<div dir="ltr"><div class="gmail_default" style="font-family:verdana,sans-serif;color:#000000">For my own curiosity, is it 40,000 k-points or 40 k-points?</div><div class="gmail_default" style="font-family:verdana,sans-serif;color:#000000"><br></div><div class="gmail_default" style="font-family:verdana,sans-serif;color:#000000">N.B., as Peter suggested, did you try using mpi, which would be four of <span style="font-family:Arial,Helvetica,sans-serif;color:rgb(34,34,34)">nmr_integ:localhost:2</span></div><div class="gmail_default" style="">I suspect (but might be wrong) that this will reduce you memory useage by a factor of 2, and will only be slightly slower than what you have. If needed you can also go to 4 mpi. Of course you have to have compiled it...</div><div class="gmail_default" style=""><br></div><div class="gmail_default" style="">N.N.B., you presumably realise that you are using 16 cores for lapw1, as each k-point has 2 cores.</div><div class="gmail_default" style="font-family:verdana,sans-serif;color:#000000"><br></div><div class="gmail_default" style="font-family:verdana,sans-serif;color:#000000"><br></div></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Mon, May 13, 2024 at 4:00 PM Michael Fechtelkord via Wien <<a href="mailto:wien@zeus.theochem.tuwien.ac.at">wien@zeus.theochem.tuwien.ac.at</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">Hello all,<br>
<br>
<br>
as far as I can see it, a job with 8 cores may be faster, but uses <br>
double of the space on scratch (8 partial nmr vectors with size <br>
depending on the kmesh per direction eg. nmr_mqx instead of 4 partial <br>
vectors) and that also doubles the RAM usage of the NMR current <br>
calculation because 8 partial vectors per direction are used.<br>
<br>
I will try the -quota 8 option, but currently it seems that calculations <br>
on eight cores are at high risk to crash because of the memory and <br>
scratch space it needs and that already for 40k points. I never had <br>
problems with calculations on 4 cores even with only 64 GB RAM and 1000k <br>
points.<br>
<br>
<br>
Best regards,<br>
<br>
Michael<br>
<br>
<br>
Am 12.05.2024 um 18:02 schrieb Michael Fechtelkord via Wien:<br>
> It shows EXECUTING: /usr/local/WIEN2k/nmr_mpi -case MS_2M1_Al2 <br>
> -mode current -green -scratch /scratch/WIEN2k/ -noco<br>
><br>
> in all cases and in htop the values I provided below.<br>
><br>
><br>
> Best regards,<br>
><br>
> Michael<br>
><br>
><br>
> Am 12.05.2024 um 16:01 schrieb Peter Blaha:<br>
>> This makes sense.<br>
>> Please let me know if it shows<br>
>><br>
>> EXECUTING: /usr/local/WIEN2k/nmr_mpi -case MS_2M1_Al2 -mode <br>
>> current -green -scratch /scratch/WIEN2k/ -noco<br>
>><br>
>> or only nmr -case ...<br>
>><br>
>> In any case, it is running correctly.<br>
>><br>
>> PS: I know that also the current step needs a lot of memory, after <br>
>> all it needs to read the eigenvectors of all eigenvalues, ...<br>
>><br>
>> PPS: -quota 8 (or 24) might help and still utilizing all cores, <br>
>> but I'm not sure if it would save enough memory in the current steps.<br>
>><br>
>><br>
>><br>
>> Am 12.05.2024 um 10:09 schrieb Michael Fechtelkord via Wien:<br>
>>> Hello all, hello Peter,<br>
>>><br>
>>><br>
