<div dir="ltr"><div>Just to chime in, if I recall correctly, the reason for needing parallel HDF5 with laserslab physics is for the ParallelIO of laser rays, which I believe you can turn off as well.</div><div><br></div><div>Scott<br></div></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Thu, Oct 8, 2020 at 1:44 PM David Blackman <<a href="mailto:drblackman@eng.ucsd.edu">drblackman@eng.ucsd.edu</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>
<p>I've just had to learn how to compile it this morning as well. I
don't know what system you're on, but you either need to link to a
HDF5 library that has been compiled specifically for parallel
operation (for me on the Stampede2 cluster I had to load phdf5
instead of hdf5), or compile the HDF5 library with mpicc instead
of gcc, or whatever compiler it is you use. There are specific
instructions on the HDF5 website for compiling with MPI-I/O.
<a href="https://support.hdfgroup.org/HDF5/PHDF5/" target="_blank">https://support.hdfgroup.org/HDF5/PHDF5/</a> <br>
</p>
<p>Good luck!<br>
</p>
<div>On 08/10/2020 05:05, Gabriel Perez
Callejo wrote:<br>
</div>
<blockquote type="cite">
<p>Hello,</p>
<p>I am new to Flash (I have just recently installed it), and I am
having some problems when running the LaserSlab example case in
parallel.</p>
<p>I have installed openmpi-4.0.5, hypre-2.12.0, hdf5-1.10.7, and
FLASH4.6.2. When I run the Sedov example using just</p>
<p><i>./setup -auto Sedov</i></p>
<p><i>mpirun -np [number of cores] flash4</i></p>
<p>It runs perfectly. However, when I do the same with the
LaserSlab example:</p>
<p><i>./setup -auto LaserSlab -2d +cylindrical +hdf5typeio -nxb=16
-nyb=16 species=cham,targ +mtmmmt +laser +uhd3t +mgd
mgd_meshgroups=6 -parfile=example.par</i></p>
<p><i>mpirun -np [number of cores] flash4</i></p>
<p>The code starts running, but freezes after <i>'Initial dt
verified'</i>.</p>
<p>Strangely enough, when the same setup command is run for
LaserSlab, but instead of running in parallel, FLASH is called
to operate in serial mode (./flash4), the code works without any
problems (it just takes a much longer time, obviously).</p>
<p>From what I have gathered, this seems to be a problem with the
hdf5 I/O, but I have been unable to solve it, and I don't know
what might be happening. Any thoughts about how to solve this?</p>
<p>Thanks everyone,</p>
<p>Gabriel Perez<br>
</p>
<p><br>
</p>
</blockquote>
</div>
</blockquote></div>