Hi, <br>I have found a problem with a benchmark calculation with pw.x . It aborts when running with the -npool option.<br><br>However, it runs normally without the npool option. The calculation uses 34 k-points. <br>
<br><br>I have tried with <br>mpirun -np 32 pw.x -npool 2<br>mpirun -np 32 pw.x -npool 4<br>mpirun -np 4 pw.x -npool 2<br>mpirun -np 4 pw.x -npool 2 <br><br>in two machines, using OpenMPI and HPMPI, single nodes and multiples nodes. All the times it fails.<br>
<br>Looking at the output I see the following messages<br>
<br>%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%<br> %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%<br> from addusdens_r : error # 1<br> from addusdens_r : error # 1<br>
expected 360.00000000, found 101.25021916: wrong charge, increase ecutrho<br> from addusdens_r : error # 1<br> expected 360.00000000, found 101.25021916: wrong charge, increase ecutrho<br> expected 360.00000000, found 101.25021916: wrong charge, increase ecutrho<br>
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%<br><br> %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%<br> %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%<br>
stopping ...<br><br><br> stopping ...<br> stopping ...<br> %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%<br> from addusdens_r : error # 1<br> expected 360.00000000, found 101.25021916: wrong charge, increase ecutrho<br>
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%<br><br> stopping ...<br><br>the previous output was with -npool 4. Using -npool 2 I get outpits like this<br><br> %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%<br>
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%<br> from addusdens_r : error # 1<br> expected 360.00000000, found 180.00040683: wrong charge, increase ecutrho<br> from addusdens_r : error # 1<br>
expected 360.00000000, found 180.00040683: wrong charge, increase ecutrho<br><br>It looks like hving problems to sum the charge densities of the pools.<br><br>Here is the input (the pseudos are from the web site)<br>
<br> &CONTROL<br> calculation = 'scf' ,<br>
restart_mode = 'from_scratch' ,<br> outdir = '.' ,<br> pseudo_dir = '.' <br> prefix = 'cdsebench' ,<br> wfcdir = '/tmp',<br>
/<br> &SYSTEM<br> ibrav = 0,<br> celldm(1) = 1.8897261,<br> nat = 40,<br> ntyp = 2,<br> ecutwfc = 30.0 ,<br>
ecutrho = 180.0 ,<br>
occupations = 'smearing' ,<br> degauss = 0.02 ,<br> smearing = 'gaussian' ,<br>! qcutz=150., q2sigma=2.0, ecfixed=24.0<br> /<br> &ELECTRONS<br>
electron_maxstep = 60,<br> conv_thr = 1.0D-6 ,<br> startingpot = 'atomic' ,<br> startingwfc = 'random' ,<br>! mixing_mode = 'TF' ,<br>
mixing_beta = 0.7D0,<br> mixing_beta = 0.7D0,<br> diagonalization = 'david' ,<br> tqr = .true.<br> /<br><br>CELL_PARAMETERS hexagonal<br> 4.373836756 0.000000000 0.000000000<br>
-2.187115631 3.787739854 0.000000000<br> 0.000000000 0.000000000 71.411016248<br><br>ATOMIC_SPECIES<br> Cd 112.41000 Cd.pbe-van.UPF<br> Se 78.960000 Se.pbe-van.UPF<br><br>ATOMIC_POSITIONS (crystal)<br>
Cd 0.000000000 0.000000000 0.000000000<br>
Cd 0.000000000 0.000000000 0.100000000<br>Cd 0.000000000 0.000000000 0.200000000<br>Cd 0.000000000 0.000000000 0.300000000<br>Cd 0.000000000 0.000000000 0.400000000<br>Cd 0.000000000 0.000000000 0.500000000<br>
Cd 0.000000000 0.000000000 0.600000000<br>Cd 0.000000000 0.000000000 0.700000000<br>Cd 0.000000000 0.000000000 0.800000000<br>Cd 0.000000000 0.000000000 0.900000000<br>Cd 0.666663821 0.333312908 0.050000151<br>
Cd 0.666663821 0.333312908 0.150000151<br>Cd 0.666663821 0.333312908 0.250000151<br>Cd 0.666663821 0.333312908 0.350000151<br>Cd 0.666663821 0.333312908 0.450000151<br>Cd 0.666663821 0.333312908 0.550000151<br>
Cd 0.666663821 0.333312908 0.650000151<br>Cd 0.666663821 0.333312908 0.750000151<br>Cd 0.666663821 0.333312908 0.850000151<br>Cd 0.666663821 0.333312908 0.950000151<br>Se 0.000000000 0.000000000 0.037565413<br>
Se 0.000000000 0.000000000 0.137565413<br>Se 0.000000000 0.000000000 0.237565413<br>Se 0.000000000 0.000000000 0.337565413<br>Se 0.000000000 0.000000000 0.437565413<br>Se 0.000000000 0.000000000 0.537565413<br>
Se 0.000000000 0.000000000 0.637565413<br>Se 0.000000000 0.000000000 0.737565413<br>Se 0.000000000 0.000000000 0.837565413<br>Se 0.000000000 0.000000000 0.937565413<br>Se 0.666665290 0.333319516 0.087565394<br>
Se 0.666665290 0.333319516 0.187565394<br>Se 0.666665290 0.333319516 0.287565394<br>Se 0.666665290 0.333319516 0.387565394<br>Se 0.666665290 0.333319516 0.487565394<br>Se 0.666665290 0.333319516 0.587565394<br>
Se 0.666665290 0.333319516 0.687565394<br>Se 0.666665290 0.333319516 0.787565394<br>Se 0.666665290 0.333319516 0.887565394<br>Se 0.666665290 0.333319516 0.987565394<br>K_POINTS automatic<br>
8 8 1 0 0 0<br><br><br>Testing the speed, version 4.1 is a bit slower than 4.0.4 (about 9% more time in this benchmark: 39.5 vs 36 minutes, using 32 cpus). I guess this is irrelevant in face of Moore's law.<br><br><br>
<br>Thank you,<br>Best regards<br clear="all"><br>-- <br>Eduardo Menendez<br>Departamento de Fisica<br>Facultad de Ciencias<br>Universidad de Chile<br>Phone: (56)(2)9787439<br>URL: <a href="http://fisica.ciencias.uchile.cl/%7Eemenendez" target="_blank">http://fisica.ciencias.uchile.cl/~emenendez</a><br>