I am currently in the progressing of theorising for a strategy game I am planing to make after I finish my current project. Since a player should only be able to see what is within vision of his Units, I would need to make multiple line-of-sight checks. The problem is, that conventional line traces are too expensive, as I would probably need a few hundred per frame, and I would like to have the flexibility to introduce even more Units (In other words line traces dont scale very well). I know only the basics of how a GPU works opposed to a CPU (Manly that it can handle very high numbers of calculations simultaneously, but that they cannot interact with each other). So my question is: Is there a way to utilise the GPUs power? If I understand it correctly, they should be ideal for this kind of calculation, since the individual line traces dont need to interact (I would only need to know which enemy units were found afterwards), and I could have far higher numbers of traces per frame. Is there a way to write code for the GPU and for the CPU and GPU to communicate? As I said, I am probably totally oblivious when it comes to that topic, but I would be happy to learn. While I am also open for other ideas of how to handle vision, I am mainly interested in the GPU/CPU side of the topic. Thank you in advance for any answers.
Any updates regarding this topic?