I have been looking into how to implement a raycaster for the F18A GPU since Omega suggested this again the other day. It's definitely something the F18A has the power to do, but as far as I can tell the problem is precision.
My starting point was this tutorial, but I find it difficult to fit the calculations into 16 bit. I have implemented the algorithm in Java, and it's working fine with floating point numbers, but when try to change the calculations to use fixed point 16-bit numbers (with 32-bit support for mul and div) I'm getting overflows all the time.
I also have the source code for the Wolfie3D raycaster that The Mole made in c. This is working with angles instead of vectors which might be better when you don't have so many bits available. But even though the code is commented I'm not sure how the code is working. Is it something you made yourself is it based on some other work that might have a tutorial?
I also tried to make my own algorithm, but, of course, I'm running into all the issues that other people have solved before.
So, do you have good references to 8/16-bit code for raycasting?