>>> That is what is really running in the background (from htop: this is <br>
>>> a new job with 4 nodes but it was the same with 8 nodes -p 1 - 8), <br>
>>> so no nmr_mpi.<br>
>>><br>
>>><br>
>>> TIME+ Command<br>
>>><br>
>>> 96.0 14.9 19h06:05 /usr/local/WIEN2k/nmr -case MS_2M1_A12 -mode <br>
>>> current -green -scratch /scratch/WIEN2k/ -noco -p 3<br>
>>><br>
>>> 95.8 14.9 19h05:10 /usr/local/WIEN2k/nmr -case MS_2M1_A12 -mode <br>
>>> current -green -scratch /scratch/WIEN2k/ -noco -p 1<br>
>>><br>
>>> 95.1 14.9 19h06:00 /usr/local/WIEN2k/nmr -case MS_2M1_A12 -mode <br>
>>> current -green -scratch /scratch/WIEN2K/ -noco -p 2<br>
>>><br>
>>> 95.5 15.4 19h08:10 /usr/local/WIEN2k/nmr -case MS_2M1_A12 -mode <br>
>>> current -green -scratch /scratch/WIEN2k/ -noco -p 4<br>
>>><br>
>>> 94.6 14.9 18h35:33 /usr/local/WIEN2k/nmr -case MS_2M1_A12 -mode <br>
>>> current -green -scratch /scratch/WIEN2k/ -noco -p 3<br>
>>><br>
>>> 93.3 15.4 18h36:24 /usr/local/WIEN2k/nmr-case MS_2M1_Al2 -mode <br>
>>> current -green -scratch /scratch/WIEN2k/ -noco -p 4<br>
>>><br>
>>> 93.3 14.9 18h33:02 /usr/local/WIEN2k/nmr-case MS_2M1_A12 -mode <br>
>>> current -green -scratch/scratch/WIEN2k/ -noco -p2<br>
>>><br>
>>> 94.0 14.9 18h38:44 /usr/local/WIEN2k/nmr -case MS_2M1_A12 -mode <br>
>>> current -green -scratch /scratch/WIEN2k/ -noco -p 1<br>
>>><br>
>>><br>
>>> Regards,<br>
>>><br>
>>> Michael<br>
>>><br>
>>><br>
>>> Am 11.05.2024 um 20:10 schrieb Michael Fechtelkord via Wien:<br>
>>>> Hello Peter,<br>
>>>><br>
>>>><br>
>>>> I just use "x_nmr_lapw -p" and the rest is initiated by the nmr <br>
>>>> script. The Line "/usr/local/WIEN2k/nmr_mpi -case MS_2M1_Al2 -mode <br>
>>>> current -green -scratch /scratch/WIEN2k/ -noco " is just <br>
>>>> part of the whole procedure and not initiated by me manually.. (I <br>
>>>> only copied the last lines of the calculation).<br>
>>>><br>
>>>><br>
>>>> Best regards,<br>
>>>><br>
>>>> Michael<br>
>>>><br>
>>>><br>
>>>> Am 11.05.2024 um 18:08 schrieb Peter Blaha:<br>
>>>>> Hallo Michael,<br>
>>>>><br>
>>>>> I don't understand the line:<br>
>>>>><br>
>>>>> /usr/local/WIEN2k/nmr_mpi -case MS_2M1_Al2 -mode current <br>
>>>>> -green -scratch /scratch/WIEN2k/ -noco<br>
>>>>><br>
>>>>> The mode current should run only k-parallel, not in mpi ??<br>
>>>>><br>
>>>>> PS: The repetition of<br>
>>>>><br>
>>>>> nmr_integ:localhost is useless.<br>
>>>>><br>
>>>>> nmr mode integ runs only once (not k-parallel, sumpara has already <br>
>>>>> summed up the currents)<br>
>>>>><br>
>>>>> But one can use nmr_integ:localhost:8<br>
>>>>><br>
>>>>><br>
>>>>> Best regards<br>
>>>>><br>
>>>>> Am 11.05.2024 um 16:19 schrieb Michael Fechtelkord via Wien:<br>
>>>>>> Hello Peter,<br>
>>>>>><br>
>>>>>> this is the .machines file content:<br>
>>>>>><br>
>>>>>> granulartity:1<br>
>>>>>> omp_lapw0:8<br>
>>>>>> omp_global:2<br>
>>>>>> 1:localhost<br>
>>>>>> 1:localhost<br>
>>>>>> 1:localhost<br>
>>>>>> 1:localhost<br>
>>>>>> 1:localhost<br>
>>>>>> 1:localhost<br>
>>>>>> 1:localhost<br>
>>>>>> 1:localhost<br>
>>>>>> nmr_integ:localhost<br>
>>>>>> nmr_integ:localhost<br>
>>>>>> nmr_integ:localhost<br>
>>>>>> nmr_integ:localhost<br>
>>>>>> nmr_integ:localhost<br>
>>>>>> nmr_integ:localhost<br>
>>>>>> nmr_integ:localhost<br>
>>>>>> nmr_integ:localhost<br>
