What is aijuboard?

aijuboard is a computer built around the Xilinx Zynq-7000 combined system-on-chip and FPGA. This campaign is for collecting funds to build a small production run of 10 units. Contributions of any amount will be much appreciated!

The idea behind it was to create a computer specifically for running Plan 9. The FPGA is particularly useful for that as it allows us to create hardware for Plan 9 essentially from scratch, for instance a "GPU" that implements precisely the Plan 9 graphics operations and takes that load off the CPU.

Of course this does not mean you can not run other operating systems (a Linux port would be particularly easy as Linux has been ported to other boards with the same CPU). Ethernet, SD card and the serial line are particularly easy to program for and make the board excellent for developing your own or porting other operating systems.

It is completely Open-Source hardware: all the board design files and all the source code (both Verilog and C) are available under either the MIT license or the MIT-like Lucent Public license. None of the chips require a non-disclosure agreement to get a copy of the datasheet and no proprietary firmware blobs are required.

What is an FPGA and what is it good for?

An FPGA (field-programmable gate array) is essentially a block of digital hardware that can be reconfigured to do almost any digital function -- limited only by size and speed. As an example we have implemented the Nintendo Entertainment System using about 10% of the FPGA's resources -- meaning you could have maybe 8 running in parallel, all without loading the CPU. By soldering a few wires we could even connect a Super Nintendo controller straight to expansion port.

The combination of the FPGA with a CPU running Plan 9 is especially powerful. With minimal effort you can combine existing Plan 9 tools and your own C code with your FPGA hardware. Plan 9 makes files and resources from other computers readily available which really comes in handy sometimes.

Specs

• 666 MHz Dual-Core ARM Cortex-A9 CPU
• latest generation Xilinx 7 Series FPGA with 46200 LUTs -- equivalent to approximately 1 million gates
• 1 GB of RAM (can be configured as 512 MB of error-corrected RAM instead)
• 2x USB 2.0
• true Gigabit Ethernet (not a USB-to-Ethernet converter)
• SD card slot (supports SD mode at 25 MB/s)
• dual-link Displayport (supports up to about 1080p@60 Hz)
• SATA 1.0 (150 MB/s)
• built-in USB-serial and USB-JTAG converter for debugging
• 26 pin expansion port hooked directly to the FPGA -- can be used for whatever you want! It will connect to nearly anything that speaks 3.3V logic levels.

The Rev 1.1 of the board that will actually go into the production may have some extra minor features. Audio support is one that I particularly have in mind, but I'm open for suggestions.

FAQ

What will the money be used for? The money covers the cost of manufacture of the boards.

What work remains to be done? There are some changes to the schematics and layout that need to be done. A manufacturer has been selected and assembly data will have to be produced in a format the manufacturer is happy with. Software and documentation needs to be written.

When are you planning to deliver? We hope to deliver by September 2015.

Will there be more boards beyond the ten offered currently? Depending on how much money is raised we will increase the number of boards manufactured which will be offered either as perks here or will be available for sale later.

Where will the boards be manufactured? The boards will be manufactured by a prototype fab in the USA.

About us

aijuboard is the work of mainly two people: Julius Schmidt (aka aiju), a physics student at the University of Cambridge, designed the hardware and wrote parts of the software. Friedemann Gerold (aka cinap_lenrek), a professional software developer and the main developer of 9front, wrote the software (and also came up with the name).

Risks & Challenges

Two prototypes have already been built and are so far fully functional (SATA has not been tested as the code for that is to be written). However there needs to be a slight redesign of the board to fix some issues with the Rev 1.0. As the changes are mostly minor it seems unlikely that they should lead to new issues, however we can not completely exclude the possibility. There is of course always the risk that a larger production run uncovers issues that were not found during prototyping. We are very confident we can succeed with this but you do need to keep in mind that you are supporting a research project and not buying an off-the-shelf product.