(no subject)
Yep, cleaned up the equations some more, it's a pretty simple process so I can do a filter of an arbitrary order, and it only takes about 5 more floating point operations per order...
So lets see... each 'iteration' used to take 512 floating point operations... now it takes 7... each set of iterations (16) used to take 8192 floating point operations, now it takes 121.. as in, the program is now 60+ times as fast and perfectly accurate to the limits of 32-bit (or 64-bit) floats.
For a typical 1-second frame, I needed to do 722 million floating point operations... now I have to do 11 million.
I'll say it again, THANK GOD FOR EULER.
So lets see... each 'iteration' used to take 512 floating point operations... now it takes 7... each set of iterations (16) used to take 8192 floating point operations, now it takes 121.. as in, the program is now 60+ times as fast and perfectly accurate to the limits of 32-bit (or 64-bit) floats.
For a typical 1-second frame, I needed to do 722 million floating point operations... now I have to do 11 million.
I'll say it again, THANK GOD FOR EULER.