The floating point unit (fpu64_rx4rsqrt) is processing unit which allows arithmetic operations on floating point numbers. The fpu64_rx4rsqrt complies fully with the IEEE 754 Standard 1.
* Reciprocal square root (RSQRT) * Square-root (SQRT) * Reciprocal (RECP) * Divide (DIV)
Data type and formats :
* Single precision * Double precision * Demoralized number * Zero value * Infinity * Not a Number (Nan's)
* Signaling Nan's * Infinity / Infinity * Zero / Zero * Infinity x Zero * Invalid Square Root * Zero Divide Exception * Overflow Exception * Underflow Exception
Rounding modes are supported:
* Round to nearest even * Round to zero * Round up * Round down
Number of clock cycles for each operation in formats double precision and single precision
* RSQRT 27 13 * SQRT 29 14 * RSCP 30 15 * DIV 30 15
The project is implemented in the Verilog language and has been tested on more than 4 million (2**22) test vectors for each operation.