Introduction
The 38th ACM/IEEE International Conference on Computer-Aided Design (ICCAD 2019) served as a vital pulse-check for the Electronic Design Automation (EDA) industry. Held in Westminster, Colorado, the conference highlighted a significant shift in the research landscape: a transition from traditional chip design optimization toward emerging frontiers such as hardware security, neural networks, and quantum computing. With a competitive 23.9% acceptance rate, ICCAD 2019 showcased the cutting edge of "Beyond CMOS" technologies and provided a roadmap for the future of integrated circuit design.
Conference Overview and Selectivity
As the premier forum for Computer-Aided Design (CAD) research, ICCAD 2019 addressed the escalating complexity of modern electronic systems. The primary challenge facing the community is that traditional EDA topics—such as placement, routing, and synthesis—are no longer sufficient on their own to meet the demands of modern hardware.
The conference received 394 regular paper submissions, which underwent a rigorous double-blind review process conducted by a Technical Program Committee (TPC) of 135 experts from both academia and industry. Ultimately, only 94 papers were accepted, reflecting the high standard of technical excellence required for the event. The program was organized into 14 distinct thematic tracks, ensuring specialized expertise in the review and presentation of diverse topics. Beyond the core technical sessions, the conference featured 11 special sessions, seven specialized workshops, and the "CADathlon" programming contest.
The Rise of New EDA Pillars
A primary takeaway from the 2019 conference was the evolution of EDA's core pillars. While traditional optimization remains essential, the high volume of submissions in hardware security and neural networks indicates that these fields have become primary pillars of the EDA research community.
Quantum Computing and Machine Learning
One of the most notable moments of the conference was a keynote from IBM that included a live demonstration of a cloud-based quantum computer. This signaled to the community that quantum computing is moving out of the realm of theoretical physics and into practical architectural consideration for CAD researchers. Simultaneously, the integration of machine learning into design flows was a recurring theme, with researchers exploring how to leverage ML to automate and optimize complex design tasks, particularly in the realm of neural network hardware.
Hardware Security
As hardware becomes increasingly globalized and complex, security has moved to the forefront. The conference highlighted the need for CAD tools that can secure hardware against modern threats, treating security as a first-class design constraint alongside power, performance, and area (PPA).
Technical Highlights and Award-Winning Research
The conference recognized several papers for their outstanding contributions to the field, highlighting specific technical trends in "Frontend" and "Backend" design.
- In-Storage Computing (Frontend Best Paper): The award was presented to Ruan, He, and Cong for their work on FPGA-based system-level emulation for in-storage computing. This research addresses the "von Neumann bottleneck" by moving computation closer to data, a necessary evolution for modern data-heavy workloads.
- Power Grid Reliability (Backend Best Paper): The award went to Moudallal, Najm, and Sukharev for their research on fixing electromigration-induced failures in power grids. As chips scale down, physical reliability remains a daunting hurdle, and this work provides critical insights into maintaining the structural integrity of power delivery networks.
- Long-Term Impact: The Ten-Year Retrospective Most Influential Paper award was given to a 2009 study on Nonvolatile Memristor Memory, underscoring the long-term importance of alternative memory technologies in the EDA landscape.
Future Directions: Beyond CMOS and Uncertainty
ICCAD 2019 also looked toward the horizon, emphasizing the need for guidelines for future nanoelectronic and nanomagnetic devices. As traditional silicon reaches its physical limits, the "Beyond CMOS" era will require entirely new CAD methodologies.
Furthermore, keynotes addressed the challenges of autonomous systems, specifically the need to combine probabilistic methodologies with machine learning to handle real-world uncertainty in robotics and self-driving cars. The report also noted the conference's geographic evolution—moving between tech hubs like Silicon Valley, Austin, and Denver—suggesting a need for the EDA community to stay connected with diverse regional industries.
Conclusion
The 38th ICCAD conference demonstrated that the EDA field is in a state of vibrant transformation. By reconciling traditional design requirements with the challenges of emerging technologies like quantum computing and in-storage processing, the community is building the foundation for the next generation of computing systems.
Summary updated