<html>
  <head>
    <meta content="text/html; charset=windows-1252"
      http-equiv="Content-Type">
  </head>
  <body bgcolor="#FFFFFF" text="#000000">
    <p>Hi everyone,<br>
      <br>
      I see you discussing this topic (the convolution box) for quite<br>
      a while now, but my stupid and naive question is, what do you<br>
      need this system for?</p>
    <p>I mean, is it only for reverb or rooms simulations?<br>
      Or do you want to model some amps or other gear (like the<br>
      Kemper Profiling thing does)?</p>
    <p>Because I personally see convolution approaches as very <br>
      inefficient methods (in terms of computing resources) when it<br>
      comes  to model a certain system response.</p>
    <p>Musically,<br>
      Frédéric</p>
    <p>PS: Not trolling the list, just curious.<br>
    </p>
    <br>
    <div class="moz-cite-prefix">Le 13/02/2017 à 17:18, cheater00
      cheater00 a écrit :<br>
    </div>
    <blockquote
cite="mid:CA+9GZUibxAwTVKBoqENvcF_OhP2y15sDCJwwFki8FYdv1M-8uA@mail.gmail.com"
      type="cite">
      <p dir="ltr">Hi Bruno,<br>
        TBH for a single instrument you don't need amazing adc or dac,
        but there are lots of i2s adc and dac boards out there geared
        for hifi enthusiasts. Some are cheap, and work well. Some are
        great quality, and you can spend as much as you want, really.</p>
      <br>
      <div class="gmail_quote">
        <div dir="ltr">On Mon, 13 Feb 2017 17:03 Bruno Afonso, <<a
            moz-do-not-send="true" href="mailto:bafonso@gmail.com">bafonso@gmail.com</a>>
          wrote:<br>
        </div>
        <blockquote class="gmail_quote" style="margin:0 0 0
          .8ex;border-left:1px #ccc solid;padding-left:1ex">
          <div dir="ltr" class="gmail_msg">Here are some random
            thoughts. Unless things have changed or are different with
            higher end sharc dev boards, I'm not sure you will get audio
            ins and outs that are acceptable for musicians, especially
            after you've gone through all the trouble to be able to use
            long and well made IRs, ie, think of nice reverbs where
            tails are really important. That means creating your own
            PCB, effectively increasing the total time to bring it to a
            useful tool that is plug and play, not only the time spent
            on learning and implementing the algorithms.
            <div class="gmail_msg"><br class="gmail_msg">
            </div>
            <div class="gmail_msg">That said, my feeling is that whoever
              implements these smarter approaches to convolution on a
              sharc platform and shares it will be eternally loved by
              the community. And once this happens likely others with
              hardware know how will chime in. That's a short step away
              from a sharc aleph-like.</div>
            <div class="gmail_msg"><br class="gmail_msg">
            </div>
            <div class="gmail_msg">An axoloti/nord-modular like
              interface and philosophy featuring really good building
              blocks exploring a powerful DSP platform would be amazing.
              Great for people wanting to develop their own DSP ideas
              but also with potential to be poor man's kyma. </div>
            <div class="gmail_msg"><br class="gmail_msg">
            </div>
            <div class="gmail_msg">cheers</div>
          </div>
          <div dir="ltr" class="gmail_msg">
            <div class="gmail_msg">b</div>
            <div class="gmail_msg"><br class="gmail_msg">
            </div>
          </div>
          <br class="gmail_msg">
          <div class="gmail_quote gmail_msg">
            <div dir="ltr" class="gmail_msg">On Mon, Feb 13, 2017 at
              8:49 AM cheater00 cheater00 <<a moz-do-not-send="true"
                href="mailto:cheater00@gmail.com" class="gmail_msg"
                target="_blank">cheater00@gmail.com</a>> wrote:<br
                class="gmail_msg">
            </div>
            <blockquote class="gmail_quote gmail_msg" style="margin:0 0
              0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
              <p dir="ltr" class="gmail_msg">It should also be said that
                the naiive algorithms have quadratic runtime complexity
                whereas the best ones have much better complexity (I
                believe n log n), so longer reverb tails that can be
                done with the optimized algorithm are simply not
                possible with the naiive approach, no matter how much
                hardware you throw at it - so that might be another
                reason to spend the time.</p>
              <br class="gmail_msg">
              <div class="gmail_quote gmail_msg">
                <div dir="ltr" class="gmail_msg">On Mon, 13 Feb 2017
                  14:40 cheater00 cheater00, <<a
                    moz-do-not-send="true"
                    href="mailto:cheater00@gmail.com" class="gmail_msg"
                    target="_blank">cheater00@gmail.com</a>> wrote:<br
                    class="gmail_msg">
                </div>
                <blockquote class="gmail_quote gmail_msg"
                  style="margin:0 0 0 .8ex;border-left:1px #ccc
                  solid;padding-left:1ex">
                  <p dir="ltr" class="gmail_msg">The simplest FFT and
                    convolution algorithms are easy to understand in
                    just hours, the really complex algorithm could take
                    weeks to implement, so if you're just doing this for
