Skip to main content
teijo
Associate II
November 24, 2014
Question

cmsis dsp library: fft length

  • November 24, 2014
  • 1 reply
  • 774 views
Posted on November 24, 2014 at 13:22

Hi,

I

f i

understand correct

ly

using fft functions like 

http://mazsola.iit.uni-miskolc.hu/DATA/segedletek/arm/cmsis/CMSIS/Documentation/DSP/html/group___r_f_f_t___r_i_f_f_t.html#ga3df1766d230532bc068fc4ed69d0fcdc

, maxim

um fft length are 204

8. Is

t

here

any possibility to use longer

dat

a lengths

?

br

Teijo

    This topic has been closed for replies.

    1 reply

    frankmeyer9
    Associate III
    November 24, 2014
    Posted on November 24, 2014 at 14:44

    I

    f i

    understand correct

    ly

    using fft functions like 

    http://mazsola.iit.uni-miskolc.hu/DATA/segedletek/arm/cmsis/CMSIS/Documentation/DSP/html/group___r_f_f_t___r_i_f_f_t.html#ga3df1766d230532bc068fc4ed69d0fcdc

    , maxim

    um fft length are 204

    8

     

    IMHO it were either code size or runtime considerations that prompted ARM to limit the input size to 2^11.

     

    Is

    t

    here

    any possibility to use longer

    dat

    a lengths

    ?

    You could either try to extend the DSP lib code, or find another suitable implementation.

    There are plenty of others around which might better suit your needs, like implementations for arbitrary input length (like bluestein's fft), or to calculate just one or a few points of the spectre.