• State
  • Locked Locked
  • Replies 1 reply
  • Subscribers 97 subscribers
  • Views 274 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

code running from DDR2 taking more time than that running through L2RAM of OMAPL138 DSP

Prajakta Ranade
Prodigy 110 points
Other Parts Discussed in Thread: OMAPL138

Hello,

I am currently using DSP side of OMAPL138.

In DSP i am calculating FFT using "dsplib_c674x_3_1_1_1" library from TI. 

I am using DSPF_sp_cfftr2_dit function to calculate single precesion radix 2 FFT.

[DSPF_sp_cfftr2_dit(float* , float* , unsigned short)]... Thia is an asm function from the above mentioned library, used as it is.

Now the issue I am faing is: When I put my entire code in L2RAM, function takes hardly 24 usec to compute FFT of single channel, But when i shift my code in DDR2 it is taking around 118 usec to execute same code!!!

I tried doing #pragma for code & data section for this particular function and putting only this much part in L2RAM and rest code in DDR2, but facing same problem.

My final aim is to run code through DDR2 only, so please provide me solution on this at earliest.

Waiting for positive solutions....

--

Thanks,

Prajakta

  • 0
    TI__Mastermind 35645 points

    HI Prajakta,

    Thanks for your post.

    I understand that, you still need to optimize your code to speed up performance when you execute the code through DDR2 and you can meet the expected performance. There are many more optimzation techniques for the TI c6000 compiler in which you can enhance your speed in computing FFT of single channel.

    I hope, you can optimize your code still more like, choosing proper compiler options, using SIMD operations, using intrinsic operations, applying the restrict keyword etc. For more details of c6000 optimization techniques, please refer section 3 in the c6000 dsp optimization app. report as below:

    http://www.ti.com/lit/an/sprabf2/sprabf2.pdf

    Also, there are c6000 optimization workshops & training materials. Kindly refer below the wiki article:

    http://processors.wiki.ti.com/index.php/Optimization_Techniques_for_the_TI_C6000_Compiler

    Thanks & regards,

    Sivaraj K

    -------------------------------------------------------------------------------------------------------
    Please click the Verify Answer button on this post if it answers your question.
    -------------------------------------------------------------------------------------------------------