I am sorry I am quite a novice about k-point parallel in wien2k, it seems I still have wrong idea. <br>If neither OpenMP nor any MPI is needed, then how the parallelization is realized? <br><br>The manual (sec. 5.5.1 k-Point Parallelization) says "Parallelization is achieved on the k-point level by distributing subsets of the k-mesh to different processors and subsequent summation of the results". In my mind it means using either OpenMP or MPI to distribute to different processors, what other means it refers to? <br>
<br>Long <br><br><br><br><br><div class="gmail_quote">On Fri, Feb 22, 2013 at 2:31 PM, Laurence Marks <span dir="ltr"><<a href="mailto:L-marks@northwestern.edu" target="_blank">L-marks@northwestern.edu</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">No, wrong. Neither OpenMP nor mpi (any flavor, MPICH, MPICH2, MVAPICH...) are needed.<br><br><div class="gmail_quote"><div class="im">
On Fri, Feb 22, 2013 at 1:28 PM, Long Zhang <span dir="ltr"><<a href="mailto:lzhang6789@gmail.com" target="_blank">lzhang6789@gmail.com</a>></span> wrote:<br>
</div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div><div class="im">
Thank you for the information. <br>
I think I had wrong idea about k-point parallel. If it doesn't use OpenMP then it's perfectly fine for me. I'll drop the idea of finegrained parallel in this installation.
<br>
<br>
Just want to confirm, for using k-point parallel only, MPICH2 is still required, is that right?
<br>
<br>
Thanks.<br></div>
Long <br>
<br>
<br>
<br>
<br>
<div class="gmail_quote"><div class="im">On Fri, Feb 22, 2013 at 11:04 AM, Laurence Marks <span dir="ltr">
<<a href="mailto:L-marks@northwestern.edu" target="_blank">L-marks@northwestern.edu</a>></span> wrote:<br>
</div><div><div class="h5"><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
As Peter said, the simplest approach is just k-point parallel which<br>
does not need OpenMP. The rate limiting steps are lapack calls which<br>
the Intel mkl is good for.<br>
<br>
Whether mpi is good for a small cluster such as you have in not so<br>
clear. Peter has benchmarked the mpi version and his conclusion is<br>
that it is not useful; I have benchmarked on my cluster and I have<br>
consistently found the mpi versions to be faster than the Intel<br>
threading. I suspect that it really depends upon many details specific<br>
to the systems, for instance cross-core optimizations, network and<br>
memory speed....<br>
<br>
Not a complete answer, but for certain the simplest approach is to<br>
just use k-point parallel.<br>
<div><br>
On Fri, Feb 22, 2013 at 9:55 AM, Long Zhang <<a href="mailto:lzhang6789@gmail.com" target="_blank">lzhang6789@gmail.com</a>> wrote:<br>
><br>
> That's true I agree. If the k-point parallel version does not need mpi, then<br>
> it's OpenMP I suppose (please correct me if it's not). But I really don't<br>
> want to run OpenMP because it will mess my other applications up.<br>
><br>
> I am building it on a small server of 2 Opteron 12-core cpus. It's just a<br>
> dual-cpu system, not a cluster. I am not sure if I can gain enough speed-up<br>
> by using 24 cores. Anyway if I want a little k-point parallel, can I choice<br>
> to use mpi? or there is no such choice at all?<br>
><br>
> Thank you very much.<br>
> Long<br>
><br>
><br>
> On Fri, Feb 22, 2013 at 3:07 AM, Peter Blaha <<a href="mailto:pblaha@theochem.tuwien.ac.at" target="_blank">pblaha@theochem.tuwien.ac.at</a>><br>
> wrote:<br>
>><br>
</div>
<div>
<div>>> I keep saying this all the time:<br>
>><br>
>> The mpi-parallel version needs special hardware !<br>
>><br>
>> It is not of much use on a single quadcore-cpu.<br>
>> A cluster needs a fast network like infiniband.<br>
>><br>
>> Only then the installation of the mpi-version make sense.<br>
