Top Page > Browsing

about computional memory consuming Date: 2005/04/22 12:57 Name: Brighton Yan Hi, with OPENMX I am performing a computation for a silicon nanowire with 96 atoms using a 1.0G pentium machine. My calculation always exit because of the lack of memory. However, the same system has been successfully calculated by other porgram. could you tell me an efficient way to perform my calculations. thanks. some of the main paramters in my *.dat file are as follows: ------------------------------------------------------------ Species.Number 2 <Definition.of.Atomic.Species Si Si6.0R-s2p2d1 Si_LDA H H4.0-s1p1 H_TM Definition.of.Atomic.Species> Atoms.Number 96 Atoms.SpeciesAndCoordinates.Unit Ang # Ang|AU <Atoms.SpeciesAndCoordinates # Unit=Ang. 1 Si -6.71929900 -6.71879300 -2.03625000 2.0 2.0 2 Si -6.71929900 -2.87920300 -2.03625000 2.0 2.0 * ……………………………… 89 H -4.79953585 -7.97960280 1.11302744 0.5 0.5 90 H -0.96471111 -7.97638915 1.10479311 0.5 0.5 91 H 2.87515445 -7.97671935 1.10921005 0.5 0.5 92 H 7.99082918 -6.72858741 2.93232688 0.5 0.5 93 H 6.70629800 -7.99382117 1.14427603 0.5 0.5 94 H 7.95571834 -2.87800754 2.99481178 0.5 0.5 95 H 7.95581231 0.96269621 2.99331933 0.5 0.5 96 H 7.95642975 4.78180792 2.99349086 0.5 0.5 Atoms.SpeciesAndCoordinates> Atoms.UnitVectors.Unit Ang # Ang|AU <Atoms.UnitVectors # unit=Ang. 30.0 0.0 0.0 0.00 30.0 0.0 0.0 0.0 5.43 Atoms.UnitVectors> scf.EigenvalueSolver band # Recursion|Cluster|Band scf.Kgrid 1 1 2 # means 4x4x4 scf.XcType LDA # LDA|LSDA-CA|LSDA-PW orderN.HoppingRanges 5.0 # default=5.0 (Ang) orderN.NumHoppings 2 # default=2

Page: [1]

Re: about computional memory consuming ( No.1 ) Date: 2005/04/13 10:55 Name: T.Ozaki Hi, In the below cases A and B, my trials were successful on a single Pentium 4 machine with 1 GByte memory. (icc ver. 5.0.1, SuSe Linux 7.1) (1) Case A Required cutoff energy 100 Ryd Used cutoff energy 122.8 128.8 121.5 Ryd SiR6.0-s2p2d1 H4.0-s1p1 Maximum memory size: 1.25 Gbyte (measured by "top" command) (2) Case B Required cutoff energy 90 Ryd Used cutoff 94.0 94.0 96.0 Ryd Si6.0R-s2p2d1 H4.0-s2 Maximum memory size: 1.02 Gbyte (measured by "top" command) For hydrogen, the use of double valence orbitals (H4.0-s2) would be effective rather than H4.0-s1p1 to span the occupied space (In addtion, it reduces the total number of basis functions). For the cutoff energy (scf.energycutoff), 90 Ryd maybe enough while the used cutoff energies are a little large (the used cutoff energies are found in the standard output). Regards, T.Ozaki

Page: [1]