Reproducible SoC
Choose a friendly FPGA board, synthesize an open RISC-V system, and make one-command builds boringly reliable.
3DVR Compute · RISC-V
Project seed · FPGA firstBuild a computer we understand.
A small, fully open RISC-V SoC project designed to be understandable end-to-end, reproducible by other builders, and useful long before we ever manufacture silicon.
Start on an FPGA. Make every block inspectable and replaceable.
Boot Linux first. Optimize later.
Small firstOne working system beats a giant unfinished architecture.
Open end-to-endRTL, tooling, software, docs, and reproducible builds.
FPGA-firstProve the design on hardware before thinking about silicon.
Linux-readyBuild toward a real Debian/Linux computer, not a demo core.
First system
The smallest useful open computer.
Keep the architecture boring enough to understand, then make each layer better in public.
RISC-V CPUA simple existing open core first; custom work only where it earns its complexity.
Memory + interconnectOpen, documented paths between CPU, RAM, storage, and peripherals.
Accelerator pathA deliberately simple compute block we can eventually expose to tinygrad.
LinuxBootloader, kernel, Debian userspace, and reproducible setup documentation.
Roadmap
Make it real in layers.
The project should remain useful at every stage instead of waiting years for a custom chip.
Boot a real OS
Bring up serial, storage, networking, and enough Linux support to use the machine for actual computing.
Tinygrad-friendly acceleration
Experiment with a small documented accelerator interface instead of depending forever on proprietary GPU stacks.
Where it leads
From FPGA to personal computer.
This project can eventually become an open compute module for 3DVR hardware. The near-term goal is learning and reproducibility; the long-term goal is a stack we could confidently put inside a Balthazar-style laptop.
FPGAUnderstand it.
BoardUse it.
ModuleIntegrate it.
SiliconEarn it.
Open hardware
Build one layer.
RTL, FPGA bring-up, Linux, compilers, tinygrad, boards, documentation—small contributions are enough to move the project forward.