How do I calculate accurate DOS?

[sch28181]

I want to calculate DOS of Pd(111) system.
At first I optimized 3layer 3x3 Pd system.
With optimized system, I tried to do DOS calculation.
This is my INCAR file.

NWRITE = 2; LPETIM=F

ISTART = 0 job : 0-new 1-cont 2-samecut

ICHARG = 2 charge: 1-file 2-atom 10-const

INIWAV = 1 electr: 0-lowe 1-rand

NGX = 46 ; NGY = 46 ; NGZ = 94
Electrronic Relaxation 1

NELM = 100; NELMIN = 4

PREC = Normal
EDIFF = 1E-05 stopping criterion for ELM
ENCUT = 270

Ionic Relaxation

LCORR = T Harris-correction to forces
LORBIT = 10
GGA = 91
DOS related values

ISMEAR = -5; SIGMA = 0.2
LWAVE = .FALSE.

Kpoint is

K-Points
0
Gamma
9 9 1
0 0 0

But the result was different from reference.
In my calculation the result was not a smooth curve.
But there are many sharp peaks.
What is the problem?

[zoowe]

you should look into the manual. There is a section called Accurate Band, DOS.

[sch28181]

what is the difference between SC calculation and non-SC calculation?

[physicalattraction]

In a non-self-consistent calculation, you need to input the position of the charges (i.e. charge density, supplied in the CHGCAR file) and these will be fixed during the run. It might be that the charge density you supply is not consistent with the ground state of the system. In a self-consistent run, VASP would update the charge density, but in a non-self-consistent run, the charge density is kept fixed.

[boris]

And so to get an accurate DOS calculation, you first perform a self-consistent calculation in order to let VASP determine the ground state charge density, then you perform a non-self-consistent calculation (ICHARG = 11) to determine the accurate DOS.