I am working on PCA.My entire design is in IEEE754 (32bit) formate.I need sin and cos function.My angle is in IEEE754 format. I need an IP core cordic algorithm in IEEE754 formate(VERILOG CODE).

Why on earth would you do a design in floating point like that?

• Floating point is expensive, and IEEE754 floating point even more so. It will use way more logic than necessary, possibly forcing you to purchase a more expensive FPGA
• A 32-bit fixed point value will provide better results than an IEEE754 floating point result
• The "CORDIC" algorithm you are asking about is a fixed point algorithm in the first place. You'd still have to convert it to floating point.

If you are at all interested, you might find this blog article useful. It describes one of several sine wave core-generation techniques used by my sine wave generation repository. They are all fixed point based implementations however.

I'd like to post again about how much better things could be with only two multiplies added to the algorithm, but for now I've only posted on how a CORDIC works to generate a sine/cosine, how it can be used to do rectangle to polar conversion, and how to build one of two rather simple table based lookups. This becomes important since, among other things, CORDIC algorithms get expensive for any significant bit precision, and with just two multiplies you can do much better for much less logic.

Dan