<html><body><div style="font-family: arial, helvetica, sans-serif; font-size: style=" font-size:="" 12pt;="" color:="" #000000"=""><div><br></div><div><div>Hello Xin,</div><div><br></div><div>Thank you for your reply. I have made a new thread. Below is the k-points obtained from kpoints.x for BCC Bravais Lattice, for a 4 4 4 0 0 0 mesh, considering the 48 symmetries. I hope this will be the same for the case with calculations for ibrav=3, with BCC primitive cell with one atom per cell.</div><div>8 (=number of k_points)</div><div><br></div><div> 1 0.0000000 0.0000000 0.0000000 1.00</div><div> 2 0.2500000 0.0000000 0.2500000 12.00</div><div> 3 0.5000000 0.0000000 0.5000000 6.00</div><div> 4 0.2500000 0.2500000 0.5000000 24.00</div><div> 5 0.0000000 0.0000000 0.5000000 6.00</div><div> 6 0.2500000 0.0000000 0.7500000 12.00</div><div> 7 -0.5000000 0.5000000 0.5000000 2.00</div><div> 8 0.0000000 0.0000000 1.0000000 1.00</div><div><br></div><div><br></div><div>Now let us compare with the k_points that were used in your calculation when a BCC supercell consisting of BCC Unit Cells is used as input with ibrav=0. So, I request you if you can kindly post the k_points considered in your calculation. Also you may kindly include your input script once more so that later on somebody can follow the discussion. </div><div><br></div><div>Thanking you,</div><div>Best regards,</div><div>Krishnendu</div></div><div><br></div><div><br></div><hr id="zwchr" data-marker="__DIVIDER__"><div data-marker="__HEADERS__"><b>From: </b>"xin tlg jin" <xin.tlg.jin@outlook.com><br><b>To: </b>users@lists.quantum-espresso.org<br><b>Cc: </b>"KRISHNENDU MUKHERJEE" <krishnendu@nmlindia.org><br><b>Sent: </b>Monday, October 31, 2022 12:28:24 PM<br><b>Subject: </b>Re: [QE-users] Parallel computing of QE7.1 vc-relax crashes when using large number of processors<br></div><br><div data-marker="__QUOTED_TEXT__">Hello Krishnendu,<br>
<br>
Thank you for the message<br>
Yes, we can start a separate thread. I didn't realize that there
could be such a problem. <br>
Actually, the reason that I use ibrav=0 for BCC W is that I want to
put some interstitials inside if the program works.<br>
<br>
Best regards,<br>
Xin<br>
<br>
<br>
<div>On 30/10/2022 05:35, KRISHNENDU
MUKHERJEE wrote:<br>
</div>
<blockquote>
<div style="font-family:'arial' , 'helvetica' , sans-serif">
<div><br>
</div>
<div>
<div>Dear Xin Jin,</div>
<div><br>
</div>
<div> Sorry that I have another matter to discuss. If you wish
we may start a separate thread on the subject I am eager to
discuss. I have some concern about your script. If you want
to do calculation for BCC W you may need to use ibrav=3.
