TODO.md

Bugfix

• Particles going out of bounds when window size is not 1000x1000. lost like 2h in this, it was the shader, the last place I expected to see this
• Particles are all white when using the current settings in main

Recorder

• Put new sandbox dimensions etc into recorder
• Need to cache things in memory and not keep getting from file somehow. fix
• Recorder is forgetting either accel or old_pos in the transition. fix

Rendering

• Change shader to support height != width

Simulation

• Finish multithreading
• Add SIMD
• Add a system where an error in adding to grid draws the current simulation, making offender particles red
• Cleanup after closing window, and prevent setting up the renderer multiple times
• Make mipmaps(????????????????)
• Check if transposing the grid would be better for cache locality
• Redo collisions to get better cache locality
• Test ridiculous overkill compile-timing everything, like the grid size
• Use different threadpool, it would be better for cache if thread N always worked on slice N Think about collision optimizations:
• Threads only need to wait for the fisrst row of the thread above them
• Collisions can maybe propagate up and never look below them, meaning update() could be done right after the collisions of each slice