|
各位大佬们我想问一下计算出现虚拟内存不足的情况forrtl: severe (41): insufficient virtual memory
我是在5*5体系的二维材料掺杂单个其他原子计算其光学性质的情况下遇到的这个问题的
这是我遇到问题时候提交计算的脚本:调用2个节点 每个节点64核 出现虚拟内存不足的情况#SBATCH -N 2
#SBATCH --ntasks-per-node=64
#SBATCH --time=100:0:0
经过超算平台管理人员测试给我修改的试用脚本:调用了4个节点 每个节点8核 可以计算出来光学性质
#SBATCH -N 4
#SBATCH --ntasks-per-node=8
#SBATCH --exclusive
#SBATCH --time=50:0:0
到这里我产生了一些疑问,在计算过程中出现的虚拟内存不足的情况,他是跟计算资源的单个节点的核数有关呢,还是说需要更多的节点才可以更好的避免这个问题?
以下是我计算光学性质所用到的INCAR设置:
SYSTEM = OPTIC
Global Parameters
ISTART = 0 (Read existing wavefunction; if there)
#ISPIN = 1 (Spin polarised DFT)
ICHARG = 11 (Non-self-consistent: GGA/LDA band structures)
LREAL = .F. (Projection operators: automatic)
ENCUT = 500 (Cut-off energy for plane wave basis set, in eV)
PREC = Normal (Precision level)
LWAVE = .F. (Write WAVECAR or not)
LCHARG = .F. (Write CHGCAR or not)
ADDGRID = .TRUE. (Increase grid; helps GGA convergence)
NCORE = 8
LOPTICS = T
NEDOS = 3000
EMIN = -20
EMAX = 20
Electronic Relaxation
ISMEAR = 0 (Gaussian smearing; metals:1)
SIGMA = 0.05 (Smearing value in eV; metals:0.2)
NELM = 200 (Max electronic SCF steps)
EDIFF = 1E-6 (SCF energy convergence; in eV)
# GGA = PS (PBEsol exchange-correlation)
Ionic Relaxation
NSW = 0 (Max ionic steps)
IBRION = 1 (Algorithm: 0-MD; 1-Quasi-New; 2-CG)
POTIM = 0.5
ISIF = 3 (Stress/relaxation: 2-Ions, 3-Shape/Ions/V, 4-Shape/Ions)
EDIFFG = -0.01 (Ionic convergence; eV/AA)
ISYM = 0 (Symmetry: 0=none; 2=GGA; 3=hybrids)
|
-
|