Hi everyone, I am currently using cordic for arctangent in my system. I need at least 12 bits of accuracy. My x,y inputs are less than |1| in the form of 2.16. I have made some optimizations to another block in my system and it turns out that I have 6 extra slots of 32x32 multiplier resource but usage of 4 would be better (because of scheduling constraints). I don't have any dividers of course.
Are there any algorithms which would need very few/small roms (which can be implemented in combinational logic) and 4 or so multipliers (and some adder/subtracters) ? Again I can't afford any dividers. Any suggestions are welcome.