As demand for artificial intelligence (AI) dedicated semiconductors surges, a new chip design method that addresses computation and power efficiency issues that existing general-purpose central processing units (CPUs) struggle to handle is gaining attention. The open-source-based instruction set architecture (ISA) "RISC-V" is representative of this trend. This design asset (IP) was devised by computer science luminary David Patterson, a professor at the University of California, Berkeley, along with his research team in 2010. Being open source, there are no separate licensing expenses compared to IPs from Intel or ARM. This enables anyone to freely design their own CPU. Companies like SiFive and Tens Torrent are designing semiconductors based on RISC-V. Market research firm Omdia forecasts that shipments of RISC-V based semiconductor chips will reach 17 billion by 2030, which accounts for a quarter of the global semiconductor production.
How does Professor Patterson, known as the "father of RISC-V," view these changes in the industry? When I emailed him, he immediately replied, "There will be more industries adopting RISC-V in the future." He noted, "Looking back at the history of the computer industry, technological advancement has occurred through openness, not closure," and emphasized, "While various new technologies like neuromorphic computers and quantum computers will emerge in the future, RISC-V will not disappear and will play an even more important role than now." I asked Professor Patterson about the future of open source and RISC-V. This is his first interview with Korean media. Below is a Q&A.
Did you expect RISC-V to receive such industry attention back in 2010 when it was first created?
"No. The initial goal of Professor Krste Asanovic and I, who were conducting research together at the time, was simple. We aimed to create a design method that was useful enough to be used in research and taught in classes. We only had hopes that other universities would perhaps follow the method we developed."
Nevertheless, many corporations around the world are adopting RISC-V. In 2020, you mentioned that the industry atmosphere shifted from "Why should we use RISC-V?" to "Why shouldn't we use RISC-V?" What is the current atmosphere like now, five years later?
"Even in 2020, the technical and business advantages of open architecture were clear. I believe that from that time, industry interest and momentum surrounding RISC-V began to solidify. Even now, five years later, that momentum has not stopped. Currently, RISC-V has established itself as the preferred design method in fields requiring small and efficient computer chips, such as air conditioning temperature controllers or smartwatches. In fact, the RISC-V design method is rapidly expanding into the automotive and data center sectors as well."
I'm curious about your insistence on open source. If you had abandoned open source, you could have gained significant commercial benefits.
"UC Berkeley has a long tradition of open source development. Its roots trace back to the 1970s computer design tool 'Spice' and the open-source operating system 'Berkeley Unix,' which was widely used before Linux in the 1980s. We believe that by making software public, more volunteers participate, enhancing quality and naturally leading to wider spread. We thought that if RISC-V were also developed as open source, more people would be interested and help evolve it together."
Chinese startup DeepSeek has released its AI model as open source, and Jim Keller, CEO of Tens Torrent, noted, "In the end, open source wins." Has the era of open source arrived?
"Over the past nearly 50 years in the computer industry, open source software and technical standards have more often triumphed over proprietary technologies controlled by specific corporations. This demonstrates that technological advancement happens through openness, not closure. Technologies accessible to everyone have ultimately led to greater participation and faster innovation. Thus, the question we must ask now is, 'Why should proprietary technology win this time?'"
What should true open source look like?
"There exists something called 'formal open source,' where even though anyone can see the code, only specific corporations can modify or set the direction. In contrast, true open source involves a structure where anyone can view the code, and the developer community actively participates in its improvement. RISC-V embodies this approach. We established the non-profit foundation 'RISC-V International' to ensure that RISC-V develops as a public good for developers and the entire industry, rather than being controlled by specific corporations. This foundation guarantees the openness and neutrality of RISC-V."
Over the past several decades, you have left behind lasting works in computer engineering history, from "RISC" to "RAID" and now RISC-V. I'm curious about the secret to your innovation.
"All the achievements you mentioned were not accomplished by me alone. It was made possible with the help of colleagues. RISC was with Professor John Hennessy (current chairman of Alphabet), RAID was with Professor Randy Katz, and RISC-V, as I mentioned earlier, was with Professor Asanovic. Additionally, the contributions of dozens of outstanding Ph.D. students who turned these achievements into reality cannot be overlooked."
If you were to advice young engineers in Korea.
"I have a career advice lecture aimed at young engineers and researchers. The title might sound a bit playful, but it actually conveys serious content. It's titled 'How To Have A Bad Career.' You can find the video on YouTube. Recently, I also wrote a column reflecting on my personal and professional experiences. The title is 'Life Lessons from the First Half-Century of My Career,' which includes helpful advice as well."
In that column and lecture, Professor Patterson advises that career success comes not from starting many tasks but from completing them. He also conveyed that "who you work with is much more important than what you do," encouraging a people-centered career design and prioritizing family in life.
How do you forecast the future of computer architecture? New technologies such as AI dedicated chips, neuromorphic computers, and quantum computers are emerging. In this process, could RISC-V be replaced?
"The strength of RISC-V lies in its robust core structure, designed to freely combine accelerators tailored for specific purposes. This allows for flexible adaptation to technological changes such as AI dedicated hardware or neuromorphic computers. If quantum computers become commercially viable, they may play a significant role in solving some problems that existing computers struggle to address, but they are not suitable for all fields. For example, AI requires vast amounts of data and memory, which is a weak point of quantum computers. Additionally, as they only operate in extremely low-temperature environments, the possibility of quantum laptops or smartphones emerging is nearly non-existent. Ultimately, quantum computers are likely to remain in data centers. Considering this, RISC-V will play an even more critical role in 2035, when it celebrates its 25th anniversary, than now. There is no reason or sign that it will disappear. Let's meet again in 2035 and discuss how accurate this prediction was."
Plus Point
Professor Patterson is considered a pioneer in the advancement of modern computers, having developed the groundbreaking chip design method RISC in 1981. This method simplifies hardware structure by reducing the number of instructions for computer CPUs. Most of the computer chips produced worldwide today are designed on the basis of RISC. The semiconductor design corporation 'ARM' is an acronym for 'Advanced RISC Machine,' indicating that the company's identity is rooted in RISC.
In 1987, Professor Patterson also created RAID, a technology for distributing data across multiple disks, which later became the technical foundation for cloud storage methods. In 2010, as the director of the Parallel Computing Laboratory (Par Lab) at UC Berkeley, he played a pivotal role in establishing the research direction and philosophical foundation when Professor Asanovic, graduate student Yoonseop Lee, and Andrew Waterman developed RISC-V. In 2017, Professor Patterson was awarded the Turing Award, regarded as the Nobel Prize for computer science, in recognition of these contributions. His book "Computer Architecture and Design" continues to serve as a core textbook in universities worldwide, including those in Korea.
In addition, "RAY," a corporation in South Korea, was founded based on related technologies.