.”
How SLM addresses both SoC and automotive OEM challenges
Simply put, SLM solutions enable increased visibility and insights that can be used to finetune not only the SoC for your next-generation vehicle but also the workload on your existing SoC based on all the data that has been collected throughout its lifecycle. A solid SLM solution allows users to monitor for problems very early in the design process, transport that data to a centralized database, analyze the data throughout the lifecycle of the vehicle, and strategically act at any point necessary. Ultimately, early warnings and accurate remediation enable hardware to scale for future updates.
SLM allows for root cause analysis, predictive maintenance, alerts for aging and degradation, and in-field voltage profiling that bring both the end customer and the OEM true value. On the predictive maintenance front, silicon analytics can provide more granular information for fast, accurate diagnosis. For example, extreme temperature warnings for ASIL-rated silicon can result in customer updates to remediate with action, service, or an OTA update which can help avoid long-term damage to crucial systems and avoid mass recalls.
Ultimately, deploying SLM for automotive SoCs directly translates to cost reductions and savings for the OEM, increases the lifetime value and lifespan of vehicles, boosts reliability and troubleshooting ability, and eases automotive chip shortages. While automotive SLM has been around in some capacity ever since chips were used in cars, the use cases for it have become more advanced as we move beyond mature nodes and into leading-edge nodes for automotive. With more sophisticated features powered by smaller chips come more challenges that SLM is prepared to solve. In addition, SLM solutions can address the predictive maintenance requirements that will come into play with new amendments in the ISO 26262 series of standards and ISO/SAE 21434 monitoring and analysis requirements.
Chris Clark, a senior manager in the Synopsys Automotive Group, is a 23-year veteran of the IT world who uses his experience in automotive systems, embedded device testing, and cybersecurity practices to help automotive organizations effectively integrate meaningful security solutions into their environments. Clark holds a master’s degree in cybersecurity from the University of Maryland University College and has also earned numerous certifications throughout his career.
Pawini Mahajan is a Staff Product Manager within the Hardware Analytics and Test group at Synopsys, specializing in silicon life cycle management solutions for Automotive. In this role she is responsible for Automotive Go-to-Market activities, understanding and communicating SLM-based solutions for automotive applications to the broad automotive ecosystem. Before joining Synopsys, she managed an engineering team at Intel focused on reducing defective parts per million (DPPM) for Intel's flagship products. In addition, she has actively participated in several notable fundraisers, including cancer awareness, women in technology and academic fund gifting. She holds a master's degree in electrical engineering from Arizona State University.