FFT/IFFT ParaCore Parameters
Parameters used when compiling the Dillon Engineering FFT/IFFT IP Core.
PARAMETER |
TYPE |
DESCRIPTION |
n |
Integer |
Number of points in the DFT (n must be power of 2) |
forward |
Select |
Forward or inverse FFT (assignment options are forward, inverse, or both). |
inputs |
Integer |
Number of data input paths (must be a power of 2) |
outputs |
Integer |
Number of data output paths (must be a power of 2) |
scale |
List |
Used for fixed-point data to selectively scale the data by 1 bit after each rank is executed. Scaling removes errors created by the overflow caused by the complex addition performed in each butterfly structure. |
type |
nType |
The nType definition specifies the fixed or floating-point data used by the FFT (see also the notes below). |
width |
Integer |
Used to define the width of fixed-point data (if floating-point data is selected, than refer to the following table for additional floating-point parameters). |
mag_out |
Boolean |
Produce magnitude output instead of complex results (note that when mag_out is active, the output port width will be reduced by half). |
hanning |
Select |
Used to select the type of Hanning Window (options are to select a formula type). |
|
The following table defines the nType parameter, which is used to define the number types to be implemented by ParaCore Architect:
PARAMETER |
TYPE |
DESCRIPTION |
type |
Select |
Select number type (options are int or float). |
width |
Integer |
For type == int, this parameter sets the integer width. For type == float, this parameter is automatically set to e_width + m_width + 1 (see below). |
e_width |
Integer |
When type == float, this parameter defines the width of the exponent (it should be set to 8 for single precision and 11 for double precision). |
m_width |
Integer |
When type == float, this parameter defines the width of the mantissa (it should be set to 23 for single precision and 52 for double precision). |
|