[gmx-users] difference in energy when using -nb cpu

[Rafael I. Silverman y de la Vega]

Hi all,
I am running a nvt position restrained equilibration step on a protein,
with the native GPU accelerated run, I get
Potential Energy  = -5.4152162e+05
Maximum force     =  2.4561725e+02 on atom 2350
Norm of force     =  5.0402722e+00

but same .tpr file I get with -nb cpu

Potential Energy  = -5.4509656e+05
Maximum force     =  1.5848671e+02 on atom 2350
Norm of force     =  3.1396613e+00

Does this indicate I have bad combination of parameters?
Thanks