The birth of the RISC-V processor architecture has undoubtedly excited many people. However, although many say that RISC-V heralds a broader open-source hardware movement, it is not easy to determine why this architecture has succeeded, and whether it will promote the emergence of more open-source chip cores is still unknown.
“RISC-V has become the preferred architecture for many chip developers, triggering a wave of innovation in the hardware development community,” said Stephano Cetola, Technical Project Director of RISC-V International. “Designers are designing various architectures based on RISC-V and developing actual chip implementations across many industries.”
This is not the first time the industry has opened up the implementation of processors, or what is known as Instruction Set Architecture (ISA), to the public domain. There are many such examples in the industry, including OpenPOWER, OpenSPARC, OpenRISC, and so on. While each architecture has achieved a certain level of momentum, they pale in comparison to the achievements RISC-V has made in a short time.
When communicating with RISC-V community members, you often hear two words—free and freedom. Some people hope that the processor core is free, while others want the freedom to use the core and do what they want. For these people, whether it is free is not very important since they will spend a lot of money to obtain what they want.
The rise of RISC-V coincides with other events in the industry.
First, Moore’s Law has slowed down, meaning that each new chip manufacturing process node no longer implies a geometric increase in chip computing power. Secondly, the rapid development of machine learning has led to a huge demand for computing power.
So, does this mean that RISC-V has simply appeared at the right time and place?
People’s views on processors have changed. Simon Davidmann, founder and CEO of Imperas Software, said: “RISC-V is driven by the hardware design community’s demand for freedom.” “Electronic products are defined by their functions, many of which are defined by software that runs on processors. Today, all software requires some form of machine learning capability. For instance, your phone needs machine learning support to take better photos, which in turn requires a lot of computing power, and so on. What people realize is that they need a lot of processors. They need their own processor architecture. You need to configure them the way you need. Off-the-shelf technologies are not enough. Therefore, there has been a shift in the electronics market, saying ‘we need the freedom to design chips, design processors, and the processor architectures on these chips.’”
Many agree with his view. Roddy Urquhart, Senior Marketing Director of Codasip, said: “As the pace of general-purpose computing power expansion of chips gradually slows down, the only way to provide higher performance is specialization. The open RISC-V ISA is modular and supports custom instructions, making it an ideal ISA for creating a wide variety of specialized processors and accelerators.”
We must see that behind all these demands is a new generation of system companies entering the market, each with its unique business motivation. But a common point among these system companies is that they are not trying to sell the chips they develop. Instead, they are selling services driven by these products in some way. They cannot find suitable products from the existing market, so they are ready to develop chips themselves while driving some necessary innovations through contribution and collaboration. In this case, RISC-V plays an important role.
RISC-V is creating breakthroughs in multiple fields, and the reasons for success in each field vary. To understand this, it is necessary to look separately at the various factors contributing to RISC-V’s success. First is the architecture itself. Second is the large number of open-source architecture implementations in development. The third aspect is the support cores around the processor core. Finally, there are the necessary tools to help companies implement and verify RISC-V processors.
It was initially created to meet a specific need. Davidmann from Imperas said: “Now it is backed by a large amount of resources.” “Initially, it came from universities and academia, where some smart people made a good thing. After coming out of Berkeley, it gained momentum from Silicon Valley and many former Berkeley graduates. Its momentum is much stronger than that of OpenRISC. Many universities need it, and they pushed its development.”
RISC-V is now an open standard ISA, driven by the University of California, Berkeley, and there is a non-profit industry organization taking care of it—RISC-V International. Many universities have created open processor cores, such as Berkeley’s Rocket core, the pulp platform from ETH Zurich, and so on. Today, there are many industry consortiums that bring together industry and academia to build open-source cores and provide them to the entire community. Examples in this regard include the CHIPS Alliance and the OpenHW Group.
Many countries have proposed projects that meet local needs. India has the Shakti project, driven by IIT Madras. In Israel, the GenPro Alliance has united industry and academia. Japan and China also have similar projects, and they are developing open-source RISC-V cores to serve their communities and specific interests.
RISC-V is the first open and customizable ISA. “Currently, the main industrial interests related to RISC-V are not about open-source implementations but about the open-source instruction set,” said Andy Heinig, head of the Advanced Systems Integration Group at Fraunhofer IIS Adaptive Systems Engineering Department, “With it, the environment is standardized, but the actual implementations are designed for specific company needs and are owned by the company. We see similar activities in the field of interfaces between chips, where people are preparing and discussing many interconnection standards. On the RISC-V side, these interconnection standards can also support interoperability between chips from different vendors, and RISC-V provides interoperability on the software side.”
