Le Potato High Performance Low Power Single Board Computer in Raspberry Pi 3 Form Factor, 2GB RAM, HDMI 2.0, eMMC, Latest Android 9 Nougat, and Linux 5.19.
A prototype is a preliminary model of something. Projects that offer physical products need to show backers documentation of a working prototype. This gallery features photos, videos, and other visual documentation that will give backers a sense of what’s been accomplished so far and what’s left to do. Though the development process can vary for each project, these are the stages we typically see:
Proof of Concept
Explorations that test ideas and functionality.
Functional Prototype
Demonstrates the functionality of the final product, but looks different.
Appearance Prototype
Looks like the final product, but is not functional.
Design Prototype
Appearance and function match the final product, but is made with different manufacturing methods.
Production Prototype
Appearance, function, and manufacturing methods match the final product.
Prototype Gallery
These photos and videos provide a detailed look at this project’s development.
Our first Libre Computer Board, code name Le Potato, is designed as a drop in hardware replacement for the Raspberry Pi 3 Model B and offers faster performance, more memory, lower power, higher IO throughput, 4K capabilities, open market components, improved media acceleration, removal of the vendor locked-in interfaces, and Android 7.1 support. This platform uses the latest technologies and is built upon proven long term available chips. It is supported by upstream Linux and has a downstream development package based on Linux 4.9 LTS that offers ready-to-go 4K media decoding, 3D acceleration, and more.
It can be used to tinker with electronics, teach programming, build media centers, create digital signage solutions, play retro games, establish bi-directional video, and unlock imaginations. It is available in 1GB and 2GB configurations while utilizing a large existing collaborative ecosystem of parts for creators to build new and exciting products and services.
Why is this project important?
We coined the term "extensible embedded computing space" or EECS (or Electrical Engineering Computer Science?) to describe a market that is under-developed in proportion to its size and age. Professionals, enthusiasts, educators, students, and hobbyists have been creating alternative world-changing ideas utilizing inexpensive extensible computing infrastructure.
We want this computing infrastructure to be open so that anyone can turn an idea into a product or service. We feel that the traditional barriers to entry for hardware and software, both in terms of cost and time, should be de-duplicated and scaled. We created this initiative called the Libre Computer Project to further this scalability via promoting and funding open hardware and software libre.
In terms of Linux and Android, we should have equivalent or better upstream and downstream support as Raspberry Pi. Linux operating systems distros will definitely favor the Raspberry Pi although this board has a greater lead on Android and can easily be supported by existing distros. Please read the software section further down for more information.
What about the proprietary ARM Mali-450 GPU?
The ARM Mali-4xx series of 3D accelerators is among the most popular 3D GPU IP due to its age, cost, and software support, albeit proprietary in nature. ARM inherited most of the design from its Norwegian acquisition of Falanx Microsystems but never open sourced the software IP. The 3D GPU utilizes compiled binary firmware tailored to specific SoCs. The GPU does not feature a 2D display engine and the binary must be tied into different display engines via customizations from the SoC vendor. The firmware connects to a passthrough kernel driver and this is the mechanism of 3D enablement for our board in our operating system images.
Initiatives to develop an open source driver never achieve the critical mass to provide resources to drive adoption by the community. Due to the number of deployments with this 3D GPU series, we know that having this would open new possibilities. We really want this to happen and will contribute resources when presented the opportunity to affect meaningful progress on this front.
It is important to make the distinction between software libre and other forms of libre. We are designing open hardware while promoting software libre. The components we use are commodity items and people have the liberty to redesign, customize, improve, and build on top of this platform.
Le Potato will have basic upstream support in Linux 4.13 via the collaboration with BayLibre. BayLibre specializes in embedded Linux development on a variety of architectures and their team played an instrumental role in our ability to get this product to market. We will continue supporting the impressive upstream work they have done for Amlogic solutions and we will push to have the full feature set of the board upstream.
SoC vendors have started to see the benefits of software libre on open source hardware and we hope this board is beginning of long and beneficial cooperation between vendors and the software libre community. While there were real previous attempts (as well as namesake attempts) to varying degree of success, we analyzed where efforts fell short and we have committed investors that share our mission and vision.
Chips manufactured today uses thousands of IP blocks created by hundreds of vendors from the transistor level to entire cores. There are no 100% verifiable designs given the number of parties involved.
We will disclose as much as possible with regards to the design to help facilitate development of this platform and add-ons to this platform. We will release the schematic design after launch. If you need a custom solution, we can help with that too.
Software and images made for Raspberry Pi must be modified to work
The Raspberry Pi series, the single board computers that put maker boards on the map, is starting to show its age. Shared USB/Ethernet bandwidth, lack of MicroSD UHS support, lack of modern codec support, and lack of 4K contribute to a sluggish experience. The CSI and DSI interfaces on the board were proprietary in nature and required too much from Broadcom to build upon. The onboard WiFi and Bluetooth on the Raspberry Pi 3 Model B were convenient but it encumbers the board with certification requirements and numerous performance issues. In many ways, the Raspberry Pi 2 Model B was a cleaner design since it did not have these features. The GPU is moving towards Linux mainline via the efforts of Eric Anholt which feels like an industry first.
