All,<div><br></div><div>if openmpi is installed on sparc, it takes priority over lam ! (it has priority 40 and lam 30)</div><div><br></div><div>here is the result of mpic++ -showme:compile on <a href="http://smetana.debian.org">smetana.debian.org</a> </div>
<div><div>-I/usr/lib/openmpi/include -I/usr/lib/openmpi/include/openmpi -pthread</div><div>and link</div><div><div>mpic++ -showme:link </div>
<div>-pthread -L/usr/lib/openmpi/lib -lmpi_cxx -lmpi -lopen-rte -lopen-pal -ldl -Wl,--export-dynamic -lnsl -lutil -lm -ldl</div></div><div><br></div><div>so even though mpi-default-dev point to lam, if openmpi gets also installed it is in practice the default implementation</div>
<div>on sparc !</div><div><br></div><div>To my opinion this is a bug in the mpi system. Adam ? Others ?</div><div><br></div><div><div>There are 3 choices for the alternative mpi (providing /usr/include/mpi).</div><div><br>
</div><div> Selection Path Priority Status</div><div>------------------------------------------------------------</div><div>* 0 /usr/lib/openmpi/include 40 auto mode</div><div>
1 /usr/include/lam 30 manual mode</div><div> 2 /usr/lib/mpich/include 10 manual mode</div><div> 3 /usr/lib/openmpi/include 40 manual mode</div></div>
<div><br></div><br><div class="gmail_quote">On Wed, Jun 9, 2010 at 8:53 PM, Christophe Prud'homme <span dir="ltr"><<a href="mailto:prudhomm@debian.org">prudhomm@debian.org</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">
I tried, without success so far, to help cmake(FindMPI.cmake) find the proper mpi implementation.<div>It still finds openmpi which breaks linkage with boost::mpi<br><br></div><div>Best regards</div><div>C.<div><div></div>
<div class="h5"><br><div class="gmail_quote">
On Wed, Jun 9, 2010 at 12:08 PM, Adam C Powell IV <span dir="ltr"><<a href="mailto:hazelsct@debian.org" target="_blank">hazelsct@debian.org</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div>On Mon, 2010-06-07 at 02:25 -0500, Steve M. Robbins wrote:<br>
> On Mon, Jun 07, 2010 at 09:02:41AM +0200, Christophe Prud'homme wrote:<br>
><br>
> > On my side life depends solely on mpi-default-dev, it seems that some other<br>
> > package don't (e.g. hdf5), isn't it a problem ?<br>
><br>
> Yes, something like that is likely the problem.<br>
><br>
> Note that libhdf5-mpi-dev is supposed to alleviate that problem as it<br>
> depends on the default MPI version. Installing that package should<br>
> not pull in any non-default MPI packages, IMHO. Adam: any comment?<br>
<br>
</div>Indeed, that's the idea: Build-Depend on libhdf5-mpi-dev and<br>
mpi-default-dev and you should have a consistent MPI implementation<br>
across both.<br>
<br>
That said, the LAM HDF5 implementation seems to be missing a couple of<br>
libraries, such that for example PETSc doesn't build on architectures<br>
where LAM is the default. I disabled PETSc HDF5 support on those<br>
arches, but haven't investigated further.<br>
<br>
-Adam<br>
<font color="#888888">--<br>
GPG fingerprint: D54D 1AEE B11C CE9B A02B C5DD 526F 01E8 564E E4B6<br>
<br>
Engineering consulting with open source tools<br>
<a href="http://www.opennovation.com/" target="_blank">http://www.opennovation.com/</a><br>
</font></blockquote></div><br></div></div></div>
</blockquote></div><br></div>