https://github.com/zenith391/numworks-rasterizer

Fast rasterizer for the NumWorks calculator
https://github.com/zenith391/numworks-rasterizer

numworks

Last synced: about 1 year ago
JSON representation

Fast rasterizer for the NumWorks calculator

• Host: GitHub
• URL: https://github.com/zenith391/numworks-rasterizer
• Owner: zenith391
• License: gpl-3.0
• Created: 2022-10-16T13:41:29.000Z (over 3 years ago)
• Default Branch: master
• Last Pushed: 2025-02-17T15:05:29.000Z (over 1 year ago)
• Last Synced: 2025-02-17T15:23:08.535Z (over 1 year ago)
• Topics: numworks
• Language: Zig
• Homepage:
• Size: 86.9 KB
• Stars: 3
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# NumWorks Rasterizer

A simple rasterizer for NumWorks calculators, made to test the feasability of running 3D graphics

on this hardware.

## Design Decisions

- A fixed-point number implementation has been added, as it has a much higher performance than

  floating-point numbers on the CPU (Cortex M7). Using a lousy benchmark quickly tested on the

  calculator, fixed point computations are about 10 to 15 times faster than floating point.

  An excellent ressource I like discussing floating point is http://x86asm.net/articles/fixed-point-arithmetic-and-tricks/

- All vector operations use fixed point numbers, including matrix multiplication

- Linear interpolation is used for triangles, even if that's incorrect (and leaves to the warping

  walls effect like in PS1 games) as it's much better for performance

## Running on your calculator

Make sure you have NodeJS installed first

1. Connect your calculator

2. You only need to execute

```sh

zig build run -Doptimize=ReleaseFast

```

(`-Doptimize=ReleaseFast` is recommended even in debug for the app to run at reasonable speed)