<div dir="ltr"><div>Hi Rob,<br><br>Please find enclosed a fully working test case to write a variable collectively, using ParallelNetCDF. <br>I took the liberty to add few things I was trying to do and for which I also struggled a little bit; the program remains however very simple. The program is first creating a small variable (kind of metadata), which is created and written by solely one thread, and then the file is re-open and the variable is written collectively by all the MPI threads.<br>
<br></div>Best,<br><br>MBR<br></div><div class="gmail_extra"><br><br><div class="gmail_quote">On Fri, Jan 25, 2013 at 7:37 PM, Rob Latham <span dir="ltr"><<a href="mailto:robl@mcs.anl.gov" target="_blank">robl@mcs.anl.gov</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div class="im">On Fri, Jan 25, 2013 at 07:08:00PM +0100, MBR MBR wrote:<br>
> Hi Wei-keng,<br>
><br>
> The problem was indeed the start... xpos, ypos & zpos started at 0, and I<br>
> re-used what I did in parallel HDF5, where the "start" index start at 0 as<br>
> in C.<br>
><br>
> Thanks, and sorry for the dumb question!<br>
<br>
</div>Hi MBR: Glad Wei-keng could fix you up so quickly. Can you contribute<br>
a driver for this routine? We are always looking for small-ish test<br>
cases we can add to our test suite, even tests that are supposed to<br>
fail.<br>
<span class="HOEnZb"><font color="#888888"><br>
==rob<br>
<br>
--<br>
Rob Latham<br>
Mathematics and Computer Science Division<br>
Argonne National Lab, IL USA<br>
</font></span></blockquote></div><br></div>