>>>>>><br>
>>>>>><br>
>>>>>> Best regards,<br>
>>>>>><br>
>>>>>> Michael<br>
>>>>>><br>
>>>>>><br>
>>>>>> Am 11.05.2024 um 14:58 schrieb Peter Blaha:<br>
>>>>>>> Hmm. ?<br>
>>>>>>><br>
>>>>>>> Are you using k-parallel AND mpi-parallel ?? This could <br>
>>>>>>> overload the machine.<br>
>>>>>>><br>
>>>>>>> How does the .machines file look like ?<br>
>>>>>>><br>
>>>>>>><br>
>>>>>>> Am 10.05.2024 um 18:15 schrieb Michael Fechtelkord via Wien:<br>
>>>>>>>> Dear all,<br>
>>>>>>>><br>
>>>>>>>><br>
>>>>>>>> the following problem occurs to me using the NMR part of WIEN2k <br>
>>>>>>>> (23.2) on a opensuse LEAP 15.5 Intel platform. WIEN2k was <br>
>>>>>>>> compiled using one-api 2024.1 ifort and gcc 13.2.1. I am using <br>
>>>>>>>> ELPA 2024.03.01, Libxc 6.22, fftw 3.3.10 and MPICH 4.2.1 and <br>
>>>>>>>> the one-api 2024.1 MKL libraries. The CPU is a I9 14900k with <br>
>>>>>>>> 24 cores where I use eight for the calculations. The RAM is 130 <br>
>>>>>>>> Gb and a swap file of 16 GB on a Samsung PCIE 4.0 NVME SSD. The <br>
>>>>>>>> BUS width is 5600 MT / s.<br>
>>>>>>>><br>
>>>>>>>> The structure is a layersilicate and to simulate the ratio of <br>
>>>>>>>> Si:Al = 3:1 I use a 1:1:2 supercell currently. The monoclinic <br>
>>>>>>>> symmetry of the new structure (original is C 2/c) is P 2/c and <br>
>>>>>>>> contains 40 atoms (K, Al, Si, O, and F).<br>
>>>>>>>><br>
>>>>>>>> I use 3 NMR LOs for K and O and 10 for Si, Al, and F (where I <br>
>>>>>>>> need the chemical shifts). The k mesh is 40k points.<br>
>>>>>>>><br>
>>>>>>>> The interesting thing is that the RAM is sufficient during NMR <br>
>>>>>>>> vector calculations (always under 100 Gb RAM occupied) and at <br>
>>>>>>>> the beginning of the electron current calculation. However, the <br>
>>>>>>>> RAM use increases to a critical point in the calculation and <br>
>>>>>>>> more and more data is outsourced into the SWAP File which is <br>
>>>>>>>> sometimes 80% occupied.<br>
>>>>>>>><br>
>>>>>>>> As you see this time only one core failed because of memory <br>
>>>>>>>> overflow. But using 48k points 3 cores crashed and so the whole <br>
>>>>>>>> current calculation. The reason is of the crash clear to me. <br>
>>>>>>>> But I do not understand, why the current calculation reacts so <br>
>>>>>>>> sensitive with so few atoms and a small k mesh. I made <br>
>>>>>>>> calculations with more atoms and a 1000K point mesh on 4 cores <br>
>>>>>>>> .. they worked fine. So can it be that the Intel MKL library is <br>
>>>>>>>> the source of failure? So I better get back to 4 cores, even <br>
>>>>>>>> with longer calculation times?<br>
>>>>>>>><br>
>>>>>>>> Have all a nice weekend!<br>
>>>>>>>><br>
>>>>>>>><br>
>>>>>>>> Best wishes from<br>
>>>>>>>><br>
>>>>>>>> Michael Fechtelkord<br>
>>>>>>>><br>
>>>>>>>> -----------------------------------------------<br>
>>>>>>>><br>
>>>>>>>> cd ./ ... x lcore -f MS_2M1_Al2<br>
>>>>>>>> CORE END<br>
>>>>>>>> 0.685u 0.028s 0:00.71 98.5% 0+0k 2336+16168io 5pf+0w<br>
>>>>>>>><br>
>>>>>>>> lcore .... ready<br>
>>>>>>>><br>
>>>>>>>><br>
>>>>>>>> EXECUTING: /usr/local/WIEN2k/nmr_mpi -case MS_2M1_Al2 <br>
>>>>>>>> -mode current -green -scratch /scratch/WIEN2k/ -noco<br>
>>>>>>>><br>
>>>>>>>> [1] 20253<br>
>>>>>>>> [2] 20257<br>
