<html>
  <head>
    <meta content="text/html; charset=ISO-8859-1"
      http-equiv="Content-Type">
  </head>
  <body bgcolor="#FFFFFF" text="#000000">
    Also, on July 1 I will have another $5K to spend and I plan to get a
    whole bunch more C6100's (rack2), since they've been such a success!
    $5K should get us around 6 more of those units, which means 24 more
    nodes... if I can find 12U and the requisite power somewhere for
    them, which are both actual problems, but nice problems to have :)<br>
    We can definitely get another 3 units, or 12 more nodes - I've got
    the rackspace and power for that.<br>
        -Josiah<br>
    <br>
    <br>
    <div class="moz-cite-prefix">On 6/27/13 2:26 PM, Thomas Helmuth
      wrote:<br>
    </div>
    <blockquote
cite="mid:CABgVVjc3EH4ac3+Bciuf6B=JEcd1XFjto3mkr0ux8+a+AxuOWA@mail.gmail.com"
      type="cite">
      <div dir="ltr">
        <div>
          <div>Hi Jaime,<br>
            <br>
          </div>
          I think I'm the main culprit right now. I have quite a few
          long, important, but not very time sensitive runs going on the
          cluster currently. Some have been going for 5 to 10 days now,
          so I'd prefer not to cancel them and lose that work, even
          though they could be going another 5 to 10 days until
          completion. I'd be happy to pause any new launches of runs for
          now, but I'd prefer to allow the already-started runs to
          finish if at all possible.<br>
          <br>
        </div>
        <div>Also, for figuring out the future sharing of the cluster,
          I'd be happy to only have my runs hog certain nodes, or have
          some other way of having them get lower priority when others
          need to use it. Let's talk about it at the lab meeting.<br>
        </div>
        <div><br>
        </div>
        -Tom</div>
      <div class="gmail_extra"><br>
        <br>
        <div class="gmail_quote">On Thu, Jun 27, 2013 at 2:23 PM, Wm.
          Josiah Erikson <span dir="ltr"><<a moz-do-not-send="true"
              href="mailto:wjerikson@hampshire.edu" target="_blank">wjerikson@hampshire.edu</a>></span>
          wrote:<br>
          <blockquote class="gmail_quote" style="margin:0 0 0
            .8ex;border-left:1px #ccc solid;padding-left:1ex">The issue
            right now is that many of Tom's Digital Multiplier processes
            seem to be taking around 9 days to finish, so once they've
            grabbed a slot, they hold on to it for a very long time,
            which effectively means that nobody else can use the cluster
            while they are running, since we don't have any
            kicking-people-out algorithm, and everybody else's jobs
            finish in more like 20 minutes.<br>
            The solution, of course, is to change the "tom" tag to only
            use every other slot or something like that, or only run two
            processes per machine or something... but then they'll take
            even longer to finish.<br>
            Discussing tomorrow at the meeting seems like a good plan.<br>
                -Josiah
            <div class="HOEnZb">
              <div class="h5"><br>
                <br>
                <br>
                On 6/27/13 2:19 PM, Lee Spector wrote:<br>
                <blockquote class="gmail_quote" style="margin:0 0 0
                  .8ex;border-left:1px #ccc solid;padding-left:1ex">
                  Hi Jaime,<br>
                  <br>
                  Fine with me personally but I'll check with my lab
                  group to see what everyone's expected needs are. I'm
                  also not sure exactly how to implement the idea if
                  people do want to run some other things... maybe by
                  having you use a subset of machines that everyone else
                  excludes?<br>
                  <br>
                    -Lee<br>
                  <br>
                  On Jun 27, 2013, at 2:14 PM, Jaime Davila wrote:<br>
                  <br>
                  <blockquote class="gmail_quote" style="margin:0 0 0
                    .8ex;border-left:1px #ccc solid;padding-left:1ex">
                    Greetings everyone,<br>
                    <br>
                    I wanted to check to see if it was possible for me
                    to grab some more CPU cycles out of the cluster for
                    a week or so. I just placed a new algorithm on my
                    account, and have it run it, but it's fairly
                    different from what I was doing before, and I would
                    rather detect quickly if I need to tweak or change
                    things, as opposed to having to wait a week to
                    realize I need to make a 10 minute change.<br>
                    <br>
                    Last time that the system load diminished some, I
                    noticed that my processes run at their top speed if
                    the number of CPUs loaded to their maximum drops to
                    about 75%, as opposed to the 97% where they are now.
