<div class="gmail_quote">On Fri, Aug 26, 2011 at 08:12, Dominik Szczerba <span dir="ltr"><<a href="mailto:dominik@itis.ethz.ch">dominik@itis.ethz.ch</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">
<div id=":9k">Do you claim I should not use MPI_Barrier at all when only programming<br>
Petsc? Is really all required synchronization managed automatically?<br></div></blockquote><div><br></div><div>Yes, apart from unconventional/discouraged things like communicating through the file system, and even then, you should usually do without barriers.</div>
<div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;"><div id=":9k">
<br>
I do some raw MPI programming before and after, but not inside the<br>
Petsc core solver.</div></blockquote></div><br><div>Common problems include having one or more of your processes hung up in this earlier phase and possible tag collisions. If you are doing nontrivial things in your pure MPI code, you should either create the communicator yourself or MPI_Comm_dup() one that you get from PETSc.</div>