>>>>>>>> [3] 20261<br>
>>>>>>>> [4] 20265<br>
>>>>>>>> [5] 20269<br>
>>>>>>>> [6] 20273<br>
>>>>>>>> [7] 20277<br>
>>>>>>>> [8] 20281<br>
>>>>>>>> [8] + Abgebrochen ( cd $dir; $exec2 >> <br>
>>>>>>>> nmr.out.${loop} ) >& nmr.err.$loop<br>
>>>>>>>> [7] + Fertig ( cd $dir; $exec2 >> <br>
>>>>>>>> nmr.out.${loop} ) >& nmr.err.$loop<br>
>>>>>>>> [6] + Fertig ( cd $dir; $exec2 >> <br>
>>>>>>>> nmr.out.${loop} ) >& nmr.err.$loop<br>
>>>>>>>> [5] + Fertig ( cd $dir; $exec2 >> <br>
>>>>>>>> nmr.out.${loop} ) >& nmr.err.$loop<br>
>>>>>>>> [4] + Fertig ( cd $dir; $exec2 >> <br>
>>>>>>>> nmr.out.${loop} ) >& nmr.err.$loop<br>
>>>>>>>> [3] + Fertig ( cd $dir; $exec2 >> <br>
>>>>>>>> nmr.out.${loop} ) >& nmr.err.$loop<br>
>>>>>>>> [2] + Fertig ( cd $dir; $exec2 >> <br>
>>>>>>>> nmr.out.${loop} ) >& nmr.err.$loop<br>
>>>>>>>> [1] + Fertig ( cd $dir; $exec2 >> <br>
>>>>>>>> nmr.out.${loop} ) >& nmr.err.$loop<br>
>>>>>>>><br>
>>>>>>>> EXECUTING: /usr/local/WIEN2k/nmr -case MS_2M1_Al2 -mode <br>
>>>>>>>> sumpara -p 8 -green -scratch /scratch/WIEN2k/<br>
>>>>>>>><br>
>>>>>>>><br>
>>>>>>>> current .... ready<br>
>>>>>>>><br>
>>>>>>>><br>
>>>>>>>> EXECUTING: mpirun -np 1 -machinefile .machine_nmrinteg <br>
>>>>>>>> /usr/local/WIEN2k/nmr_mpi -case MS_2M1_Al2 -mode integ -green<br>
>>>>>>>><br>
>>>>>>>><br>
>>>>>>>> nmr: integration ... done in 4032.3s<br>
>>>>>>>><br>
>>>>>>>><br>
>>>>>>>> stop<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" rel="noreferrer" target="_blank">http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien</a><br>
> SEARCH the MAILING-LIST at: <br>
> <a href="http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html" rel="noreferrer" target="_blank">http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html</a><br>
<br>
-- <br>
Dr. Michael Fechtelkord<br>
<br>
Institut für Geologie, Mineralogie und Geophysik<br>
Ruhr-Universität Bochum<br>
Universitätsstr. 150<br>
D-44780 Bochum<br>
<br>
Phone: +49 (234) 32-24380<br>
Fax: +49 (234) 32-04380<br>
Email: <a href="mailto:Michael.Fechtelkord@ruhr-uni-bochum.de" target="_blank">Michael.Fechtelkord@ruhr-uni-bochum.de</a><br>
Web Page: <a href="https://www.ruhr-uni-bochum.de/kristallographie/kc/mitarbeiter/fechtelkord/" rel="noreferrer" target="_blank">https://www.ruhr-uni-bochum.de/kristallographie/kc/mitarbeiter/fechtelkord/</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" rel="noreferrer" target="_blank">http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien</a><br>
SEARCH the MAILING-LIST at: <a href="http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html" rel="noreferrer" target="_blank">http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html</a><br>
</blockquote></div><br clear="all"><div><br></div><span class="gmail_signature_prefix">-- </span><br><div dir="ltr" class="gmail_signature"><div dir="ltr">Professor Laurence Marks (Laurie)<div>Northwestern University<br><div><a href="http://www.numis.northwestern.edu" target="_blank">Webpage</a> and <a href="http://scholar.google.com/citations?user=zmHhI9gAAAAJ&hl=en" target="_blank">Google Scholar link</a></div><div>"Research is to see what everybody else has seen, and to think what nobody else has thought", Albert Szent-Györgyi</div></div></div></div>