15-07-2022 | | By Robin Mitchell
In a surprising report, RISC-V International reported that over 10 billion RISC-V cores have been delivered to market solutions, and yet it would seem that RISC-V processors are a rare find. What does the report claim, where are these RISC-V processors, and does this report suggest that RISC-V will become a major platform for future devices?
In a recent report, the CEO of RISC-V International announced that over 10 billion RISC-V cores have been delivered to the market. Despite the architecture having only been officially announced 12 years ago, RISC-V has quickly garnered media attention due to its open-source nature that allows engineers to create their own implementations of RISC-V with no need for licensing or royalties.
The report also mentions how RISC-V has been introduced into various industries, including embedded designs, wearables, AI, data centres, HPS, automotive, and IoT. Additionally, it also predicts that by 2027 the number of RISC-V devices globally will reach 25 billion. Considering the young nature of RISC-V, it also mentions that now is a prime time for engineers to learn the new architecture and start gaining experience ready for a potential future revolution.
At Electropages, we have been reporting the growth of RISC-V for some time now, and the past year has seen some breakthrough reports in the field of RISC-V. For example, it was not that long ago that SiFive announced the world’s first RISC-V PC platform or that Alibaba reported that it was close to creating a RISC-V version of Android.
All of these achievements are essential for RISC-V to become a major player, but are there really 10 billion RISC-V devices already out there? If that is true, how is that even possible when globally, 25.6 billion microcontroller devices (from all companies and architectures) are sold each year? And if it is true, where are all of these RISC-V devices?
While there is no hard data on the true number of RISC-V devices currently in use, it would appear that the report could very well be accurate, but it is crucial to understand what it means when the term “RISC-V device” actually means. RISC-V itself isn’t a physical product but instead a description of an instruction architecture. Two RISC-V processors could be entirely different at the transistor level, but so long as both can execute the same binary instructions and produce the same result, then both are considered to be RICS-V compatible. As such, it is possible to build a RISC-V processor on the latest silicon technology or from discrete valves taking up an entire warehouse; either design would be considered a RISC-V processor.
As such, a RISC-V processor could be implemented as a virtual machine running on top of another pre-existing solution, or it could be running on an FPGA as a soft-core that can be upgraded over time. Either way, 10 billion RISC-V cores may exist, but they certainly are not all dedicated silicon devices.
RISC-V microcontrollers do exist, and many of these are found on the Chinese market, with two famous examples being the ESP32-C3 which has a single-core RISC-V processor and the GD32V device range from GigaDevice (they are well known for their range of STM32 compatible processors). SiFive has also seen popularity through its development of RISC-V-based SoCs for use in PC applications, and Google has developed their Titan M2 security module that also uses a RISC-V core.
But even though RISC-V processors are becoming available, they are nowhere near as established as mainstream devices such as ARM-based microcontrollers. Looking at most IC distributors, you would be hard-pressed to find RISC-V devices in mainstream use or readily available. Thus, it is highly likely that many RISC-V cores are being used in closed-source projects that are not available on the open market. Additionally, those that are available may be more popular in countries where access to key technologies is becoming increasingly challenging (such as China).
It’s hard to believe that there are 10 billion RISC-V devices currently in use, but even if this figure is off by two orders of magnitude, that is still a hell of a lot of RISC-V devices. If RISC-V is being used as a soft-core or as a proprietary solution, then it may turn out that the future of RISC-V is not in off-the-shelf solutions but as a self-contained processor for specific applications that need a secondary processor not tied to royalties or licenses.
Of course, off-the-shelf RISC-V devices will continue to be developed, but engineers choosing these devices over established solutions would have to ask themselves what advantage they are getting. RISC-V microcontrollers are unlikely to be significantly cheaper than an ARM equivalent, and ARM architecture has immense software support along with support from ARM itself. RISC-V would entirely depend on the wider community for new libraries, and those who develop RISC-V products would likely keep them closed-source.
If RISC-V can demonstrate itself as a viable platform that can compete with ARM and x86, then it stands a very good chance at dominating the field of processor design. If all processors are capable of running the same RISC-V code, then it could enable a new generation of cross-platform code and allow engineers to develop software on more unified platforms.