The ability to modify is very important. Shubhodeep Roy Choudhury, CEO and co-founder of Valtrix Systems, said: “Open-source, custom extension-supporting ISAs like RISC-V give processor designers incredible freedom.” “At the same time, it also brings a very interesting verification challenge. To ensure that all designs meet requirements and function correctly, the design of test generators needs to change. They need high configurability to validate custom features and traditional/benchmark features.”
Moving from an open ISA to open-source processors is a significant leap. “The concept of open-source IP is very appealing because it conjures up the idea of free IP,” said Andy Jaros, Vice President of IP Sales and Marketing at Flex Logix. “However, open-source does not equal free. Most companies will obtain licenses for pre-implemented RISC-V cores from numerous IP vendors (such as Open5, Andes, and many others) unless they want to invest significant resources in IP development. This saves the investment in development, verification, software development, etc., and provides many guarantees.”
Many companies are developing competing cores, which also promotes the innovation process in the implementation field. “The real value of RISC-V is that it has become a competitor to Arm, not because it is open-source,” Jaros added. “We see many RISC-V core vendors providing choices to the market and promoting competition. On the Arm side, you can only get Arm processor cores from Arm.”
Another driving factor is the rapidly increasing number of cores, making instance-based patent fee pricing seem quite expensive. “People want processors spread across their designs,” Davidmann said. “They want many small processors, but existing Arm licensing terms make it difficult to achieve that. Of course, they are also expensive, but more importantly, people cannot modify Arm cores at will. I do not believe that RISC-V’s success is because it is cheap or low-cost. If you just want to do the same thing as an Arm core, you should definitely buy an Arm core because it has been thoroughly validated and designed very well—that’s exactly what you want. The only reason to use RISC-V is that you want the freedom to change it and add your own things to it.”
Even with all these favorable factors, RISC-V might not have succeeded without the thriving ecosystem around it. “Many important tools developed by the open-source community are key factors in helping people develop a variety of processors based on RISC-V, such as chip technology design kits, design verification kits, implementation tools, etc.,” said Cetola of RISC-V International. “This has also democratized VLSI design, promoted the development of higher-level design description languages, and complex open-source automation tools have accelerated design progress, all of which have greatly enhanced RISC-V’s capabilities. With design tools and toolchains, RISC-V will soon become truly ubiquitous.”
The OpenHW Group is one of the collaborative projects driving the industry to achieve this goal. It is developing various processor cores and supporting peripheral IP for these cores. In addition, it is establishing a complete set of tools for designing and verifying these cores. “They do things differently,” Davidmann said: “One is that they will give you the source code so that you can modify it yourself. More importantly, they also provide you with a verification environment, so that if you make changes, you can verify that it works. If someone just throws a core at you and you change some of the code, you might break something. You need a complex verification environment to know if you have broken it. And that is what makes OpenHW unique in the open-source hardware field because they provide a complete verification environment. If you add a new instruction, you can know if you have broken other instructions. I do not think people will just take an OpenHW core and start using it. That makes no sense. If you want to save money, you can do that. But the significance lies in that you can take it to extend the core, and it is a very good starting foundation. That is the key. You are not starting from scratch.”
Can this open-source momentum expand beyond the processor core?
Processors are just a small part of a complete SoC. An SoC also needs memory controllers and memory interfaces, USB, PCI, and more components. These components are not special for products, and many people hope they are open-source as well.
The problem is that these components are very complex; they contain analog parts, and these parts are often custom-designed and implemented for specific foundries and process technologies. While controllers can also be built in an open-source manner, we can say that without the tight integration of digital and analog parts, the product that comes out may be substandard.
LowRISC is an organization established in the UK. It initially wanted to build an open-source system similar to Raspberry Pi. Today, it develops hardware and software within a completely collaborative framework, including RISC-V cores and the software compilation infrastructure supporting it.
Recently, Google created a specification and IP for silicon root of trust. It open-sourced this work and commissioned lowRISC to manage it. Part of the focus here is that openness and transparency will ultimately improve security and trustworthiness, rather than just providing the ability to modify specifications.
RISC-V has realized and facilitated innovation. While free may be important to some in the industry, the real key is freedom. This freedom has brought together like-minded talents, companies, and organizations, allowing them to jointly explore new fields. Its contribution to innovation is not only in breadth but more importantly in depth. While RISC-V may turn more hardware modules into open-source, perhaps the most important takeaway is that people can quickly adopt the open specifications of processors and implement them.
Once upon a time, when engineers stopped developing their custom processors, the tools for processor development and verification disappeared because, in the 1980s, those processors offered almost no differentiation. Now, the processor field has become highly differentiated again, and the industry is collaborating to develop the necessary tools. One unanswered question is whether they can create open-source tools faster than the EDA industry can provide off-the-shelf products.
Original link:
Are you also “watching”? 👇