                    Maybe things will be that way this time around,
                    maybe not? In either case, my grabbing more cpu
                    cycles right now would be extremely useful.<br>
                    <br>
                    Thoughts?<br>
                    <br>
                    Thanks a lot,<br>
                    <br>
                    Jaime<br>
                    <br>
                    -- <br>
                    ******************************************************<br>
                    Jaime J. Dávila<br>
                    Associate Professor of Computer Science<br>
                    Hampshire College<br>
                    jdavila at hampshire dot edu<br>
                    <a moz-do-not-send="true"
                      href="http://helios.hampshire.edu/jdavila"
                      target="_blank">http://helios.hampshire.edu/jdavila</a><br>
                    *******************************************************<br>
                    <br>
                    _______________________________________________<br>
                    Clusterusers mailing list<br>
                    <a moz-do-not-send="true"
                      href="mailto:Clusterusers@lists.hampshire.edu"
                      target="_blank">Clusterusers@lists.hampshire.edu</a><br>
                    <a moz-do-not-send="true"
                      href="https://lists.hampshire.edu/mailman/listinfo/clusterusers"
                      target="_blank">https://lists.hampshire.edu/mailman/listinfo/clusterusers</a><br>
                  </blockquote>
                  --<br>
                  Lee Spector, Professor of Computer Science<br>
                  Cognitive Science, Hampshire College<br>
                  893 West Street, Amherst, MA 01002-3359<br>
                  <a moz-do-not-send="true"
                    href="mailto:lspector@hampshire.edu" target="_blank">lspector@hampshire.edu</a>,
                  <a moz-do-not-send="true"
                    href="http://hampshire.edu/lspector/"
                    target="_blank">http://hampshire.edu/lspector/</a><br>
                  Phone: <a moz-do-not-send="true"
                    href="tel:413-559-5352" value="+14135595352"
                    target="_blank">413-559-5352</a>, Fax: <a
                    moz-do-not-send="true" href="tel:413-559-5438"
                    value="+14135595438" target="_blank">413-559-5438</a><br>
                  <br>
                  _______________________________________________<br>
                  Clusterusers mailing list<br>
                  <a moz-do-not-send="true"
                    href="mailto:Clusterusers@lists.hampshire.edu"
                    target="_blank">Clusterusers@lists.hampshire.edu</a><br>
                  <a moz-do-not-send="true"
                    href="https://lists.hampshire.edu/mailman/listinfo/clusterusers"
                    target="_blank">https://lists.hampshire.edu/mailman/listinfo/clusterusers</a><br>
                </blockquote>
                <br>
              </div>
            </div>
            <span class="HOEnZb"><font color="#888888">
                -- <br>
                Wm. Josiah Erikson<br>
                Assistant Director of IT, Infrastructure Group<br>
                System Administrator, School of CS<br>
                Hampshire College<br>
                Amherst, MA 01002<br>
                <a moz-do-not-send="true"
                  href="tel:%28413%29%20559-6091" value="+14135596091"
                  target="_blank">(413) 559-6091</a></font></span>
            <div class="HOEnZb">
              <div class="h5"><br>
                <br>
                _______________________________________________<br>
                Ci-lab mailing list<br>
                <a moz-do-not-send="true"
                  href="mailto:Ci-lab@lists.hampshire.edu"
                  target="_blank">Ci-lab@lists.hampshire.edu</a><br>
                <a moz-do-not-send="true"
                  href="https://lists.hampshire.edu/mailman/listinfo/ci-lab"
                  target="_blank">https://lists.hampshire.edu/mailman/listinfo/ci-lab</a><br>
              </div>
            </div>
          </blockquote>
        </div>
        <br>
      </div>
    </blockquote>
    <br>
    <pre class="moz-signature" cols="72">-- 
Wm. Josiah Erikson
Assistant Director of IT, Infrastructure Group
System Administrator, School of CS
Hampshire College
Amherst, MA 01002
(413) 559-6091
</pre>
  </body>
</html>