Note that for ibrav=3 the (primitive) lattice vectors are in
the form:</div>
<div><br>
</div>
<div>ibrav=3 cubic I (bcc)</div>
<div> v1 = (a/2)(1,1,1), v2 = (a/2)(-1,1,1), v3 =
(a/2)(-1,-1,1) which is inbuilt in QE (so that you need not
input CELL_PARAMETERS) and you only need to put an atom in
the position 0.00 0.00 0.00.</div>
<div>With those input QE determines the k_points for BCC
Bravais Lattice.</div>
<div><br>
</div>
<div>You have input the atomic positions in terms of supercell
made of BCC unit cell. And you have put ibrav=0. With those
input I am afraid most probably the k_points generated would
be that of a Simple Cubic structure.</div>
<div>What we can do is next week I would generate the k_points
for BCC with a mesh of 4 4 4 0 0 0 and post it. And you can
see whether it is matching with the k_points considered in
your calculation and discuss further.</div>
<div><br>
</div>
<div>Thank you,</div>
<div>Best regards,</div>
<div>Krishnendu</div>
</div>
<div><br>
</div>
<div>------------------------------------------------------------------------------------------------------------------------------------------</div>
<div><br>
</div>
<div>
<div>Xin Jin wrote on 28/Oct/2022 </div>
<div><br>
</div>
<div><br>
</div>
<div>Dear Quantum Espresso Forum,</div>
<div><br>
</div>
<div>I encountered a problem related to the parallel computing
using QE7.1 </div>
<div>for vc-relax.</div>
<div><br>
</div>
<div>I was trying to perform a vc-relax for a 3*3*3 BCC
tungsten super cell. </div>
<div>The code works fine for non-parallel computing, also
works fine for </div>
<div>parallel computing if the number of processors is smaller
than 10.</div>
<div><br>
</div>
<div>However, if the number of processors is larger than 10, I
will get </div>
<div>following MPI error:</div>
<div>/*** An error occurred in MPI_Comm_free//</div>
<div>//*** reported by process [3585895498,2]//</div>
<div>//*** on communicator MPI_COMM_WORLD//</div>
<div>//*** MPI_ERR_COMM: invalid communicator//</div>
<div>//*** MPI_ERRORS_ARE_FATAL (processes in this
communicator will now abort,//</div>
<div>//*** and potentially your MPI job)/</div>
<div><br>
</div>
<div>For parallel computing, I am using
/OpenMPI/3.1.4-gcccuda/. (In </div>
<div>addition, it seems that If I use OpenMPI V4, the
simulation speed will </div>
<div>be much slower than that of V3.)</div>
<div><br>
</div>
<div>Another thing is that, if I decrease the size of the
supper cell, for </div>
<div>example to 2*2*2, then there is no problem in the
parallel computing </div>
<div>even if I use more than 30 processors.</div>
<div><br>
</div>
<div>Could you help me look at this problem, please?</div>
<div><br>
</div>
<div>The input for QE can be found below.</div>
<div><br>
</div>
<div>Thank you in advance!</div>
<div><br>
</div>
<div>Xin Jin</div>
<div><br>
</div>
<div>/&control//</div>
<div>//</div>
<div>// calculation='vc-relax' //</div>
<div>// restart_mode='from_scratch', //</div>
<div>// prefix='W_relax', //</div>
<div>// pseudo_dir="../../PP_files",//</div>
<div>// outdir='./'//</div>
<div>//</div>
<div>// ///</div>
<div>////</div>
<div>//</div>
<div>// &system//</div>
<div>// ibrav= 0, //</div>
<div>// celldm(1)=5.972,//</div>
<div>// nat= 54, //</div>
<div>// ntyp= 1,//</div>
<div>// ecutwfc = 50,//</div>
<div>// ecutrho = 500,//</div>
<div>// occupations='smearing', smearing='mp',
degauss=0.06//</div>
