This repo is a revamp of an old proof of concept that I wrote in 2013 to answer the question "How much text can you render in a web browser at once?", for which the answer is "At least the entire Linux kernel, or as much text as will fit in your GPU's ram".
Wideboard uses WebGL 1, some special shaders, and some simple GPU-compatible data structures to render arbitrarily large amounts of monospace, unformatted text in a web browser. It has been tested with 56000 files from the Linux repo (31 million lines of source, or around 1.2 gigabytes of raw text) and runs at 60 fps while using single-digit percentages of a 3080ti GPU. It should run equally well on integrated GPUs, barring memory limits.
To launch Wideboard, run "python3 -m http.server" in in the root of this repo and then go to http://localhost:8000. By default it renders this README.md message and all files with extensions (h|hpp|c|cc|cpp|sh|js|ts|txt|md) under /docs.
You can drag and mousewheel-zoom the view ala Google Maps and it should render at 60 frames per second even when displaying tens of millions of lines of text on a weak integrated GPU. Arrow keys and pageup/pagedown scroll the view, escape resets the view.
To render the entire Linux kernel source tree in your browser, check out https://github.com/torvalds/linux under /docs and refresh the page. This will use a lot of GPU ram (something about Chrome is bloating GPU memory usage by ~4x) and there's no real error checking if you run out, so proceed with caution.
NOTE - If you're running the server under WSL, loading will take ages as WSL seems to add ~300 milliseconds of delay per HTTP connection. Running directly under Windows or on Linux will load files 30x faster.
Above this message you'll see three squares - the left one contains lines of ASCII text packed into a 4096x4096 texture, the middle one contains one texel per line of source text that encodes a 24-bit pointer into the ASCII text and an 8-bit line length, the right square is the font used (Terminus). Wideboard will create more textures as needed to hold text, up to the capacity of your GPU.
