FFT Dynamic Point - 2022.1 English

Vitis Model Composer User Guide (UG1483)

Document ID
Release Date
2022.1 English


AI Engine/DSP/Window IO


FFT Dynamic Point implementation targeted for AI Engines. This block supports performing different lengths of FFT on different input data windows. Each window must be preceded by a 256-bit header which describes run-time parameters of that window. The format of the header is described below.

Figure 1. Header Format
Table 1. Header Format
Index Field Name Description
1 Direction
  • 0 = Inverse FFT
  • 1= Forward FFT
2 Point size (radix 2 stages) Point size described as power of 2. For example, value '5' describes a point size of '32'
3 Reserved Reserved
4 Status (output only)
  • 0 = Legal point size
  • 1 = Illegal point size

Example Header Format

To implement the 32-point size FFT of 'cint16' datatype, the format of the header that should be preceded to the input data window should be as shown.

Example (Input to Dynamic Point FFT)

To implement the FFT of different sizes, say, 64, 32 and 128, of type cint32, the format of the input should be as follows.

[1 complex(6,0) 0 0 ones(1,64) 1 complex(5,0) 0 0 ones(1,32) 1 7 0 0 ones(1,128)]  


Input/Output Data Type:
  • Describes the type of individual data samples input/output of the dynamic point FFT. It can be cint16, cint32, cfloat types.
FFT Maximum Size:
  • Specifies the maximum FFT size that is supported by Dynamic point FFT. You can perform different lengths of FFT on different input data windows. It must be a power of 2 with a minimum value of 16. The maximum value supported by the library element is 65536.
Input Window Size:
  • Specifies the number of samples in the input window excluding the header. The value must be in the range 8 to 1024 and the default value is 64.
Scale Output down by 2^:
  • Describes the power of 2 shift down applied to the accumulation of FIR terms before output. It must be in the range 0 to 61.
Target Output Throughput (MSPS):
  • Specifies the output sampling rate of the DDS function in Mega Samples per Second (MSPS). The value must be in the range 1 to 1000 and the default value is 200.
Specify the Number of Cascade Stages:
  • When this option is not enabled, the tool will determine the FFT configuration that best achieves the specified input sampling rate. When the option is enabled, the Number of cascade stages can be specified (which describes the number of AI Engine processors to split the operation over). This allows resources to be traded for higher performance, but the specified input sampling rate constraint may not be achieved. The value must be in the range of 1 to 9.