Thanks a lot Eyvaz for your reminder. Actually, every time before I run ./configure I load moduli "pgi52 and mpich-pgi". <br> <br> Best<br><br><b><i>Eyvaz Isaev <eyvaz_isaev@yahoo.com></i></b> wrote:<blockquote class="replbq" style="border-left: 2px solid rgb(16, 16, 255); margin-left: 5px; padding-left: 5px;"> Hi,<br><br>My apologies interfering. It might be that I did not<br>read the threat completely.<br><br>The problem looks like the same I had before on a<br>Linux cluster. It turned out that I had to load some<br>moduli before typing "./configure". They were mpich,<br>blas, lapack, fftw, and ifort. Sure, compiled<br>libraries have to be compatible with a compiler you<br>would like to use. <br>I.e. if you like ifort compiler choose intel_mpich,<br>but not gnu_mpich. <br><br>So, try the command "module available" and choose<br>these what you need using "module load XXXXXX". Then<br>./configure<br>It worked perfectly in my
case.<br><br>Bests,<br>Eyvaz.<br><br><br><br>--- stargmoon <stargmoon @yahoo.com=""> wrote:<br><br>> Thanks a lot Dr.Kohlmeyer, I will ask our sysadmin<br>> and let you know if I can successfully compile the<br>> parallel version.<br>> <br>> Best regards,<br>> <br>> Stargmoon<br>> <br>> Axel Kohlmeyer <akohlmey @cmm.chem.upenn.edu=""> wrote:<br>> On 9/18/06, stargmoon wrote:<br>> > Dear Dr.Kohlmeyer,<br>> ><br>> > Thanks for your reply.<br>> ><br>> > I am just one user of our cluster, that is, I am<br>> not the system<br>> > administrator. I do not know the detail of the<br>> installation of softwares<br>> <br>> so please _ask_ your sysadmin, this is the person<br>> 'in the know'<br>> about the details. as i wrote before, all<br>> machines/clusters are<br>> slightly different, so there is no way to predict<br>> what might be a problem.<br>> <br>> > there. But MPICH
was there, and can be launched by<br>> "module add". And I was<br>> > told that MPICH was installed ourselves, not<br>> bundled with the machines.<br>> <br>> so then your sysadmin _has_ to know...<br>> <br>> > The command line I used to do "configure" is<br>> "./configure<br>> > MPI_LIBS="-L/opt/mpi/tcp/mpich-pgi/lib -lmpich<br>> -lfmpich"<br>> ><br>> > Could you please give me more hints what I should<br>> do to figure out this<br>> > problem?<br>> <br>> as i wrote before, please try to compile and run<br>> _another_ MPI program,<br>> best one of the MPI tutorial examples, as they are<br>> rather trivial.<br>> once you get that working, let us know what command<br>> line you needed<br>> for the successful compile that produced a usable<br>> executable,<br>> we can look into getting QE compiled.<br>> <br>> ><br>> > By the way, have you ever used pathscale to<br>>
compile espresso? It seems to<br>> > be possible from the update information for<br>> espresso3.1.1. We have pathscale<br>> > and mpich for pathscale work for VASP on our<br>> cluster.<br>> <br>> i managed to do it a long time ago (including a few<br>> manual hacks).<br>> but i didn't have access to a machine with pathscale<br>> for quite a while.<br>> with the addition of the iotk library, probably a<br>> few more tweaks are/were<br>> needed. i would not expect a large difference<br>> between pathscale and PGI<br>> on AMD64 machines. most of the speed comes from the<br>> design of the<br>> cpu itself and usually numerical codes like QE are<br>> faster, if you _lower_<br>> the optimization (and especially avoid IPA/IPO and<br>> heuristic vectorization).<br>> QE already takes a lot of advantage of SIMD<br>> instructions through the<br>> use of optimized BLAS/LAPACK libraries (i.e. ACML on<br>>
