Electronic design automation (EDA) is often treated as a semiconductor-industry concern, but its direction also matters to industrial automation. The tools and design methods used to build processors, embedded systems and specialized chips influence the hardware available for control, edge computing, machine vision and other data-intensive tasks. The 2020 Asia and South Pacific Design Automation Conference (ASP-DAC) highlighted several developments that engineers and technical decision-makers should understand—not as ready-made plant solutions, but as signals of where computing hardware and its design processes are evolving.
ASP-DAC 2020 brought academic and industry participants together around integrated-circuit design and automation. Its program covered topics including AI processors, memory architectures, reliability, testing, security and emerging technologies. Taken together, these themes show how design teams are working to balance computing performance with constraints such as power, reliability and implementation complexity.
Electronic design automation connects chip design to system needs
EDA encompasses the methods and software used to design, analyze, verify and prepare electronic circuits and systems for implementation. The conference program covered work across this design chain, from system-level modeling and high-level synthesis to physical design, timing, verification and manufacturing considerations.
For industrial automation professionals, the connection is indirect but important. A control or edge-computing application depends on the capabilities and constraints of its underlying hardware. Advances in chip architecture can create new options, while design-for-test, verification and reliability methods help address the challenge of making complex hardware dependable. These developments do not, by themselves, establish suitability for a particular industrial environment; that still requires application-specific evaluation.
AI accelerator design focuses on more than raw compute
AI hardware was a prominent theme in the program. Sessions and tutorials addressed AI chip technologies, application-specific processors, neural-network computing and accelerator architectures. The Designers’ Forum also examined AI accelerators, including computing-in-memory approaches and the organization of accelerator resources for AI applications.
The practical lesson is that accelerator design involves choices across architecture, software and implementation—not simply adding more processing capacity. Application-specific processors are designed around particular workloads, while high-level synthesis and customized compilers can help translate application requirements into hardware implementations. For industrial teams considering AI at the edge, these topics are relevant when assessing whether a workload can be supported within the available device and system constraints.
Conference presentations also addressed neural architecture search and compression for efficient deep learning. These are approaches to adapting models and their implementation to hardware constraints. They are part of a wider design problem: matching computational needs to the resources available, rather than assuming that a model can be deployed unchanged on every platform.
In-memory computing addresses data movement
Memory architecture and near- or in-memory computing received dedicated attention. The underlying challenge is that moving data between memory and a processor can limit performance, particularly for data-intensive workloads. In-memory computing seeks to perform selected operations within or close to memory, reducing the need to move data back and forth for those operations.
ASP-DAC’s tutorials discussed ferroelectric field-effect transistors (FeFETs) as a possible basis for integrated memory and computing functions, including content-addressable memory and compute-in-memory circuits. These are emerging-device and design research topics, not a general-purpose replacement for conventional computing. Their relevance to automation is as a potential direction for specialized processing, where system designers would need to consider the technology’s circuit, architecture and application-level behavior together.
Reliability and test remain central design concerns
As chip designs become more complex, testing and reliability cannot be left until the end of development. The conference included work on design for test (DFT), testing, validation, simulation and verification, alongside a tutorial on machine learning for the reliability of integrated circuits and systems. A separate special session addressed designing reliable and robust circuits in the nanometer era.
DFT methods help make a chip’s internal behavior more accessible to manufacturing test and diagnosis. Verification and simulation, meanwhile, support checking that a design behaves as intended. For industrial automation decision-makers, these disciplines matter because the computing hardware selected for a system must be evaluated not only for function, but also for how it can be tested and supported. The conference program highlighted the methods being explored; it did not prescribe qualification criteria for industrial equipment.
Security and emerging technologies broaden the design challenge
Hardware security was another thread, spanning security engineering for in-vehicle networks, hardware-based security for the Internet of Things, and security testing. The Designers’ Forum also featured work on reconfigurable cryptographic hardware. These topics reflect a need to consider security at multiple levels, from circuit design to embedded-system architecture.
The program also covered emerging technologies such as monolithic three-dimensional integrated circuits, negative-capacitance field-effect transistors and non-volatile memory. Their potential implications vary, and moving from device research to a practical system requires work across several abstraction levels. This is why the conference’s emphasis on cross-layer design—from devices and circuits through architecture and applications—is notable.
What industrial automation teams can take away
ASP-DAC 2020 offers a useful snapshot of research and industry discussion in electronic design automation. For automation engineers, system integrators and operations leaders, the main takeaways are:
- Evaluate computing at the system level. Architecture, memory, software and workload characteristics interact; a chip feature alone does not determine application suitability.
- Ask how hardware is tested and verified. DFT, simulation and verification are integral parts of design, especially as functionality and complexity grow.
- Track specialized computing carefully. AI accelerators and in-memory approaches may suit particular workloads, but require application-specific assessment.
- Include security and reliability in design discussions. These concerns span hardware and embedded-system architecture rather than sitting apart from implementation.
The practical implication is not that every emerging chip technology is ready for deployment in a factory. Rather, EDA research helps shape the hardware options that industrial systems may use in the future. Understanding its themes can help technical teams ask better questions about compute architecture, testability, security and workload fit when evaluating embedded and edge platforms.