By Zhangxi Tan, Lin Zhang, Yi Li, and David Patterson of the RISC-V International Open Source Laboratory Tsinghua-Berkeley Shenzhen Institute
RISC-V, the royalty-free open-source alternative to proprietary instruction sets, is growing globally in popularity. This has led to many free and open-source processor designs, a free and open software stack and an international organization that maintains and promotes the instruction set. However, while key pieces of an open-source ecosystem for RISC-V already exist, many parts are missing to match the completeness of ecosystems for the older, proprietary instruction sets.
Enter…the PicoRioTM open-source project. Stewarded by the RISC-V International Open Source Laboratory (also known as RIOS Lab) – a nonprofit research lab at Tsinghua-Berkeley Shenzhen Institute (TBSI), the PicoRio will be a RISC-V-based small-board computer available at an affordable price point.
The PicoRio will aim to offer the following features and benefits:
- Independently maintained: The RIOS Lab will be the solo non-profit organization that governs the architecture development, ensures compliance and will publish the design. It will also be the gatekeeper for both hardware and software, from software on a chip (SoC) and firmware/drivers to high-level software and documentation. The PicoRio will be vendor-agnostic and not proprietary, and the RIOS Lab will work with academic and commercial organizations that will commit to its expansion and volume manufacturing.
- Open source: Unlike Raspberry Pi, which uses proprietary Broadcom SoCs, PicoRio will open source as many components as possible, including the central processing unit (CPU) and main SoC design, chip package and board design files, device drivers and firmware. The exceptions are foundry-related IPs (e.g., TSMC SRAM configurations), commercial high-speed interfaces and complex commercial IP blocks like GPU. Nevertheless, our goal is to reduce the commercial closed-source IPs for each successive release of PicoRio, with the long-term goal of having a version that is as open as is practical.
- High-quality IPs: The openness of PicoRio will not come at the cost of lower-quality IP blocks. A major goal of the RIOS lab is to develop open-source, industrial-strength hardware IP to help the RISC-V ecosystem catch up with those of the older, proprietary ISAs. Thus, PicoRio aims at a high-quality silicon release using open-source IP. The IP blocks will have gone through rigorous real tapeout verifications that meet industry quality. Also, we will open-source our verification process, to help improve transparency and trustworthiness.
- Low power and low cost: The target metrics of PicoRio are long battery life and low cost, rather than offering high-performance and requiring large amounts of memory, making it a better match for RISC-V. In contrast, Raspberry Pi uses power-hungry ARM processors. For example, the idle power consumption in the latest version of Raspberry Pi has risen from 0.4 Watts to 2.7 Watts².
PicoRio three-phase development
To give focus to such an effort, it helps to have concrete targets. We aim to release a new PicoRio version every year and we will divide the development of PicoRio into three phases:
First Phase (PicoRio 1.0):
Second Phase (PicoRio 2.0):
In addition to improving the v1.0 hardware, RIOS Lab is partnering with Imagination Technologies to integrate a complete multimedia pipeline alongside our open 64-bit RISC-V cores to provide a modern, affordable Linux-based PC-like environment. This will include Imagination’s PowerVR GE7800 XE series GPU to enable developers to run graphics-intensive applications (such as web browsers).
Imagination’s efficient PowerVR GPUs
The PowerVR Series7XE Series, which includes the GE7800, is a range of ultra-efficient, silicon-proven, highly-featured GPUs that have been designed specifically to deliver the right mix of performance, features and silicon size for the most demanding and cost-sensitive markets.
Imagination’s GPUs are designed to maximize performance in a given power envelope. Their tile-based deferred rendering architecture significantly reduces system-memory bandwidth requirements, which in turn increases performance and reduces power. It does this by capturing the whole scene before rendering, so occluded pixels can be identified and rejected before they are processed. The hardware also splits up the geometry data into small rectangular regions (tiles) that are processed as one image. Each tile is rasterized and processed separately, and as the size of the render is so small, this allows all data to be kept on the chip.
Imagination embraces the RISC-V ecosystem
Imagination’s PowerVR GPU Driver (DDK v1.13 release onwards) supports RISC-V as a target application processor and will be easily integrated with the PicoRio single-board computer platform.
Third Phase (PicoRio 3.0):
Building upon the v2.0 hardware, we plan to further improve the CPU performance to bring PicoRio to the level of a pad computer/laptop.
An example of what could be done with PicoRio 3.0 is a RISC-V-based pad computer/laptop, where all application software and storage runs in the Cloud (like a Chromebook). Such devices are much more popular than conventional laptops in K-12 schools because they are cheaper to purchase, and are easier to maintain due to no system administration being required. Another benefit is that since most of computing and storage is on the Cloud, it could use a slower, cheaper processor, require less storage and would need much less software to be ported to it.
We look forward to the arrival of the PicoRio and believe it will be a major boost for the RISC-V ecosystem. However, it will only be the beginning. After all, there is no “end product” when it comes to open source – by its very nature, it will constantly evolve as different users and developers collaborate from all over the world.
Zhangxi Tan is the Co-Director of the RIOS Lab, an Adjunct Professor at TBSI, and the Founder and President of RiVAI Technologies Co. LTD. Prior to these roles, Dr. Tan joined Pure Storage as the company’s first chip designer, and successfully guided the delivery of FlashBladeTM. Dr. Tan holds more than 20 US patents, and he is also the inventor of the FPGA-based architecture simulator (RAMP Gold). Dr. Tan earned Bachelor and Master degrees in electrical engineering and computer science from Tsinghua University; Master and Ph.D. degrees in computer science from the University of California, Berkeley, where he was supervised by Prof. David Patterson.
Lin Zhang is the executive director of RIOS Lab, a professor and doctoral supervisor of Tsinghua University. His research interests include communication networks, wireless sensor networks, sensor data mining, and artificial intelligence systems. In recent years, he made several breakthroughs in the implementation and development processes of many National Natural Science Foundation projects, “863” Projects and major national engineering, and has played important roles in the Olympic Game related projects as well as the promotion of smart city informatization. He has served as Chairman of the Graduate School Student Association of Tsinghua University, Executive Chairman of the National Association of Students, Deputy Director of the Department of Electronic Engineering of Tsinghua University, and has been awarded the “Excellent Teachers” title in Tsinghua University Graduate School. He is a member of the Chinese Institute of Electronics (Information Theory Branch) and a senior member of the IEEE. He has published more than 100 academic papers in journals and conferences in China and at abroad and has won the IEEE / ACM Sensys 2010, 2016, CASE 2013, IPSN 2014, and IEEE PES 2017 Best Paper / Technical Presentation Award.
Yi Li is the assistant director of RIOS Lab. She was leading strategic initiatives for Amazon AI Services. Launched Amazon Prime Photos and API-driven service Amazon Rekognition for image recognition and video recognition at AWS re: Invent 2016&2017.Before Amazon, Li has performed as Chief Executive Officer (CEO) for Orbeus Inc (acquired by Amazon). As the first Chinese team funded by the top U.S. incubator (TechStar), Orbeus tapped the power of deep learning to provide scalable image and face recognition solutions for businesses and consumers. With expertise in state-of-the-art machine learning and ImageNet (award winners in 2013 and 2014), Orbeus developed ReKognition API (a cutting-edge facial and image recognition platform), and managed the invention of PhotoTime (a mobile application for auto-tagging and search photos). Orbeus was acquired by Amazon in 2015.