• State
  • Locked Locked
  • Replies 1 reply
  • Subscribers 98 subscribers
  • Views 399 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.

Using two 6713 in series

Rajan Chauhan
Prodigy 30 points

I have found the need to use two DSK6713 boards, such that board one is doing some form of processing to a signal, then that processed signal is fed into another boards audio input to be further processed. Each processing task is very intensive therefore they cant be done on the same board, and the 6713 boards are the only thing i have access to.

Other than the latency introduced from my own code and processing, what sort of latency am I looking at by connecting two boards together?

Thank you,

Rajan

  • 0
    TI__Guru* 84110 points

    Rajan,

    Have you given up on your other thread here concerning 'an insanely large amount of clock cycles' ? I think Shankari is waiting on some answers to help you there. Do you know how low you need to get that number to let you avoid moving to two boards?

    There are quite a few components to the latency you will see using the analog I/O between boards. The only practical way to know the answer will be to measure it, using the type of signal that you expect to send across the patch cable. You would want to use two patch cables so you can send the signal from Board 1 to Board 2, and then send the signal directly back from Board 2 to Board 1. Then on Board 1 you can compare the signals in a pair of memory buffers to see how much lag and distortion there is between the original signal and the returned signal.

    If both boards will be at full capacity in doing the intensive processing, you will need to make good use of the EDMA for data transfers and use this in your measurement of latency. Most of the time, we setup ping-pong buffers for audio data so the processing can be done in the most efficient way on a buffer of data instead of a single sample at a time. You will have less latency with the single sample method, but total processing time may be worse. These are things you can learn about in the C6713 workshop we have archived - go to TI.com and search for "c6713 workshop" (no quotes). I think the course title has DSP Integration in it.

    At the very least, please keep your options open between a more efficient 1-board solution and a more difficult 2-board solution.

    Regards,
    RandyP