why did the EDIFFG not work?

[ichthyornis]

Hi,
I relaxed a system with about 60 atoms, using default EDIFF and EDIFFG values and NSW=120. But, it ended after 3 ionic steps and the energy difference of two ionic step is -.557072E-01. the log file like this:

N E dE d eps ncg rms rms(c)
... ...
1 F= -.45300547E+03 E0= -.45300547E+03 d E =-.453005E+03
curvature: 0.00 expect dE= 0.000E+00 dE for cont linesearch 0.000E+00
trial: gam= 0.00000 g(F)= 0.338E+00 g(S)= 0.000E+00 ort = 0.000E+00 (trialstep = 0.100E+01)
search vector abs. value= 0.338E+00
bond charge predicted
... ...
2 F= -.45279581E+03 E0= -.45279581E+03 d E =0.209660E+00
trial-energy change: 0.209660 1 .order 0.226872 -0.337705 0.791449
step: 0.3189(harm= 0.2991) dis= 0.00745 next Energy= -453.059882 (dE=-0.544E-01)
bond charge predicted
... ...
3 F= -.45306118E+03 E0= -.45306118E+03 d E =-.557072E-01
curvature: -0.15 expect dE=-0.150E-03 dE for cont linesearch -0.434E-04
trial: gam= 0.00000 g(F)= 0.101E-02 g(S)= 0.000E+00 ort =-0.990E-02 (trialstep = 0.864E+00)
search vector abs. value= 0.101E-02
reached required accuracy - stopping structural energy minimisation
Fri Jun 2 07:40:34 CST 2006

Could anyone tell me what hapen to this and how to avoid it? thanks in advance.

sincerely,
ichthyornis