>><br>
>> We have a k-point parallel version, which does not need mpi, but is<br>
>> perfect for a small cluster with a common NFS filesystem.<br>
>><br>
>><br>
>> Am 22.02.2013 06:20, schrieb Long Zhang:<br>
>>><br>
>>><br>
>>> Hello,<br>
>>><br>
>>> I am trying to install the current version 12.1 on Ubuntu 12.04<br>
>>> (64-bit). I met some problem related to mpi, can someone please give me<br>
>>> some hint? Thanks in advance.<br>
>>><br>
>>> The situation is like this, on a freshly installed Ubuntu 12.04 64bit, I<br>
>>> installed current version Intel Fortran Composer XE 2013.2.146, and<br>
>>> installed MPICH2 and FFTW3 via Ubuntu Software Center.<br>
>>><br>
>>> Before run the siteconfig_lapw, I checked<br>
>>> which ifort<br>
>>> /opt/intel/composer_xe_2013.2.146/bin/intel64/ifort<br>
>>> which mpif90<br>
>>> /usr/bin/mpif90<br>
>>><br>
>>> and did the following 3 lines in the terminal<br>
>>> source /opt/intel/composer_xe_2013.2.146/bin/compilervars.sh intel64<br>
>>> export<br>
>>><br>
>>> LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/opt/intel/composer_xe_2013.2.146/mkl/lib/intel64<br>
>>> export MKLROOT=/opt/intel/composer_xe_2013.2.146/mkl/<br>
>>><br>
>>> Then I run the siteconfig_lapw, and selected system as "ifort12+mkl"<br>
>>> (the first one), selected compilers as "ifort" and "cc", selected<br>
>>> "finegrained parallel" and "mpif90", and got error message like this:<br>
>>><br>
>>> ----------------------------------------------- cut1<br>
>>> -----------------------------------------------<br>
>>> Compile time errors (if any) were:<br>
>>> SRC_hf/compile.msg:make[1]: *** [modules.o] Error 1<br>
>>> SRC_hf/compile.msg:make: *** [rp] Error 2<br>
>>> SRC_hf/compile.msg:make[1]: *** [modules.o] Error 1<br>
>>> SRC_hf/compile.msg:make: *** [cp] Error 2<br>
>>> SRC_lapw0/compile.msg:make[1]: *** [modules.o] Error 1<br>
>>> SRC_lapw0/compile.msg:make: *** [para] Error 2<br>
>>> SRC_lapw1/compile.msg:make[1]: *** [modules.o] Error 1<br>
>>> SRC_lapw1/compile.msg:make: *** [rp] Error 2<br>
>>> SRC_lapw1/compile.msg:make[1]: *** [modules.o] Error 1<br>
>>> SRC_lapw1/compile.msg:make: *** [cp] Error 2<br>
>>> SRC_lapw2/compile.msg:make[1]: *** [modules.o] Error 1<br>
>>> SRC_lapw2/compile.msg:make: *** [rp] Error 2<br>
>>> SRC_lapw2/compile.msg:make[1]: *** [modules.o] Error 1<br>
>>> SRC_lapw2/compile.msg:make: *** [cp] Error 2<br>
>>> ----------------------------------------------- cut1<br>
>>> -----------------------------------------------<br>
>>><br>
>>> I looked into SRC_lapw1/compile.msg, and the detailed error is:<br>
>>><br>
>>> ----------------------------------------------- cut2<br>
>>> -----------------------------------------------<br>
>>> mpif90 -FR -mp1 -w -prec_div -pc80 -pad -ip -DINTEL_VML -DFFTW3<br>
>>> -traceback -DParallel -c modules_tmp_.F<br>
>>><br>
>>> --------------------------------------------------------------------------<br>
>>> The Open MPI wrapper compiler was unable to find the specified compiler<br>
>>> gfortran in your PATH.<br>
>>><br>
>>> Note that this compiler was either specified at configure time or in<br>
>>> one of several possible environment variables.<br>
>>><br>
>>> --------------------------------------------------------------------------<br>
>>> make[1]: *** [modules.o] Error 1<br>
>>> make[1]: Leaving directory `/home/lzhang200/wien2k/SRC_lapw1'<br>
>>> make: *** [rp] Error 2<br>
>>> ----------------------------------------------- cut2<br>
>>> -----------------------------------------------<br>
>>><br>
>>> It seems OpenMPI was invoked and it looked for gfortran. I don't have<br>
>>> gfortran installed at all, but I found I have openmpi installed. I then<br>
