It’s an exciting time to be involved in open source. Linux powers the world’s most critical devices, a story to which Red Hat has always been a champion. Today we look further afield. We look to a world of not what “is” but what “could be”. It’s informed by the path we’ve walked and the lessons we’ve learned along the way. It’s a story of re-aligning goals in the spirit of community.
Red Hat’s Emerging Technologies blog includes posts that discuss technologies that are under active development in upstream open source communities and at Red Hat. We believe in sharing early and often the things we’re working on, but we want to note that unless otherwise stated the technologies and how-tos shared here aren’t part of supported products, nor promised to be in the future.
In the prior millennium, computers were designed behind closed doors. Compaq didn’t ask what folks wanted in an operating system… they *built* the operating system. As long as it conformed to POSIX, your software would be portable and could be compiled to run. It was a world of clearly defined amateurs working on MINIX versus those working on UNIX. One where folks might learn to design a processor using RISC if they were at UC Berkeley, but as soon as they graduated they were on to working on Alpha, MIPS, SPARC, x86, or a number of other smaller players.
If you were lucky enough to work on an academic team going through the full tape-out process of manufacturing a silicon processor, the licensing terms of any proprietary ISA would make it impossible to publish your research. This created a challenge, all of the practicing you did as a student could never _quite_ emulate what you would experience in the “real world” even at the pinnacle of the research.
This is changing, due to the adoption of RISC-V. The work on RISC-V for FPGAs at the Red Hat Collaboratory at Boston University exemplifies the transformative power of open source in hardware development. By leveraging the open RISC-V instruction set architecture, researchers can create customizable softcore processors that offer unprecedented flexibility and efficiency in FPGA designs. This approach allows for the optimal partitioning of workloads between specialized FPGA circuits and general-purpose cores, resulting in improved functionality, power consumption, and development turnaround times. The open nature of RISC-V enables developers to extend the ISA easily and securely, facilitating seamless integration with additional IP blocks and custom hardware. This level of customization and accessibility is democratizing hardware design, making it more approachable for software developers and fostering innovation in fields ranging from edge computing to data center applications. As the RISC-V ecosystem continues to grow, supported by a vibrant community and robust toolchains, it paves the way for a new era of open hardware development that promises to accelerate research and drive technological advancements across various domains.
The Importance of Licensing
The success of Linux accelerated as it was adopted by enterprises. Red Hat was founded on the idea that the way the world consumed software could be changed. We believed that there were enterprise users who would, if given the choice, reward the development of open source software. That there were users who saw the intrinsic value of freedom missing in proprietary software. This belief hasn’t changed and guided our collaboration in launching the RISC-V Software Ecosystem (RISE) project.
The work of RISC-V International (RVI), RISE, and The OpenHW Group is allowing members to collaborate at an unprecedented level and Red Hat is proud of our membership in these communities. We see the opportunity to show that licensing is important. Linux wasn’t the first “open source” operating system, but the combination of tools and kernel released under GNU Public License ensured future collaboration in a way that the BSD derivatives didn’t.
The open-source hardware movement has made significant strides in addressing licensing concerns for processor IP. The Solderpad Hardware License, specifically designed for open hardware designs, offers a permissive license that builds upon the widely understood Apache 2.0 license. This license provides a clear framework for collaboration and sharing in the hardware design space, addressing the concerns that previously hindered university researchers from using certain open-source hardware descriptions. Furthermore, organizations like the Free and Open Source Silicon (FOSSi) Foundation are playing a crucial role in promoting and protecting the open-source silicon chip movement. As the custodian of this movement, FOSSi Foundation enables everyone to collaborate, innovate, and enjoy the benefits of open-source chip design. Their efforts, including organizing conferences like ORConf and publishing newsletters, are fostering a vibrant community that is shaping the future of open hardware.
Zero Cost Compute & New Capabilities
In the area of microcontrollers, the “race to the bottom” has already begun. One can already buy a 48 MHz RISC-V processor from a reputable manufacturer, sold by a reputable vendor for 25 cents. This has been Red Hat’s reality all along. While we compete with “proprietary” software we also compete with “free as in beer” code. This is likely a future reality for silicon vendors: Tomorrow, Intel, ARM, and other silicon vendors need to compete not just on the strength of their products, but also against the market in a race to zero cost compute.
The recent development of Flex-RV, a bendable RISC-V microprocessor, showcases the transformative potential of open-source hardware. This groundbreaking device, fabricated using indium gallium zinc oxide thin-film transistors, opens up a world of possibilities for emerging applications in fast-moving consumer goods, healthcare wearables, and single-use medical devices. Imagine smart labels that can bend and conform to packaging, or flexible neural interfaces that can be implanted with minimal invasiveness. The open nature of the RISC-V instruction set architecture was crucial in making this innovation possible, allowing researchers to customize the processor and add machine learning capabilities without the burden of restrictive licenses or exorbitant fees. This is a prime example of how open access to fundamental technologies can spark creativity and drive progress in unexpected directions, potentially revolutionizing fields from supply chain management to personalized medicine.
In these ways, Linux changed many of our lives. It was a platform for individuals to engage in an endless self-study of the architecture of computers. More than that, it allowed a fan(atic) to explore what they found *interesting* and not just what their employer deemed “necessary.” It allowed a generation of upstarts to sneak in through the back door with practical knowledge gained through direct experience. As Linux took over the servers of the world this direct experience became even more broadly applicable. Now, we live in a world where one can look at the “sources” for a whole hardware project like the CORE-V MCU development kit. This project can allow the user an unprecedented level of visibility into every aspect of the system. As these platforms continue to scale, we will provide new routes for users to learn about the nuances of system design decisions from cache hierarchies and ordering decisions all the way down to microarchitecture choices.

In this world where one can refine an open design for a processor tailored to their particular needs, not everyone will want to have to maintain an operating system for the duration of the existence of that processor. This is where “generic” Linux distributions exist today. However, in order for a distribution to make economical sense, a distributor would have to recompile the distribution for each of the processors that people may design, which seems difficult to envision. This is actually one of the flaws we have seen in the ARM space and the reason why, on non-server type ARM processors, the commercially available Linux distributions are scarce.
Red Hat, along with some of the main Linux distributions, is hard at work defining a way where extensibility of a processor design does not necessarily rhyme with having to recompile a kernel. There are ways to inform the distribution, at boot time, of the extensions that were made in a particular design. The good news is that the RISC-V community is taking those proposals seriously and the future that we are dreaming of is becoming closer everyday.