AMD64).<br>> <br>> cheers,<br>> axel.<br>> <br>> ><br>> > Best,<br>> ><br>> > Stargmoon<br>> ><br>> > Axel Kohlmeyer wrote:<br>> > On 9/17/06, stargmoon wrote:<br>> > > Dear pwscf community,<br>> > ><br>> > > I tried to compile Espresso-3.1.1 recently on<br>> our PC cluster (AMD64).<br>> > > However, after I run ./configure, I am told that<br>> "Parallel environment not<br>> > > detected". I checked the config.log, since there<br>> is no problem in seaching<br>> > > for the MPI compilers (mpif90, mpif77 and<br>> mpicc), I think it must be the<br>> > MPI<br>> ><br>> > there are two stages of the search. a) whether the<br>> executables exists<br>> > and b) whether they can produce working binaries.<br>> ><br>> > > library problem. Therefore, I tried to set<br>> "MPI_LIBS" (there is only<br>> > >
libmpich.a in there) in the ./configure command<br>> line, but it did not work<br>> ><br>> > i would have expected a libfmpich.a, too.<br>> ><br>> > > either. Could anybody please tell me what kind<br>> of MPI libraries I have to<br>> > > point to the "configure" in order to get<br>> parallel compilation?<br>> ><br>> > this is impossible to tell, without knowing any<br>> details about your system.<br>> ><br>> > what parallel software are you using (it looks<br>> like MPICH) and did you<br>> > install it yourself or was it bundled with the<br>> system? do mpif77 and<br>> > mpif90 point to a sufficient fortran<br>> ><br>> > can you compile and run any of the (trivial) MPI<br>> example programs<br>> > that usually ship with MPI packages, and if yes,<br>> please describe the<br>> > commandline you use for that. based on that<br>> information, we
may<br>> > be able to help you.<br>> ><br>> > especially on linux machines, there are almost<br>> always a few kinks<br>> > to be worked out in the installation.<br>> ><br>> > regards,<br>> > axel.<br>> ><br>> > ><br>> > > Thanks in advance!<br>> > ><br>> > > stargmoon<br>> > ><br>> > ><br>> > > ________________________________<br>> > > Get your email and more, right on the new<br>> Yahoo.com<br>> > ><br>> > ><br>> ><br>> ><br>> > --<br>> ><br>><br>=======================================================================<br>> > Axel Kohlmeyer akohlmey@cmm.chem.upenn.edu<br>> http://www.cmm.upenn.edu<br>> > Center for Molecular Modeling -- University of<br>> Pennsylvania<br>> > Department of Chemistry, 231 S.34th Street,<br>> Philadelphia, PA 19104-6323<br>> > tel:
1-215-898-1582, fax: 1-215-573-6233,<br>> office-tel: 1-215-898-5425<br>> ><br>><br>=======================================================================<br>> > If you make something idiot-proof, the universe<br>> creates a better idiot.<br>> > _______________________________________________<br>> > Pw_forum mailing list<br>> > Pw_forum@pwscf.org<br>> > http://www.democritos.it/mailman/listinfo/pw_forum<br>> ><br>> ><br>> ><br>> > ________________________________<br>> > Stay in the know. Pulse on the new Yahoo.com.<br>> Check it out.<br>> ><br>> ><br>> <br>> <br>> -- <br>><br>=======================================================================<br>> Axel Kohlmeyer akohlmey@cmm.chem.upenn.edu <br>> http://www.cmm.upenn.edu<br>> Center for Molecular Modeling -- University of<br>> Pennsylvania<br>> Department of Chemistry, 231 S.34th
Street,<br>> Philadelphia, PA 19104-6323<br>> tel: 1-215-898-1582, fax: 1-215-573-6233, <br>> office-tel: 1-215-898-5425<br>><br>=======================================================================<br>> If you make something idiot-proof, the universe<br>> creates a better idiot.<br>> _______________________________________________<br>> Pw_forum mailing list<br>> Pw_forum@pwscf.org<br>> http://www.democritos.it/mailman/listinfo/pw_forum<br>> <br>> <br>=== message truncated ===<br><br><br>__________________________________________________<br>Do You Yahoo!?<br>Tired of spam? Yahoo! Mail has the best spam protection around <br>http://mail.yahoo.com <br>_______________________________________________<br>Pw_forum mailing list<br>Pw_forum@pwscf.org<br>http://www.democritos.it/mailman/listinfo/pw_forum<br></akohlmey></stargmoon></blockquote><br><p>
<hr size=1>Do you Yahoo!?<br>
Get on board. <a href="http://us.rd.yahoo.com/evt=40791/*http://advision.webevents.yahoo.com/mailbeta">You're invited</a> to try the new Yahoo! Mail.