Open-source Design Tools for Sovereign Chip Development

Abstract

Chip design is a key component in the development of innova tive microelectronics and greater expertise in this field is vital to improving competitiveness and reducing Europe’s and Germa ny’s associated geopolitical vulnerability. To make the most of Germany’s potential, the Federal Ministry of Research, Technol ogy and Space (BMFTR), is implementing targeted measures to strengthen the chip design ecosystem with the Microelectronics Design Initiative. As part of the Microelectronics Design Initiative, the aim of the DI-QDISC subproject is to systematically analyse the potential and challenges of open-source design tools for sovereign chip development and to develop policy options for appropriate further development. This means investigating the ecosystem of open-source design resources and tools for electronic design automation (EDA) along the entire microelectronics develop ment chain. The focus is on the current status of, development prospects for and existing obstacles to the use of these tools for analogue, RF, digital and mixed-signal design as well as for microelectromechanical systems (MEMS). In addition to techno logical aspects, attention is also being paid to economic and application-related issues. Even today, open-source design tools are complementing pro prietary solutions and contributing to European sovereignty in chip design, especially because of their central role in academic training and education. However, they are not yet capable of meeting every industrial requirement. In both digital and analogue design, open-source design tools hold particular promise for mature technology nodes where open-source design tools can reduce barriers to entry by providing an easily accessible and inexpensive alternative to proprietary solutions for parts of the process chain, in particular at the design space exploration phase at the beginning of the design process. Practical examples from education, research and industry have al ready demonstrated that innovation mechanisms from the open source software sector can be transferred to electronic design automation in microelectronics. The innovation mechanisms used in open software development also have the potential in future to accelerate innovation processes in chip design. The largely intangible nature of chip design makes such an approach promising, as new solutions for design tools – similarly to AI Empowering small and medium-sized enterprises in chip de sign through open-source development tools is an important contribution to maintaining Germany’s international competitive position in cutting-edge technologies. In comparison with other regions, Germany’s technological performance has declined in recent years. Differentiating features in key industrial markets are increasingly being developed at the chip design level, which is why there is a need for deeper value creation through research and development. The project group recommends the following measures to strengthen Germany’s and Europe’s technological sovereignty and innovation capacity in microelectronics and to establish a self-supporting ecosystem for open-source design tools. § Availability of skilled personnel: in view of the impending loss of expertise due to retirement and the growing impor tance of microelectronics, training capacity at universities should be expanded in all areas relevant to chip design. Internationally competitive working conditions and pro grammes such as the Humboldt Professorships could attract top international talent to Germany. In addition, the next generation should have the opportunity to gain hands-on experience with open-source tools for chip design at an early stage as part of their own small chip development projects. § Industrial maturity by corporate participation: an independ ent, largely industry-driven open-source EDA user forum (e.g. in the form of a foundation) could act as a central clearing house between users and developers in order to further de velop open-source design tools in a targeted manner. Direct collaboration means common standards can be developed and seamlessly integrated into industrial processes. Embed ded in a long-term semiconductor strategy, this could foster a market-driven innovation policy. In addition, when selecting chip design tools, companies should – where possible – give preference to those that support open standards to ensure compatibility with open-source solutions. It should always be checked whether internally developed solutions can be shared with the open-source community. § Government kick-start: as part of its economic development initiatives, government could set up three programmes to support different points in the development ecosystem. 1. Government should support the establishment of a design platform as a “one-stop shop” to enable design, including for SMEs and start-ups. This platform should include an unbroken open-source tool chain and various offers for tape-outs. For companies wishing to make their own developments available research – can be transparently shared, tested and optimised. to the open-source community, this “one-stop shop” could guarantee long-term maintenance as a trustworthy plat form. This organisation could also orchestrate the integration of new solutions and the further development of existing tools and host them on its own platform. 2. So far, German industry has played only a very minor role in the further development of open-source tools for chip devel opment. A sponsored EDA development voucher is therefore proposed as an incentive to engage with open-source design tools alongside a corporation’s actual core business and to 1 develop them further to industrial maturity. If they have a use case, companies will be able to apply for a voucher that they can use to award development contracts for the further devel opment of open-source EDA tools which will then be available under an open-source licence. 3. Europe’s entire semiconductor ecosystem would benefit from the establishment of an unbroken development chain. Government support should therefore be provided for the development of open-source PDKs, for example 22 nm1 at GlobalFoundries and 130 nm at IHP.