Open Silicon · Open Software · On-Device AI

The Parallel
Compute Bridge.

VectoNexus is an open-source hardware and software ecosystem built to unify low-level silicon architectures with high-performance operational stacks — a research-oriented team bridging the gap between how a chip thinks and how software runs on it.

Explore Projects Read the Manifesto

100%

Open Source

2 → 1

Hardware + Software, Unified

MIT

Licensed by Default

Built in the open, on

RISC‑V Linux Kernel Yocto Project On-Device AI TinyML

The Name, Decoded

Why VectoNexus works.

To low-level hardware engineers and system software developers, the name explains our core mechanics instantly.

Vecto — the software logic

Short for Vector. In hardware, vector processing handles parallel math instructions — the crucial engine behind modern CPUs and NPUs. In software, vectors dictate data memory paths and computational direction.

Nexus — the core connection

The central meeting point where multiple independent systems join together — where silicon, firmware, compilers, and applications stop being separate worlds.

The narrative

The name literally translates to “The Central Hub for Parallel Compute.” It positions VectoNexus as the core bridge where software logic seamlessly matches physical hardware channels.

What We Build On

Three pillars. One stack.

Every project we ship stands on open access, from the transistor to the model weights.

Open Hardware

Datapaths, ISAs, and reference designs published down to the schematic and RTL — no black-boxed silicon.

Open Software

Toolchains, compilers, and system software released under permissive licenses so anyone can audit, fork, and build.

On-Device AI

Inference that runs at the edge — close to silicon, without shipping raw data off the device.

Featured Build

ARM single-cycle datapath, visualized.

View Repository

Computer Organization ARM ISA Datapath Simulation

A from-scratch, cycle-accurate visualization of the classic ARM single-cycle datapath.

Every instruction traced through fetch, decode, and execute — control signals, ALU operations, register file reads/writes, and memory access rendered as a living diagram instead of a static textbook figure. Built to make computer organization tangible for anyone learning how a CPU actually moves data.

Instruction-by-instruction control signal tracing
Live ALU, register file, and memory-path visualization
Fully open source — built for classrooms and self-learners alike

Open on GitHub

datapath.trace

// cycle 04 — instr: ADD R3, R1, R2
IF  → PC[0x0010] → IMEM.fetch()
ID  → opcode ADD, Rd=R3, Rn=R1, Rm=R2
EX  → ALU.compute(R1, R2, "ADD")
       ctrl.ALUSrc  = 0
       ctrl.RegWrite = 1
MEM → bypass (no memory op)
WB  → RF[R3] <= ALU.result

// next: PC <= PC + 4

Bridge software logic to physical silicon — with us.

We're a research-oriented team building in the open. If you care about open hardware, open toolchains, or on-device intelligence, there's a bridge for you here.

Contribute on GitHub Get in Touch