<div dir="ltr"><div>Actually, the way I have implemented the streaming, it still allocates the 30Go complete volume and compute it piece by piece. One thing you could try is to remove the streamerBP object, connect directly the reconstruction to the writer "writer->SetInput(pfeldkamp->GetOutput());" and set the streaming in the writer "writer->SetNumberOfStreamDivisions(args_info.divisions_arg);". Then it never allocates the whole volume in memory. If that works for you, I think you can open a PR on github with this change, that makes a lot more sense in my opinion.</div></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Tue, Feb 11, 2020 at 8:46 PM vincent <<a href="mailto:vl@xris.eu" target="_blank">vl@xris.eu</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
  
    
  
  <div bgcolor="#FFFFFF">
    <p>Hi Simon,</p>
    <p>yes, I used both in my command line.  I have 64 Go RAM on the
      machine, so that shouldn't be the issue.  For the sake of
      completeness, I also tried the subset option in combination with
      the divisions option, going as low as 1, but to no avail.</p>
    <p>I'll investigate further tomorrow.</p>
    <p>Thank you again for your help,</p>
    <p>Vincent<br>
    </p>
    <div>On 2020-02-11 8:08 p.m., Simon Rit
      wrote:<br>
    </div>
    <blockquote type="cite">
      
      <div dir="ltr">
        <div>Have you tried the combination of both? To be clear,
          --divisions acts on the reconstructed volume so it should be
          ~7 Go with the "--divisions 4" option (instead of
          2000*2000*2000*4/1024/1024/1024=29.8 Go otherwise).</div>
        <div>The --lowmem option acts on the projections and you have
          250 Mo (instead of 2048*2048*1500*4/1024/1024/1024=23.4 Go
          otherwise).</div>
        <div>The message "Failed to allocate memory for image" seems to
          be a CPU memory issue. Are you sure you have about 10 Go
          available to run this reconstruction?<br>
        </div>
      </div>
      <br>
      <div class="gmail_quote">
        <div dir="ltr" class="gmail_attr">On Tue, Feb 11, 2020 at 7:31
          PM vincent <<a href="mailto:vl@xris.eu" target="_blank">vl@xris.eu</a>> wrote:<br>
        </div>
        <blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
          <div bgcolor="#FFFFFF">
            <p>Hi Simon, <br>
            </p>
            <p>I am afraid I forgot to mention something in my last
              email.  I tried to use the lowmem option, as you suggested
              a while ago in the list for the same problem, but I am
              afraid I am still getting the same error.</p>
            <p>kind regards,</p>
            <p>Vincent<br>
            </p>
            <div>On 11.02.20 17:36, Simon Rit wrote:<br>
            </div>
            <blockquote type="cite">
              <div dir="ltr">
                <div>Hi Vincent,</div>
                <div>There is a way to do such a thing in rtkfdk with
                  the --divisions option, see code <a href="https://github.com/SimonRit/RTK/blob/master/applications/rtkfdk/rtkfdk.cxx#L190-L196" target="_blank">here</a>. <br>
                </div>
                <div>I also don't really understand either what's going
                  on in your bottom reconstruction, it seems to be a
                  geometric problem. Have you checked an axial slice?</div>
                <div>Simon</div>
              </div>
              <br>
              <div class="gmail_quote">
                <div dir="ltr" class="gmail_attr">On Tue, Feb 11, 2020
                  at 4:21 PM vincent <<a href="mailto:vl@xris.eu" target="_blank">vl@xris.eu</a>>
                  wrote:<br>
                </div>
                <blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">Hello RTK
                  community,<br>
                  <br>
                  I am afraid that my question might not be directly
                  related to the <br>
                  excellent implementation we are all using, but it
                  might still be <br>
                  interesting for some of you.<br>
                  <br>
                  I have a stack of 1500 projections of size 2048*2048. 
                  I obviously can't <br>
                  reconstruct the full resolution volume on my graphics
                  card, as it is too <br>
                  big.  So my solution was to split the sinogram into N
                  parts, for which <br>
                  each reconstructed volume would fit in my GPU memory
                  and then reassemble <br>
                  them.  I did a test with a 700*820*900 sinogram, that
                  I cut in two parts <br>
                  of 700*410(+a small overlap)*900.<br>
                  <br>
                  While the reconstruction of the whole volume was
                  acceptable, I got a <br>
                  weird issue with the split ones: the one corresponding
                  to the top of the <br>
                  image is also ok, but the bottom one is very blurry. 
                  The three images <br>
                  can be found at the following links:<br>
                  <br>
                  <a href="https://ibb.co/vLk9ZhQ" rel="noreferrer" target="_blank">https://ibb.co/vLk9ZhQ</a><br>
                  <a href="https://ibb.co/m4pm0LT" rel="noreferrer" target="_blank">https://ibb.co/m4pm0LT</a><br>
                  <a href="https://ibb.co/Jyf1yKM" rel="noreferrer" target="_blank">https://ibb.co/Jyf1yKM</a><br>
                  <br>
                  I used the same calibration parameters for the three
                  reconstruction.  I <br>
                  visually checked the split sinograms and they looked
                  fine.<br>
                  <br>
                  <br>
                  Any insight will be much appreciated !<br>
                  <br>
                  <br>
                  Thanks in advance,<br>
                  <br>
                  kindest regards,<br>
                  <br>
                  Vincent<br>
                  <br>
                  _______________________________________________<br>
                  Rtk-users mailing list<br>
                  <a href="mailto:Rtk-users@public.kitware.com" target="_blank">Rtk-users@public.kitware.com</a><br>
                  <a href="https://public.kitware.com/mailman/listinfo/rtk-users" rel="noreferrer" target="_blank">https://public.kitware.com/mailman/listinfo/rtk-users</a><br>
                  <br>
                </blockquote>
              </div>
            </blockquote>
          </div>
          _______________________________________________<br>
          Rtk-users mailing list<br>
          <a href="mailto:Rtk-users@public.kitware.com" target="_blank">Rtk-users@public.kitware.com</a><br>
          <a href="https://public.kitware.com/mailman/listinfo/rtk-users" rel="noreferrer" target="_blank">https://public.kitware.com/mailman/listinfo/rtk-users</a><br>
        </blockquote>
      </div>
    </blockquote>
  </div>

</blockquote></div>