Just pretending some chick wrote a page for my ai algorythm (but its speculation)->
its a method based synomyc memory, an algoythm to make 2 things 1 thing in different styles throughout the total process, and the style changes through out it, its in a fixed anatomy!, it could be shown as a filter ordering of screens, start to finish, raw sensor to motor, with a few utility loops and branches to diagnose it, reconstruct and probe it,its NOT a greater filter, it has parts, it is brittle, its a lesser filter, theres no way in the world it would never pass the turing test properly, apart from fooling a few inconspicuously, without being told what was driving it, which would give it away completely, but thats not to mean it wasn't a good piece of artificial intelligence, the way it works of raw natural sensors was a huge effort for it. (computer vision measurement, which throws people into fits of impossibility, and is actually turning the real world into a virtualized version of it, and you really guage the face like hitlers book on how to spot a jew with your ruler.) The older engineers didnt have the huge parallel throughput we have now doing everything image based, my best raytracing algorythm is image based, and handles fractals with gi in real time, and its image based. If you said you wanted a ray per pixel in realtime back in '92 they would have said you were dreaming too much, but just saying that wolfenstien was a 1d raytracer, you could handle 1 line back then, so now we go mesh that squared over and over sequentially every day with a consumer gpu in '15. its different now, but theres no end to the amount of guys still running slow easy to write threadless code, but we eat em for breakfast.
But to help the little dummies, far beyond my time the APU came out, and its supposed to be coded on one thread and its multithreaded already??? I dont know about that, but its something like that. managed. But I have to say something weak and say I dont use CUDA, I just use direct compute (which is the girls version i guess), and its a managed gpu language. (As in taking 100 years to compile optimizations for you.)
So good luck compiling your APU code. hehe Actually dealing with a girly compiler, you have to make sure you only compile the shaders you changed, then it runs smoothly, your production process. Same goes for these APU compilers I guess.
