Re: Bitwidth required for an 8K point FFT

On 11月29日, 下午2時53分, "sudarshan_onkar" <sudarshan.on...@xxxxxxxxx>
wrote:
HI,
I am doing a FFT processor for N=8192 in fixed point. Since
there is truncation and scaling at every stage, precision will be lost in
fixed point. So how many bits i should keep in intermediate stages so
that no considerable loss occurs.

Regards
Sudarshan


It depends on your application. There is no definite answer for that.
.

Relevant Pages

  • Bitwidth required for an 8K point FFT
    ... there is truncation and scaling at every stage, precision will be lost in ...
    (comp.dsp)
  • Re: How to improve the bad condition number of Xb
    ... for limited precision calculations (e.g., ... powers of two, double scaling with powers of two,and improving ... While comparing INV,QR and PINV approaches to the current problem ... When I scaled with invC ...
    (comp.soft-sys.matlab)
  • Re: "typedef" in F95/2003?
    ... real:: foo ... The precision is not lost in the assignment. ... doesn't do a good enough job of hiding the underlying definition of the ...
    (comp.lang.fortran)
  • Re: I think a problem occured when i used long()
    ... > I did the folloing in python shell: ... fractions get truncated. ... by scaling your data so it's an integer. ... For greater fractional precision, ...
    (comp.lang.python)
  • Re: FIR filter O/p width
    ... When later truncation creates so little noise that the noise from the ... What is the meaning of 6.2831853071795864 +/- ..001? ... and numbers can have any presumed precision. ...
    (comp.dsp)