<html>
  <head>
    <meta content="text/html; charset=ISO-8859-1"
      http-equiv="Content-Type">
  </head>
  <body text="#000000" bgcolor="#FFFFFF">
    Hi again,<br>
    <br>
    I did a try on my asterisk 11.2.1 compiled on Ubuntu 12.04 (64 bit)
    with a simple Pentium 4 CPU (Intel(R) Pentium(R) D CPU 2.80GHz). I
    connected 5 SIP-Users with a ConfBridge. This is my picture:<br>
    <br>
    <img src="cid:part1.07000500.05090102@ovm-group.com" alt=""><br>
    <br>
    <img src="cid:part2.04010108.00060601@ovm-group.com" alt=""><br>
    <br>
    Please give a a hint where I can change "you parameters" like
    denoise etc. So I will try to change these settings on my box also.
    I have no experience with Confbridge.<br>
    <br>
    -Thorsten-<br>
    <br>
    <div class="moz-cite-prefix">Am 08.02.2013 19:34, schrieb Hristo
      Trendev:<br>
    </div>
    <blockquote
cite="mid:CAF5NwdMCi1hd3euNqi-phHydQAr6vQzkCAq-jtH5mBMBfdQMLw@mail.gmail.com"
      type="cite">
      <div dir="ltr">Hi, the quad-core server is a dedicated asterisk
        server. I duplicated the tests on a virtual server (running on
        another physical server) only to rule out the possibility of
        hardware problem with the first sever.
        <div>
          <br>
        </div>
        <div style="">Hristo</div>
      </div>
      <div class="gmail_extra"><br>
        <br>
        <div class="gmail_quote">On Fri, Feb 8, 2013 at 11:41 AM,
          Thorsten G&ouml;llner <span dir="ltr">&lt;<a
              moz-do-not-send="true" href="mailto:tg@ovm-group.com"
              target="_blank">tg@ovm-group.com</a>&gt;</span> wrote:<br>
          <blockquote class="gmail_quote" style="margin:0 0 0
            .8ex;border-left:1px #ccc solid;padding-left:1ex">
            <div text="#000000" bgcolor="#FFFFFF"> Hi,<br>
              <br>
              perhaps it is a problem with your Host-Guest-Setup? Did
              you try the Asterisk-Setup on a dedicated server without
              virtualization?<br>
              <br>
              -Thorsten-<br>
              <br>
              <div>Am 07.02.2013 11:42, schrieb Hristo Trendev:<br>
              </div>
              <div>
                <div class="h5">
                  <blockquote type="cite">
                    <div dir="ltr">Hi Thorsten,
                      <div><br>
                      </div>
                      <div>Thanks for your reply. I did check core show
                        translations, but the following&nbsp;<a
                          moz-do-not-send="true"
href="http://lists.digium.com/pipermail/asterisk-users/2012-November/276132.html"
                          target="_blank">http://lists.digium.com/pipermail/asterisk-users/2012-November/276132.html</a>
                        suggests that the values displayed are no longer
                        representing the computation cost only. However
                        to answer your question:</div>
                      <div><br>
                      </div>
                      <div>G722 to SLIN16 cost is 9000, reverse
                        direction is 6000</div>
                      <div>ALAW to SLN16 cost is 17000, reverse
                        direction is 14500<br>
                      </div>
                      <div><br>
                      </div>
                      <div>G722 to SLN cost is 9600, reverse direction
                        is 8250<br>
                      </div>
                      <div>ALAW to SLN cost is 9000, reverse direction
                        is 6000</div>
                      <div><br>
                      </div>
                      <div>With regards to the CPU usage per core -
                        inside the VM, where only one core is available,
                        the CPU was close to 100% when the problem
                        started to apear, on the physical server with 4
                        cores, the cores were evenly loaded at about
                        30-40%. A single call into the conference
                        consumed between 10-20% depending on whether I
                        have denoise enabled or not.</div>
                      <div><br>
                      </div>
                      <div>There is no dahdi board installed, I only use
                        the dahdi module for conference timer (note that
                        the problem is also present with the&nbsp;<span
                          style="font-family:arial,sans-serif;font-size:13px">timerfd

                          timing module).</span></div>
                      <div><br>
                      </div>
                      <div>BR,</div>
                      <div>Hristo</div>
                    </div>
                    <div class="gmail_extra"><br>
                      <br>
                      <div class="gmail_quote">On Wed, Feb 6, 2013 at
                        1:57 PM, Thorsten G&ouml;llner <span dir="ltr">&lt;<a
                            moz-do-not-send="true"
                            href="mailto:tg@ovm-group.com"
                            target="_blank">tg@ovm-group.com</a>&gt;</span>
                        wrote:<br>
                        <blockquote class="gmail_quote" style="margin:0
                          0 0 .8ex;border-left:1px #ccc
                          solid;padding-left:1ex">Did you check<br>
                          asterisk -rx "core show translation recalc 10"<br>
                          <br>
                          Am 06.02.2013 13:56, schrieb Thorsten G&ouml;llner:
                          <div>
                            <div><br>
                              <blockquote class="gmail_quote"
                                style="margin:0 0 0 .8ex;border-left:1px
                                #ccc solid;padding-left:1ex"> Sorry - I
                                just read you alsways checked the cpu
