Greenforest I/O

Working inventions that open new ways to compute, communicate, and build.

A serial multiplier accepts the next word without a drain gap. One FPGA pin carries complete radio paths. Cartilage installs and replaces spatial circuit regions. Raw WebGL turns a browser texture into a visible computing machine.

Greenforest I/O develops the mechanism itself—circuit, shader, fabric, RTL, board, or physical process—so an engineer can run it, understand it, and carry it into a product or research program.

Stream the multiplier Open the FPGA systems Run raw WebGL compute Build upward to Cartilage Develop Cartilage Commission hard R&D

Three Machines, Three Scales

A clock-by-clock arithmetic schedule, a GPU-wide state machine, and a recursively reconfigurable fabric all expose the structure that conventional stacks hide.

One FPGA Practice Spans Signal, Protocol, Logic, And Configuration

The same command of clocks, state, and physical pins connects radio, Ethernet, arithmetic, and runtime reconfiguration.

These systems receive and transmit RF, answer ARP in direct RTL, carry FPGA images through UDP, reprogram an adjacent iCE40, turn multiplexers into tangible tiles, and travel from Verilog through place-and-route to configured hardware. Enter the complete FPGA systems map.

Open The Boundary That Controls The Product

Inherited stacks often hide the exact timing, state, routing, or physical constraint that decides whether a new system can exist.

Greenforest I/O turns that hidden boundary into explicit machinery: a circuit schedule, shader pipeline, reconfigurable substrate, radio path, or fabrication process that a team can inspect and extend. This work creates leverage before convention hardens into an assumed law.

Bring The Boundary Everyone Else Treats As Fixed

Strong projects include computation trapped behind a black box, arithmetic shaped incorrectly for its surrounding graph, opaque hardware/software interfaces, FPGA architectures that need a complete physical path, and process ideas that demand reasoning from physics back to working machinery.

Ways to engage: funded R&D, research architecture, product development, implementation, technical due diligence, FPGA/WebGL/semiconductor/radio collaboration, fabrication or laboratory access, and an institutional home for Cartilage and active-device development.

Choose an engagement Open the working systems Send the hard boundary

Research Programs That Reach Beyond Conventional Computing

Company-scale problem map · 210 openings 210 Problems We Have Learned To Call Normal

The map connects three company-sized layers: structurally bounded authority for agents and machines, live reconfigurable physical computation, and physical intelligence built from minimal apparatus. Each layer supports a focused subsidiary and a portfolio of products.

Long-form collection · 53 essays Miscellanies

Fifty-three finished essays develop living circuits, self-reconfiguring machines, software and communication, physical models, mind and society, language, building, learning, and creative tools.

Manufacturing program The Missing Maker Fab

The maker loop already produces boards, enclosures, firmware, and fixtures. The missing machine fabricates active devices with nonlinear gain, restoration, fanout, interconnect, and enough repeatability for useful logic at civic scale.

Universal computing substrate Boolean Algebra Is All That Is Required

The construction uses Boolean state and switching for application logic, ownership, configuration transport, timing, and recursive circuit replacement. MUX Algebra supplies the nested-selector derivation and all sixteen two-input functions.

Transformer training from source Four-Layer Transformer Training Run

The complete run joins architecture, tokenizer, command, loss history, generated samples, and source through more than 50,000 training iterations.

Procedural renderer Cheap Pixelless Textures With 2D SDFs

A self-contained scanline renderer uses UV-space signed-distance material tests and procedural geometry to construct a forest scene without ordinary bitmap texture art.

Build From The Mechanisms

Cartilage in browser and RTL Cartilage Core in WebGL and SystemVerilog

Run one exact 252-bit installation through the compact WebGL1 model and local-clocked SystemVerilog RTL, then follow the architecture paper and Verilator testbench into the mechanism.

Browser GPU lineage WebGL And Computation From First Principles

Create the context, compile the shaders, allocate textures, execute through a framebuffer, and watch state become logic and addition directly on the GPU.

Stateful local-rule machines Cellular Automata Experiments, 2019–2021

Local rules move packed state through reversible routes, conservative dynamics, machine-like flows, and the architectural path that produced Cartilage.

Technical history Programmable Shaders Before The GPU

RenderMan shading leads through real-time shader languages, Brook, CUDA, AlexNet, and mixed-precision matrix engines—a direct technical lineage from programmable color to AI arithmetic.

135 dated technical essays LinkedIn Publication Index

Technical arguments, invention programs, collaboration openings, and original source records form a dated map from 2018 through 2026.

Research Programs Ready For A Lab

Magnetics pursues amplification and circuit behavior beyond semiconductor fabrication. Learning machinery wires gradients directly through primitive operations. Fabrication brings active substrates within reach of civic-scale builders.

The maker-fab program and wafer-diced substrate architecture connect the material path to products. Bring laboratory access, development capital, a customer problem, or an institutional home.