energy in case of spiral calculations

[xianghjun]

To learn how to perform spiral calculations,
I considered a 1D Fe chain along the z direction.
POSCAR
========================
Fe
1.000000
8 0 0
0 8 0
0 0 2
1
Direct
0.5 0.5 0.5
========================

I found E(q=(0,0,0.5)) differes a lot from the energy of the AFM state:
E(q=(0,0,0.5))-E(AFM)=44 meV/Fe
And much more surprisingly, E(q=(0,0,1.0)) is not equal to the energy of the FM state:
E(q=(0,0,1.0))-E(FM)=312 meV/Fe.
Given QSPIRAL is the propagation vector of the spiral in direct coordinates of the reciprocal lattice, I would expect E(q=(0,0,1.0))=E(FM).

So what is the reason for such discrepancy? Do I miss something?
Thank you.

Btw, from my test calculations, I found E(q=(0,0,0))=E(FM).

Here is my INCAR file:
===============================
IALGO=38
EDIFF=1E-05
EDIFFG = -0.05

ISMEAR=-5
ISPIN=2
SIGMA=0.1

NSW=0
IBRION=2

LPLANE = .TRUE.
NPAR = 1
LSCALU = .FALSE.
NSIM = 4
LREAL=AUTO
ISIF=2
PREC=normal
MAGMOM= 2 0 0

LORBIT=11

LSPIRAL=.TRUE.
LASPH=.TRUE.
QSPIRAL = 0.0 0.0 1.00
LZEROZ=.TRUE.
LNONCOLLINEAR =.TRUE.
ICHARG=1
ISYM=0
=============================

[xianghjun]

When I increase the cutoff energy to 500 eV,
E(q=(0,0,1.0)) becomes close to E(q=(0,0,0.0)).