forum@abinit.org
Subject: The ABINIT Users Mailing List ( CLOSED )
List archive
- From: "Sam M. Vinko" <Sam.Vinko@physics.ox.ac.uk>
- To: forum@abinit.org
- Subject: [abinit-forum] Parallelization over kpts
- Date: Sat, 3 Oct 2009 12:02:56 +0100
Hello, I have a question regarding the way abinit splits up cpu resources accros the kpt/band/fft levels of parallelization (on abinit 5.8.4). My calculation is a 32 atom Al supercell, looking at 64 bands, and I'd like to sample the BZ with a 8x8x8 MP grid which, as I've excluded all symmetries, contains 256kpt. I have 128 cpus for this calculation and I'd like to distribute resources in the most efficient way so I've used the input option PARAL_KGB -128 which should tell me all the possible ways of splitting the cpus between bands/ffts/kpts; however, I seem to only get one option: For dataset= 1 a possible choice for less than 128 processors is: nproc npkpt npband npfft bandpp 256 256 1 1 2 Indeed, it turns out that any way of distributing 128 cpus for this job crashes. Does this mean that you always need at least a number of cpus larger than the k-grid? Or is there some subtlety I am missing? Sample input file with only the relevant option (takes about a minute to run) is appended below. Thanks in advance for your answers, Sam #***************************************************************************** #Definition of parallelizations paral_kgb -128 #npband 32 #npfft 1 #npkpt 4 #Definition of occupation numbers occopt 3 #FD distribution tsmear 0.003 #Definition of the unit cell acell 3*15.3271402862 # 7.6635701431 per cell, from minimizing forces rprim 1.0 0.0 0.0 0.0 1.0 0.0 0.0 0.0 1.0 #Other optimizations optcell 0 wfoptalg 4 fftalg 401 fft_opt_lob 2 nloalg 4 iscf 17 #Definition of the atom types ntypat 1 znucl 13 #Definition of the atoms natom 32 typat 32*1 xred 0.00 0.00 0.00 0.50 0.50 0.50 0.25 0.25 0.00 0.25 0.00 0.25 0.00 0.25 0.25 0.50 0.00 0.00 0.00 0.50 0.00 0.00 0.00 0.50 0.50 0.50 0.00 0.50 0.00 0.50 0.00 0.50 0.50 0.75 0.25 0.00 0.25 0.75 0.00 0.00 0.75 0.25 0.00 0.25 0.75 0.75 0.00 0.25 0.25 0.00 0.75 0.75 0.75 0.00 0.75 0.00 0.75 0.00 0.75 0.75 0.25 0.25 0.50 0.25 0.50 0.25 0.50 0.25 0.25 0.75 0.75 0.50 0.75 0.50 0.75 0.50 0.75 0.75 0.25 0.50 0.75 0.25 0.75 0.50 0.50 0.25 0.75 0.50 0.75 0.25 0.75 0.25 0.50 0.75 0.50 0.25 #Definition of the planewave basis set ecut 10.0 # Maximal kinetic energy cut-off, in Hartree pawecutdg 20.0 nband 64 #Exchange-correlation functional #ixc 11 # GGA #Definition of the k-point grid kptopt 1 # Option for the automatic generation of k points ngkpt 8 8 8 #istwfk 256*1 #nshiftk 1 #shiftk 0.0 0.0 0.0 #Definition of the SCF procedure nstep 10 # Maximal number of SCF cycles #tolvrs 1.0e-15 # Stops SCF cycle when potential residue lower than # specified number. Is better than toldfe. toldfe 1.0d-2 # Will stop when, twice in a row, the difference # between two consecutive evaluations of total energy # differ by less than toldfe (in Hartree) # Symmetries of the lattice: nsym 1 chkprim 0 #***************************************************************************** ____________________________________ Sam M. Vinko Atomic & Laser Physics Clarendon Laboratory - University of Oxford Parks Road, Oxford OX1 3PU, UK Tel: +44 (0) 1865 272 301 E-mail: sam.vinko@physics.ox.ac.uk |
- [abinit-forum] Parallelization over kpts, Sam M. Vinko, 10/03/2009
Archive powered by MHonArc 2.6.16.