[gmx-users] Strange T-coupling error

[Senthil Kandasamy]

I have been using the following parameters for T-coupling in a
lipid-bilayer/protein/water system

Tcoupl	=Berendsen
tc-grps = POC Protein BULK ; Bulk is water +ions
tau_t	= 0.1	0.1	0.1
ref_T	= 310  310     310

For all my simulations, this seems to give reasonable temperature
control ( fluctuations of about 1 degree K).

Recently, I started running these simulations on a different, new, (AMD)
cluster. All the simulations on the newer machine seem to have
temperature control problems with the same parameters. The temperature
jumps to ~650 K after a few ps and stays there. There is absolutely no 
difference in the mdp files between my earlier simulations and my new
simulations (Except that older simulations were on 4 processors and
newer ones are on single processors). I have been playing around with
tau_t  and still the temperature can not be controlled closely. At tau_t
= 0.005 ps, the temperature control is better but T is still high (320
K). This is absolutely strange... Nose-hoover does the same thing too,
though better. On my earlier simulations, the Berendsen coupling did a
really tight control. 

I do not remember having any installation problems on the newer cluster.
So I can not really track the source of the error to the gromacs
installation. I have checked my mdp file and mdout.mdp multiple times to
ensure that it is the same. Has anyone seen anything bizarre like this
before? Any suggestions on troubleshooting this mess?

Thanks.

my mdp file:
title               =  POPC +MSI78
cpp                 =  /lib/cpp
;define		    =  -DPOSRES_MSI78
constraints         =  all-bonds
constraint_algorithm = lincs
integrator          =  md
dt                  =  0.002	; ps !
nsteps              =  50000000    ; total  100 ns.
nstxout             =  2500
nstvout             =  2500
nstfout             =  2500
nstlog              =  500
nstenergy           =  500
nstxtcout           =  500
nstlist             =  10
ns_type             =  grid
coulombtype	    =  PME
rlist               =  0.9
rcoulomb            =  0.9
rvdw                =  1.2
pbc		    =  xyz
comm_mode	    =  linear
nstcomm		    =  1
optimize_fft	    =  yes 
comm_grps	    =  POC   SOL 
; Berendsen temperature coupling is on in three groups
Tcoupl              =  berendsen
tc-grps		    =  POC	BULK	Protein	;	Cl	Na
tau_t               =  0.1	0.1 	0.1	;	0.01	0.01	
ref_t               =  310	310	310	;	310	310
; Energy monitoring
energygrps          = POC    	SOL 	 Protein	Cl	Na
; Isotropic pressure coupling is now on
Pcoupl              = berendsen
Pcoupltype          = anisotropic
tau_p               =  5	5	5	0	0	0
compressibility     =  4.5e-5	4.5e-5 	4.5e-5	0	0	0	
ref_p               =  1.0	1.0	1.0	0	0	0
; Generate velocites is on at 310 K.
gen_vel             =  yes
gen_temp            =  310.0
gen_seed            =  31415