This is a very beautiful board and ASUS's aesthetic efforts are top knotch. ASUS TinkerOS is based on Android 6.0 with the Tablet UI. Gigabit Ethernet is great for use as a NAS but you are stuck with a MicroSD card. While UHS support is listed, it has not been enabled in software as far as we could tell. The board has an IPEX connector which allows for better wireless range and performance. It also features an OpenGL ES 3.0 capable GPU, the ARM Mali-T-760.
The out-of-order Cortex-A17 CPU is fast but it is also a legacy design. Not having the ARMv8 instruction set prevents this board from taking advantage of security and performance enhancements made in Linux and Android for ARM's latest ARMv8 ISA. The power consumption is also extremely high resulting in thermal throttling. This prevents the board from achieving its performance potential. It requires an very high quality MicroUSB power supply capable of 2.2A compared with 1.5A for the Raspberry Pi 3 and this board.
The PINE A64+ 2GB is a very different beast since it is about twice the size of the other boards. It was the most successful Kickstarter SBC and had great marketing. The software development end of this board relied more on the community than any other board. The linux-sunxi community provided much of the support for this board.
Will it run existing software?
Le Potato will run both ARMv8, ARMv7, ARMv6 compiled code compiled on Linux. However, it requires a different boot loader and kernel to be installed on the MicroSD card for existing Raspberry Pi images to be bootable. We will provide tooling on Github to do this. If the software on the image is tailored to the features of a specific board, software work will need to be done to utilize the features of this board instead.
Android TV based on Android 7.1 have been running with great stability on the board since this SoC has been in-use on those platforms for almost a year. Linux 4.9 and Linux 3.14 buildroots are available. While basic mainline Linux support will arrive in 4.13, some SoC features are still WIP.
The following will be available at launch:
Android 7.1 and 6.0
Ubuntu 16.04 LTS with Linux 4.9
Debian 9 with Linux 4.9
The following require a few weeks to months to package:
Raspbian
RetroArch
LibreELEC
Debian 8
The rest we will probably leave up to the community.
There will not be a Libre Computer Board OS as is the trend these days. We simply hope that the Debian/Ubuntu/Fedora communities can add packages that hold binaries and scripts for users to bootstrap images with.
The images will include the necessary blob bits for 3D and media. Most of the media decoding will run on top of the GStreamer framework in the Linux 4.9 kernel.
Will it be compatible with existing add-ons?
Le Potato has a 40 pin GPIO header that maintains as much compatibility as possible with the existing standard set forth by the Raspberry Pi 3 Model B. The placement on the header of I2C, SPI, PWM, 5V, 3.3V, and GPIO are equivalent. Raspberry Pi ecosystem has quite a few years of support from numerous vendors and we can only play catch-up at this point. Most Raspberry Pi HATs can be supported by the hardware but the software bits to get them working on this board is simply not there yet.
Why did you not add onboard WiFi or Bluetooth?
The certification process for single board computers with radios is incredibly painful and long. We want to make this product as convenient as possible to procure in any country so we refrained from adding radio components. Additionally, the performance of onboard radios are generally poor while USB dongles are cheap, certified in many countries, and offer better performance.
That being said, we did design the 40 pin GPIO header to include the SDIO wires normally used for WiFi and Bluetooth. We plan to release mezzanines with dual-band WiFi, Bluetooth, and more.
Since the board is working with a small form-factor, the freed PCB space from not having onboard radio has allowed us to expose most of the SoC's capabilities in an unmodified fashion.
What is the status of the project?
The product design, initial software support, and testing have been completed by our internal team. We have already completed a manufacturing run of 100 boards on June 1 for validation and age testing. Our production run of 1000 boards has started and will be completed by the end of July. Your contributions will be used towards the manufacturing and distribution of a larger batch of boards.
The first batch of boards have since been through a battery of stress tests including a week long marathon of ssvb's cpuburn-a53, eMMC write, and 4K H.265 decoding without issue. We are still ironing out software kinks for launch such as GPIOs, LEDs, and MicroSD UHS but we have not had any stability issues so far.
We will be providing the following:
Schematic Diagrams
Buildroot with Linux 3.14 and Linux 4.9 with blob bits
What does the case look like?
We designed a new actively cooled case to go along with the product. It has transparent front for the LEDs and IR receiver, 20mm fan, slit for a 40 pin ribbon cable for expansion, extra space to prevent broken MicroSD cards, bottom ventilation slits, and bottom wall anchors. We will provide the final photography once the first production samples come back.
When will eMMC modules be available?
We are still determining the logistics for end-user flashing the eMMC before we release the modules. These are not flashable via a MicroSD adapter since they are built already for eMMC 5.0. They will come in 8GB, 16GB, 32GB, 64GB, and 128GB. 128GB eMMC may be delayed due to the move towards UFS 2.1.
When will this product go on sale?
We are aiming to have the boards available globally in October after the deliveries of the boards to Kickstarter backers. If we get too many pre-orders, we would have to push back retail launch.
Risks and challenges
We have 50 years of combined experience in the EDM and semiconductor industries. This is a product designed by top professionals in the industry. You will be hard pressed to find equivalent engineering quality in competitive products.
We are working with industry experts such as BayLibre and LoveRPi to bring this product to you with the necessary software, logistics, and accessories.