>>> used Ubuntu Software Center to search openmpi and uninstalled related<br>
>>> packages. I restarted the system and repeated the siteconfig_lapw.<br>
>>> However I still get error message, not same but similar, it still looks<br>
>>> for gfortran, for examplethe SRC_lapw1/compile.msg shows:<br>
>>><br>
>>><br>
>>> ----------------------------------------------- cut3<br>
>>> -----------------------------------------------<br>
>>> mpif90 -FR -mp1 -w -prec_div -pc80 -pad -ip -DINTEL_VML -DFFTW3<br>
>>> -traceback -DParallel -c modules_tmp_.F<br>
>>> /usr/bin/mpif90: line 370: gfortran: command not found<br>
>>> make[1]: *** [modules.o] Error 127<br>
>>> make[1]: Leaving directory `/home/lzhang200/wien2k/SRC_lapw1'<br>
>>> make: *** [rp] Error 2<br>
>>> ----------------------------------------------- cut3<br>
>>> -----------------------------------------------<br>
>>><br>
>>> I did not specify to use gfortran anywhere in the setupof<br>
>>> siteconfig_lapw, I don't see why mpif90 keeps looking for it. How can I<br>
>>> fix this problem? Thanks a lot.<br>
>>><br>
>>> Regards,<br>
>>> Long Zhang<br>
>>><br>
>>><br>
>>> _______________________________________________<br>
>>> Wien mailing list<br>
>>> <a href="mailto:Wien@zeus.theochem.tuwien.ac.at" target="_blank">Wien@zeus.theochem.tuwien.ac.at</a><br>
>>> <a href="http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien" target="_blank">
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien</a><br>
>>><br>
>><br>
>> --<br>
>> Peter Blaha<br>
>> Inst.Materials Chemistry<br>
>> TU Vienna<br>
>> Getreidemarkt 9<br>
>> A-1060 Vienna<br>
>> Austria<br>
>> <a href="tel:%2B43-1-5880115671" value="+4315880115671" target="_blank">+43-1-5880115671</a><br>
>><br>
>> _______________________________________________<br>
>> Wien mailing list<br>
>> <a href="mailto:Wien@zeus.theochem.tuwien.ac.at" target="_blank">Wien@zeus.theochem.tuwien.ac.at</a><br>
>> <a href="http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien" target="_blank">
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien</a><br>
><br>
><br>
<br>
<br><span><font color="#888888">
<br>
</font></span></div><span><font color="#888888">
</font></span></div><span><font color="#888888">
<span><font color="#888888">--<br>
Professor Laurence Marks<br>
Department of Materials Science and Engineering<br>
Northwestern University<br>
<a href="http://www.numis.northwestern.edu" target="_blank">www.numis.northwestern.edu</a>
<a href="tel:1-847-491-3996" value="+18474913996" target="_blank">1-847-491-3996</a><br>
"Research is to see what everybody else has seen, and to think what<br>
nobody else has thought"<br>
Albert Szent-Gyorgi<br>
</font></span>
<div>
<div>_______________________________________________<br>
Wien mailing list<br>
<a href="mailto:Wien@zeus.theochem.tuwien.ac.at" target="_blank">Wien@zeus.theochem.tuwien.ac.at</a><br>
<a href="http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien" target="_blank">http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien</a><br>
</div>
</div>
</font></span></blockquote>
</div></div></div>
<br>
</div>
</blockquote></div><div class="HOEnZb"><div class="h5"><br><br clear="all"><div><br></div>-- <br>Professor Laurence Marks<br>Department of Materials Science and Engineering<br>Northwestern University<br><a href="http://www.numis.northwestern.edu" target="_blank">www.numis.northwestern.edu</a> <a href="tel:1-847-491-3996" value="+18474913996" target="_blank">1-847-491-3996</a><br>
"Research is to see what everybody else has seen, and to think what nobody else has thought"<br>Albert Szent-Gyorgi
</div></div><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" target="_blank">http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien</a><br>
<br></blockquote></div><br>