This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

Re: fft time in evm6678l

Other Parts Discussed in Thread: TMS320C6670

from"TMS320C6670 Breakthrough performance for process-intensive applications"

 

C66x @1.2 GHz Single precision floating-point FFT, 2048 pt. radix 4 costs 14.60 us.

 

but the code in dsplib doesn't achieve this performance , maybe only half the speed...

i also don't know why...

  • Can you provide more detail about the scenario?  What type of memory is the data located in?  How are you measuring performance?  With a simulator?  or some other way?  Are you running this on the EVM6670L?  Can you provide a cutdown project?

    Regards,

    Dan

     

  • Hi, DanRinkes,

    I tried the code in "..\dsplib_c66x_3_0_0_8\packages\ti\dsplib\src\DSPF_sp_fftSPxSP\c66\DSPF_sp_fftSPxSP_66_LE_ELF" on EVM6678L

    here the result and the scenario:

    • [C66xx_0] DSPF_sp_fftSPxSP Iter#: 9 Result Successful N = 2048 radix = 2 natC: 97296 optC: 33197 cycles
    • [C66xx_0] DSPF_sp_fftSPxSP Iter#: 8 Result Successful N = 1024 radix = 4 natC: 40476 optC: 14762 cycles

     

    • both code and data (in, out, and twiddle factors) placed in L2 SRAM
    • ccxml: texas instruments xds100v1 usb emulator
    • I use clock() as well as on-chip Timer to measure performance, the results are almost the same
    and the project file: 7750.DSPF_sp_fftSPxSP.zip
    How can I replicate the results “C66x @1.2 GHz Single precision floating-point FFT, 2048 pt. radix 4 costs 14.60 us.” on EVM6678L?
    Thanks!
  • it seems that use the code in dsplib within simulator will get the same result in the white paper...

    but the same code on evm board will get the cycle doubled...

    i don;t know why.... whether some special consideration or configuration is needed for the emulator board....

     

    thanks....