<div>// ///</div>
<div>//</div>
<div>// &electrons//</div>
<div>// diagonalization='david',//</div>
<div>// conv_thr = 1.0d-8,//</div>
<div>// mixing_beta = 0.5,//</div>
<div>// ///</div>
<div>////</div>
<div>// &ions//</div>
<div>// ///</div>
<div>//</div>
<div>// &cell//</div>
<div>// press = 0.0,//</div>
<div>// ///</div>
<div>////</div>
<div>//ATOMIC_SPECIES//</div>
<div>// W 183.84 W.pbe-spn-kjpaw_psl.1.0.0.UPF//</div>
<div>////</div>
<div>//CELL_PARAMETERS {alat}//</div>
<div>// 3.0 0.0 0.0//</div>
<div>// 0.0 3.0 0.0//</div>
<div>// 0.0 0.0 3.0 //</div>
<div>////</div>
<div>//ATOMIC_POSITIONS {alat}//</div>
<div>//W 0.00000 0.00000 0.00000//</div>
<div>//W 0.50000 0.50000 0.50000//</div>
<div>//W 1.00000 0.00000 0.00000//</div>
<div>//W 1.50000 0.50000 0.50000//</div>
<div>//W 2.00000 0.00000 0.00000//</div>
<div>//W 2.50000 0.50000 0.50000//</div>
<div>//W 0.00000 1.00000 0.00000//</div>
<div>//W 0.50000 1.50000 0.50000//</div>
<div>//W 1.00000 1.00000 0.00000//</div>
<div>//W 1.50000 1.50000 0.50000//</div>
<div>//W 2.00000 1.00000 0.00000//</div>
<div>//W 2.50000 1.50000 0.50000//</div>
<div>//W 0.00000 2.00000 0.00000//</div>
<div>//W 0.50000 2.50000 0.50000//</div>
<div>//W 1.00000 2.00000 0.00000//</div>
<div>//W 1.50000 2.50000 0.50000//</div>
<div>//W 2.00000 2.00000 0.00000//</div>
<div>//W 2.50000 2.50000 0.50000//</div>
<div>//W 0.00000 0.00000 1.00000//</div>
<div>//W 0.50000 0.50000 1.50000//</div>
<div>//W 1.00000 0.00000 1.00000//</div>
<div>//W 1.50000 0.50000 1.50000//</div>
<div>//W 2.00000 0.00000 1.00000//</div>
<div>//W 2.50000 0.50000 1.50000//</div>
<div>//W 0.00000 1.00000 1.00000//</div>
<div>//W 0.50000 1.50000 1.50000//</div>
<div>//W 1.00000 1.00000 1.00000//</div>
<div>//W 1.50000 1.50000 1.50000//</div>
<div>//W 2.00000 1.00000 1.00000//</div>
<div>//W 2.50000 1.50000 1.50000//</div>
<div>//W 0.00000 2.00000 1.00000//</div>
<div>//W 0.50000 2.50000 1.50000//</div>
<div>//W 1.00000 2.00000 1.00000//</div>
<div>//W 1.50000 2.50000 1.50000//</div>
<div>//W 2.00000 2.00000 1.00000//</div>
<div>//W 2.50000 2.50000 1.50000//</div>
<div>//W 0.00000 0.00000 2.00000//</div>
<div>//W 0.50000 0.50000 2.50000//</div>
<div>//W 1.00000 0.00000 2.00000//</div>
<div>//W 1.50000 0.50000 2.50000//</div>
<div>//W 2.00000 0.00000 2.00000//</div>
<div>//W 2.50000 0.50000 2.50000//</div>
<div>//W 0.00000 1.00000 2.00000//</div>
<div>//W 0.50000 1.50000 2.50000//</div>
<div>//W 1.00000 1.00000 2.00000//</div>
<div>//W 1.50000 1.50000 2.50000//</div>
<div>//W 2.00000 1.00000 2.00000//</div>
<div>//W 2.50000 1.50000 2.50000//</div>
<div>//W 0.00000 2.00000 2.00000//</div>
<div>//W 0.50000 2.50000 2.50000//</div>
<div>//W 1.00000 2.00000 2.00000//</div>
<div>//W 1.50000 2.50000 2.50000//</div>
<div>//W 2.00000 2.00000 2.00000//</div>
<div>//W 2.50000 2.50000 2.50000//</div>
<div>//</div>
<div>//K_POINTS {automatic}//</div>
<div>//4 4 4 0 0 0//</div>
<div>/</div>
<div>-------------- next part --------------</div>
<div>An HTML attachment was scrubbed...</div>
<div>URL:
<a href="http://lists.quantum-espresso.org/pipermail/users/attachments/20221028/87b8c648/attachment.html" target="_blank" rel="nofollow noopener noreferrer"><http://lists.quantum-espresso.org/pipermail/users/attachments/20221028/87b8c648/attachment.html></a></div>
</div>
<div><br>
</div>
</div>
<br>
<br>
</blockquote>
<br><br></div></div>
<br>
<br></body></html>