Even a 3MB driver is unacceptable to me. Every project you release will need this driver. Why canīt it simply be a dll or lib like all other physics engines? No one owns this physics card flop anyway.
What does "next-gen" mean?
I already said it in the other thread but "next-gen" is nothing but marketing nonsense. It makes absolutely no sense.
I agree that it probably will become better with nVidia. But since CUDA is free, the other engines also will support GPUs sooner or later.
They're integrating it into their PPUs (and, should it become a de facto standard, it'll ultimately work well with other PPUs).
nVidia wonīt build PPUs and it wonīt bake physics into hardware. GPUs already are very fast vector processors which can be used for physics via
CUDA (and of course the CUDA technology will be improved in future generations).
