<p>Unfortunately tracking down problems with mpi is not very easy. I suggest that you check some of the tests that come with you mpi installation to verify that they work, and there may be some standard tests in scalapck as well. It is most likely that there is something wrong in libraries, how mpi is called in parallel_options or similar. The parallel version of lapw0 is simpler, does lapw0 -p work (with an appropriate .machines file)?</p>
<div class="gmail_quote">On Jan 4, 2011 2:07 AM, "Markus Kaukonen" <<a href="mailto:markus.kaukonen@iki.fi">markus.kaukonen@iki.fi</a>> wrote:<br type="attribution">> Dear Wien2K developers,<br>> <br>
> I managed to get in file uplapw1.error<br>> ---begin file----<br>> ** Error in Parallel LAPW1<br>> ** LAPW1 STOPPED at Tue Jan 4 09:50:31 EET 2011<br>> ** check ERROR FILES!<br>> Error in LAPW1<br>
> 'Unknow' - SIGSEGV, contact developers<br>> ---end file----<br>> <br>> and in gra111.scf1up_1<br>> ---begin file----<br>> ATOMIC SPHERE DEPENDENT PARAMETERS FOR ATOM C 1<br>> :e__0001: OVERALL ENERGY PARAMETER IS 0.3000<br>
> OVERALL BASIS SET ON ATOM IS LAPW<br>> :E0_0001: E( 0)= -0.7800 E(BOTTOM)= -2.976 E(TOP)= -200.000<br>> APW+lo<br>> :E0_0001: E( 0)= 0.3000<br>> LOCAL ORBITAL<br>
> :E1_0001: E( 1)= 0.3000<br>> APW+lo<br>> <br>> ATOMIC SPHERE DEPENDENT PARAMETERS FOR ATOM C 1<br>> :e__0002: OVERALL ENERGY PARAMETER IS 0.3000<br>> OVERALL BASIS SET ON ATOM IS LAPW<br>
> :E0_0002: E( 0)= -0.7800 E(BOTTOM)= -2.976 E(TOP)= -200.000<br>> APW+lo<br>> :E0_0002: E( 0)= 0.3000<br>> LOCAL ORBITAL<br>> :E1_0002: E( 1)= 0.3000<br>> APW+lo<br>
> <br>> MPI-parallel calculation using 5 processors<br>> Scalapack processors array (row,col): 5 1<br>> ---end file----<br>> <br>> The serial version runs fine with identical input (graphene with 2<br>
> atom unit cell).<br>> Parallel and serial version are compiled with portland compilers<br>> (pgi/10.2-0) and acml libraries were used for BLAS and LAPACK.<br>> for parallel version fftw-2.1.5 scalapack and blacs were compiled by pgi/10.2-0<br>
> <br>> Question: is there an easy fix or should one try another compiler?<br>> <br>> terveisin, Markus<br>> <br>> <br>> <br>> -- <br>> --www=<a href="http://www.iki.fi/markus.kaukonen">http://www.iki.fi/markus.kaukonen</a><br>
> --<a href="mailto:Markus.Kaukonen@iki.fi">Markus.Kaukonen@iki.fi</a><br>> --office: N102 Nano building FIN-02015 TKK<br>> --home: Viinirinne 3 F 12, 02630 Espoo, FIN<br>> --tel: h 045-1242068, w 4518694, 050-5112785<br>
> --Rikos ei kannata, eika maatalous<br>> --Suomessa. (Paimio 1998) ---<br>> _______________________________________________<br>> Wien mailing list<br>> <a href="mailto:Wien@zeus.theochem.tuwien.ac.at">Wien@zeus.theochem.tuwien.ac.at</a><br>
> <a href="http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien">http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien</a><br></div>