                    yourself the break even point is: will you otherwise
                    earn $200 - the difference between a dev board with
                    the cheapest DSPs and most powerful ones - in those
                    several weeks? If not, you might want to look into,
                    uh, flipping burgers or pizza delivery as a career
                    move. If you are doing this for production, or for
                    other people to build themselves, you want something
                    relatively inexpensive, though. The dev time might
                    be warranted if you do not mind spending it as a
                    learning experience.</p>
                  <br class="gmail_msg">
                  <div class="gmail_quote gmail_msg">
                    <div dir="ltr" class="gmail_msg">On Mon, 13 Feb 2017
                      10:38 Thomas Strathmann, <<a
                        moz-do-not-send="true"
                        href="mailto:thomas@pdp7.org" class="gmail_msg"
                        target="_blank">thomas@pdp7.org</a>> wrote:<br
                        class="gmail_msg">
                    </div>
                    <blockquote class="gmail_quote gmail_msg"
                      style="margin:0 0 0 .8ex;border-left:1px #ccc
                      solid;padding-left:1ex">On 12/02/17 22:39, <a
                        moz-do-not-send="true"
                        href="mailto:rsdio@audiobanshee.com"
                        class="gmail_msg" target="_blank">rsdio@audiobanshee.com</a>
                      wrote:<br class="gmail_msg">
                      > So, when combining FIR and FFT processing for
                      convolution, you'll<br class="gmail_msg">
                      > need MAC, bit-reversed addressing,
                      automatically-wrapped buffer<br class="gmail_msg">
                      > pointers, and possibly other special
                      instructions for maximum<br class="gmail_msg">
                      > efficiency at a given instruction clock rate.
                      Hopefully the DSP you<br class="gmail_msg">
                      > choose will have example code in optimized
                      assembly for a partitioned<br class="gmail_msg">
                      > convolution, and you won't have to piece all
                      of this together<br class="gmail_msg">
                      > yourself. Yes, you could do it all in
                      Standard C on a general purpose<br
                        class="gmail_msg">
                      > ARM or XMOS, but you'll need a higher clock
                      rate and more code to do<br class="gmail_msg">
                      > the same amount of work.<br class="gmail_msg">
                      <br class="gmail_msg">
                      I'm wondering: How precious would development time
                      hav to be to warrant<br class="gmail_msg">
                      going with a DSP and optimized assembly code
                      instead of taking the more<br class="gmail_msg">
                      blunt approach with a fast CPU and some plain C
                      code? From following<br class="gmail_msg">
                      this discussion I get the impression that the
                      answer to that question is<br class="gmail_msg">
                      "Very" but is that true?<br class="gmail_msg">
                      <br class="gmail_msg">
                              Thomas<br class="gmail_msg">
                      _______________________________________________<br
                        class="gmail_msg">
                      Synth-diy mailing list<br class="gmail_msg">
                      <a moz-do-not-send="true"
                        href="mailto:Synth-diy@synth-diy.org"
                        class="gmail_msg" target="_blank">Synth-diy@synth-diy.org</a><br
                        class="gmail_msg">
                      <a moz-do-not-send="true"
                        href="http://synth-diy.org/mailman/listinfo/synth-diy"
                        rel="noreferrer" class="gmail_msg"
                        target="_blank">http://synth-diy.org/mailman/listinfo/synth-diy</a><br
                        class="gmail_msg">
                    </blockquote>
                  </div>
                </blockquote>
              </div>
              _______________________________________________<br
                class="gmail_msg">
              Synth-diy mailing list<br class="gmail_msg">
              <a moz-do-not-send="true"
                href="mailto:Synth-diy@synth-diy.org" class="gmail_msg"
                target="_blank">Synth-diy@synth-diy.org</a><br
                class="gmail_msg">
              <a moz-do-not-send="true"
                href="http://synth-diy.org/mailman/listinfo/synth-diy"
                rel="noreferrer" class="gmail_msg" target="_blank">http://synth-diy.org/mailman/listinfo/synth-diy</a><br
                class="gmail_msg">
            </blockquote>
          </div>
        </blockquote>
      </div>
      <br>
      <fieldset class="mimeAttachmentHeader"></fieldset>
      <br>
      <pre wrap="">_______________________________________________
Synth-diy mailing list
<a class="moz-txt-link-abbreviated" href="mailto:Synth-diy@synth-diy.org">Synth-diy@synth-diy.org</a>
<a class="moz-txt-link-freetext" href="http://synth-diy.org/mailman/listinfo/synth-diy">http://synth-diy.org/mailman/listinfo/synth-diy</a>
</pre>
    </blockquote>
    <br>
  </body>
</html>