My Community

Hello,

I have a question about the running time in vasp. I have two jobs, i.e. 33410, 33411. By comparing these two jobs, it is very strange to find that the time consuming in the first SCF step (which is DAV) reduced significantly (from 4834 to 789), while the total time consuming in the LOOP+ are very close (11528 and 11294). Why?

What is the difference between the sum of LOOPs and LOOP+? Is this difference indicated in the OUTCAR?

33410:
LOOP: VPU time 4834.99: CPU time 4861.65
LOOP: VPU time 575.67: CPU time 579.36
LOOP: VPU time 582.99: CPU time 586.59
LOOP: VPU time 574.28: CPU time 578.27
LOOP: VPU time 562.13: CPU time 566.35
LOOP: VPU time 510.53: CPU time 513.69
LOOP: VPU time 512.49: CPU time 515.47
LOOP: VPU time 499.57: CPU time 502.14

CHARGE: VPU time 10.12: CPU time 10.25
FORLOC: VPU time 0.00: CPU time 0.00
FORNL : VPU time 183.53: CPU time 184.89
STRESS: VPU time 459.53: CPU time 462.18
FORCOR: VPU time 0.03: CPU time 0.03
FORHAR: VPU time 0.01: CPU time 0.01
MIXING: VPU time 0.00: CPU time 0.00
LOOP+: VPU time11528.87: CPU time11655.46

33411:
LOOP: VPU time 789.50: CPU time 793.54
LOOP: VPU time 585.24: CPU time 589.41
LOOP: VPU time 585.94: CPU time 589.59
LOOP: VPU time 586.13: CPU time 589.24
LOOP: VPU time 544.49: CPU time 547.90
LOOP: VPU time 476.07: CPU time 478.26
LOOP: VPU time 476.78: CPU time 479.39
LOOP: VPU time 475.91: CPU time 478.44
LOOP: VPU time 457.05: CPU time 459.54

CHARGE: VPU time 7.94: CPU time 7.97
FORLOC: VPU time 0.00: CPU time 0.00
FORNL : VPU time 308.42: CPU time 309.80
STRESS: VPU time 436.49: CPU time 439.26
FORCOR: VPU time 0.06: CPU time 0.06
FORHAR: VPU time 0.01: CPU time 0.01
MIXING: VPU time 0.00: CPU time 0.00
LOOP+: VPU time11294.72: CPU time11504.52


Best,
Binglun

Can it be that the job 33411 is reusing the charge density of job 33410?

The charge density is stored in CHGCAR and can be reused in consequent calculations:
http://cms.mpi.univie.ac.at/vasp/vasp/I ... .html#3135