A microcontroller based cluster computer for computing image convolutions. Designed and implemented to fulfill the final project requirement for "ECE 574: Cluster Computing" at the University of Maine, Orono. Read all about it here: Final project report (quite rushed).

TODO (you're pretty savvy, right? Just look at pico_cluster.png and check out which GPIO pins are used for what in the code eh?)

1. Install the rpi pico sdk somewhere on your computer: pico-sdk

2. Enter a linux environment (WSL)

3.  mkdir build
    cd build
    export PICO_SDK_PATH=../relative/path/to/the/cloned/directory/named/pico-sdk
    cmake ..
    make
   

4. Copy the head_pico.uf2 file to the head node pico

5. Copy the compute_pico.uf2 file to each compute pico

1. Connect the head node to your PC with a microUSB cable.
2. Execute ./client.py with an image to compute a convolution over.

Use the --help command line argument for more in-depth usage. For example, ./client.py assumes the first available serial connection is the head node, but you can specify it with the --port argument.

Documentation related to the project is contained in docs/. docs/Journal.md contains notes and thoughts that I felt like committing to memory during the development of the project. docs/writeup/ contains a copy of the Overleaf project used to compile the final report. In theory I could compile this locally but I haven't bothered to learn to do that yet.