The chip industry is facing new challenges and opportunities in data management due to the increasing amount of data collected from design to manufacturing. The complexity of designs, customised and domain-specific, along with the rising demand for reliability and traceability, compounds the issue. Chiplets from different processes and foundries, new materials like glass substrates and ruthenium interconnects, and the terabytes of data generated by EDA and verification tools weekly or daily, add to this complexity.
Despite the potential insights and improved designs more data offers, managing current data volumes remains a challenge. The industry needs to rethink methodologies and processes and invest in new tools and approaches. This shift causes concern in an industry accustomed to cautious, step-by-step progress based on silicon- and field-proven strategies. AI and ML are increasingly integrated into design tools to identify anomalies and patterns in large datasets, but this continuous update of tools and algorithms complicates decisions about investment, data focus, and sharing.
Different companies have unique methodologies for data harvesting and report generation, often leading to excessive data accumulation without effective utilisation. The opportunity lies in re-architecting systems and processes to handle data efficiently, thus leveraging it for better outcomes. This involves significant changes, from data collection to organisation. Alphawave Semi’s CTO, Tony Chan Carusone, emphasises the need to leverage data in new ways, highlighting the disruptive nature of this transformation as it requires tearing apart and re-architecting existing systems.
The influx of data presents massive challenges. Figuring out what information is critical to keep, managing the continuous data flow, and optimising its use are key concerns. Data analytics tailored to the chip industry’s specific needs can help pinpoint issues like timing violations.
Security is another major concern due to the number of tools, companies, and people involved in the design process. Ensuring data security without hindering day-to-day operations is crucial, especially given the geopolitical scenarios and collaborative nature of the industry. The security protocol must be robust to manage sensitive data access and transfer, especially with globally distributed design teams.
To address these issues, identifying essential data is the first step. Smart algorithms can optimise storage and transfer times. Technologies developed in response to the data flood allow easy access to crucial information. Storing data in a standardised, accessible, and sortable way can mitigate less technical problems, such as institutional knowledge loss when engineers switch jobs.
AI plays a growing role in EDA design and data management, promising advanced data analytics that can recognise trends and make predictions. However, the costs involved in disk space, compute resources, and licenses make some clients hesitant to adopt AI fully. Despite the potential for AI to generate more data, its ability to manage and utilise data effectively could outweigh these concerns.
Hardware engineers can learn from software development to organise data collection, run parallel verification and testing, and maintain rigorous check-in and check-out processes for test cases. Automating these processes allows for efficient problem-solving and data analysis.
While new tools and AI increase productivity, effective data usage remains crucial. Proper data storage and organisation strategies, supported by powerful tools, are essential. Handling the data influx is a delicate balance, requiring effective communication and resource investment for top-down management. Achieving this can make the data valuable and improve chip design processes.
Alphawave IP Group plc (LON:AWE) is a semiconductor IP company focused on providing DSP based, multi-standard connectivity Silicon IP solutions targeting both data processing in the Datacenter and data generation by IoT end devices.
