Opportunities And Challenges With Open-Source Hardware In System Development

For many years now, there has been a trend toward open source in the field of system development. It can be seen in software libraries for the product itself as well as in development tools.

A clear motivation for open source lies in the fact that not charging license fees makes a product more attractive on the market. It may also enable further development of the software component, depending on the license model. However, few companies will make use of this option, as it leaves them dependent on the further development of the main product. Another advantage of open-source developments is that they often enjoy a larger and more widely distributed user community, which can take care of maintenance and further development and clarify unresolved questions. The more specialized the open-source component, the smaller this benefit.

Having a distributed developer community, as is common in the case of open source, can also be a disadvantage. Manufacturers of safety-critical products, such as those used in the automotive sector, absolutely need their suppliers to be able to point to a specific software developer so they can respond to liability and warranty issues. Otherwise, legal requirements are left unmet and incalculable risks arise.

“Open source” commonly refers to software, but a lively open-source hardware community has also been developing for some years now. There is, of course, a fundamental difference: With hardware, the design data can be disclosed and distributed free of charge as an alternative to licensed IP. However, the production of silicon is always associated with costs for mask sets and wafer runs. Depending on the planned number of units, these costs can be reduced by using multi-project wafers (MPWs).

One of the best-known representatives of open-source hardware is the RISC-V processor’s open instruction set architecture (ISA). This is a topic of growing interest for an ever-expanding community. The architecture is finding its way into many chip design projects for non-safety-critical applications. However, additions have now also been made to the architecture for safety-critical areas.

As with software IP, the added value of open source in hardware lies in the transparency of the installed components. Theoretically, it is possible to test the design IP for errors and exclude backdoors or malicious code. The large amount of work this involves can be reduced by distributing tasks within the community. In this way, open source can always help protect and create trustworthy electronics.

Another aspect of open-source hardware is the availability of technology libraries for production in a specific foundry. The process design kit (PDK) required for this is usually free of charge anyway, as the money is earned with the production of the silicon. However, foundries generally only provide the PDK for their technologies upon the signing of a non-disclosure agreement (NDA). This is to ensure that knowledge about the technology is not disseminated in an uncontrolled way. The PDK contains critical information on the technology’s possibilities and its limitations, and may even allow conclusions to be drawn about the steps in the production process.

In sum, an open-source PDK is a sensible way to provide start-ups and research institutions with low-threshold access to semiconductor technologies. This also greatly reduces the administrative workload for design projects in education. The SkyWater PDKs were a good example of this, and their discontinuation has left a gap in the market. But the Europe-based foundry IHP, for instance, offers an open-source PDK for its 130nm technology. Although this is a comparatively large technology node, it takes into account special requirements such as radiation-hardened applications.

The market for development tools in chip design is currently heavily dominated by a few commercial providers. While open-source alternatives are emerging here, too, they mainly occupy niches in the field of research and teaching. One reason for this is certainly the huge amount of effort required to put together a complete development process, from a high-level description to physical implementation, sign-off, and validation using tools from a wide variety of sources. Another challenge is that the foundries have to support the open-source tool with their technology libraries. This effort is already very high for the few commercial providers of electronic design automation (EDA).

Open source is a promising alternative at various points in the development of integrated hardware/software systems. The additional opportunities offered by a broad community and transparency are offset by challenges such as compatibility and risks in the area of warranties. It remains to be seen what subsectors can be filled and which of these will prevail on the market. Development will certainly continue in this direction.