                                usage. Are all cores at 100%? Is it the
                                atserisk process which consumes it all?<br>
                                <br>
                                Am 06.02.2013 13:54, schrieb Thorsten
                                G&ouml;llner:<br>
                                <blockquote class="gmail_quote"
                                  style="margin:0 0 0
                                  .8ex;border-left:1px #ccc
                                  solid;padding-left:1ex"> Did you watch
                                  the cpu usage (for example with top)?<br>
                                  You have a board installed which does
                                  use dahdi? Did you check the command
                                  "dahdi_test"?<br>
                                  Maybe a (performance) problem of the
                                  software ec?<br>
                                  <br>
                                  Am 06.02.2013 11:13, schrieb Hristo
                                  Trendev:<br>
                                  <blockquote class="gmail_quote"
                                    style="margin:0 0 0
                                    .8ex;border-left:1px #ccc
                                    solid;padding-left:1ex"> Hi,<br>
                                    <br>
                                    I have been experimenting with
                                    ConfBridge from the asterisk-11
                                    stable SVN branch (and with 11.2.0
                                    also) for the last 3 weeks and I see
                                    a problem, which what I believe is
                                    performance related. I just wanted
                                    to ask if someone else has made any
                                    tests and what is the maximum number
                                    of participants that they've seen in
                                    a conference.<br>
                                    <br>
                                    I was never able to get more than 8
                                    participants (mixed G722 and G711a)
                                    on a conference (actually that's per
                                    server limit) with almost all
                                    settings on default, except for
                                    dsp_drop_silence and denoise which
                                    are enabled.<br>
                                    <br>
                                    I tested on Debian squeeze, 64-bit,
                                    quad-core Xeon server @2.4GHz and
                                    also on another virtual server with
                                    similar processor (just one core
                                    available to the VM). While this is
                                    not the latest and greatest CPU, I
                                    would certainly expect it to handle
                                    more than 8 calls.<br>
                                    <br>
                                    To be honest, I was in fact able to
                                    get it working for up to 20
                                    participants (most with G711), when
                                    I switched from res_timing_timerfd
                                    to res_timing_dahdi and turned off
                                    denoise, but that's still not normal
                                    I believe, especially with most
                                    participants on mute and with
                                    dps_drop_silence enabled and nothing
                                    else running on the server.<br>
                                    <br>
                                    The problem itself is, that once I
                                    get over the "critical" number of
                                    participants, the voice starts to
                                    break up and it's impossible to
                                    understand the person who's talking.
                                    This is certainly not bandwidth
                                    related because all tests were made
                                    on the LAN and besides I could see
                                    that the CPU was sometime close to
                                    100%.<br>
                                    <br>
                                    Did someone observe something
                                    similar?<br>
                                    <br>
                                    BTW, once the first participant
                                    enters the conference I start seeing
                                    probably over 50 messages per second
                                    saying:<br>
                                    <br>
                                    bridging.c:757
                                    bridge_channel_join_multithreaded:
                                    Going into a multithreaded waitfor
                                    for bridge channel 0x292d708 of
                                    bridge 0x28f3658 <br>
                                  </blockquote>
                                </blockquote>
                              </blockquote>
                              <br>
                            </div>
                          </div>
                        </blockquote>
                      </div>
                    </div>
                  </blockquote>
                </div>
              </div>
            </div>
          </blockquote>
        </div>
      </div>
    </blockquote>
    <br>
  </body>
</html>