Methods and apparatus for twiddle factor generation for use with a programmable mixed-radix DFT/IDFT processor
View Patent ↗Twiddle factor generation for use with a programmable mix-radix vector processor (“PVP”) capable of calculating discrete Fourier transform (“DFT/IDFT”) values. In an exemplary embodiment, an apparatus includes look-up table logic that receives twiddle control factors and outputs a selected twiddle factor scaler value (TFSV), a base vector generator that generates a base vector values based on the selected TFSV, and a twiddle column generator that generates a twiddle vector from the base vector.
1. An apparatus, comprising:
look-up table (LUT) logic containing at least one LUT for storing a set of radix twiddle coefficients and configured to receive twiddle control factors, the LUT logic configured to output a selected twiddle factor scaler value (TFSV) in accordance with the twiddle control factors and one or more radix twiddle coefficients retrieved from the LUT;
a base vector generator coupled to the LUT logic and that generates a base vector value based on the selected twiddle factor scaler value; and
a twiddle vector generator coupled to the base vector generator and that generates a twiddle vector in response to the base vector value.
2. The apparatus of claim 1 , wherein the look-up table logic comprises one or more address generation units that generate an address associated with the received twiddle control factors.
3. The apparatus of claim 2 , wherein the look-up table logic comprises one or more look-up tables that generate the selected TFSV based on the address.
4. The apparatus of claim 3 , wherein the look-up table logic comprises a first LUT that stores TFSVs for both radix 3 and radix 6.
5. The apparatus of claim 4 , wherein the look-up table logic comprises a second LUT that stores TFSVs for radix 4 and a third LUT that stores TFSVs for radix 5.
6. The apparatus of claim 3 , wherein the base vector generator comprises a cross multiplier that iterative multiplies the selected TFSV to generate the base vector values.
7. The apparatus of claim 6 , wherein the base vector generator comprises a serial to parallel converter to convert base values to a base vector.
8. The apparatus of claim 6 , wherein the twiddle vector generator comprises a vector cross multiplier that iteratively multiplies the base vector to generate the twiddle vector.
9. The apparatus of claim 8 , wherein the twiddle vector generator comprises a last step coefficient LUT that includes step coefficients to compute twiddle vectors for a last radix stage.
10. The apparatus of claim 1 , wherein the twiddle vector is output to a twiddle multiplication unit in a vector pipeline that carries up to 12 samples per cycle.
11. A method for generating twiddle values in a programmable mixed-radix processor, the method comprising:
receiving twiddle control factors, by look-up table (“LUT”) logic, containing information relating to twiddle factor;
searching an LUT in the LUT logic to identify at least one radix twiddle coefficient in accordance with the twiddle control factors;
obtaining a selected twiddle factor scaler value (TFSV) in response to the radix twiddle coefficient;
generating a base vector value based on the twiddle factor scaler value; and
generating a twiddle vector from the base vector.
12. The method of claim 11 , further comprising generating an address associated with the received twiddle control factors.
13. The method of claim 12 , further comprising generating the selected TFSV based on the address.
14. The method of claim 13 , wherein the generating the selected TFSV value comprises accessing a first look-up table (LUT) that includes TFSVs for both radix 3and radix 6 factorizations.
15. The method of claim 14 , wherein the generating the TFSV value comprises accessing a second LUT that includes TFSVs for radix 4 factorizations and a third LUT that includes TFSVs for radix 5 factorizations.
16. The method of claim 13 , wherein generating the base vector values comprises iteratively cross multiplying the TFSV.
17. The method of claim 16 , further comprising serial to parallel converting the base vector values to generate a base vector.
18. The method of claim 17 , further comprising iteratively cross multiplying the base vector to generate the twiddle vector.
19. The method of claim 18 , further comprising utilizing a last stage coefficient LUT to generate the twiddle vector for a last radix stage.
20. The method of claim 11 , further comprising outputting the twiddle vector to a twiddle multiplication unit in a vector pipeline that carries up to 12 samples per cycle.