A parameterizable floating point operations (written in Verilog)
Implemented operations:
- Multiplication
- Addition
- Conversion (real-to-integer)
- Conversion (integer-to-real)
- Absolute Value
- Comparison
- Complex Multiplication
- Complex Addition
- Complex Conjugation
NUM_W
- Number WidthEXP_W
- Exponent WidthMAN_W
- Mantissa WidthTRIGOUT
- Output Trigger (1 - enable, 0 - disable)
- FP16 :
NUM_W
= 16,EXP_W
= 5,MAN_W
= 10 (Half-precision) - FP32 :
NUM_W
= 32,EXP_W
= 8,MAN_W
= 23 (Single-precision) - FP64 :
NUM_W
= 64,EXP_W
= 11,MAN_W
= 52 (Double-precision) - FP128:
NUM_W
= 128,EXP_W
= 15,MAN_W
= 112 (Quadruple-precision) - FP256:
NUM_W
= 256,EXP_W
= 19,MAN_W
= 236 (Octuple-precision)