[gmx-users] Re:Re: g_nmeig_d error:cannot allocate region

[sarbani chattopadhyay]

 Hi,
I had checked and found that gromacs has been compiled in 64 bit mode. However still I am getting this error with "g_nmeig_d " command which says
g_nmeig_d(1892) malloc: *** mmap(size=18446744072353271808) failed (error code=12)
*** error: can't allocate region


 I am running this on a 8 node machine,each with 2GB of RAM ( g_nmeig_d runs on a single node).

Is it required to increase the RAM of 1 node to get rid of this error?
Any suggestion will be very helpful.
Thanking You,
Sarbani



On Tue, 02 Mar 2010 20:01:51 +0530 wrote
>
>
>
Hi,
>

>
The allocation that causes the error should allocate about 3 GB.
>
My guess is that you compiled Gromacs in 32bit mode, whereas 64bit mode
>
is required for allocation of more than 2 GB.
>

>
Berk
>

>
Date: Tue, 2 Mar 2010 13:35:58 +0000
>
To: gmx-users at gromacs.org
>
From: sarbani_c84 at rediffmail.com
>
Subject: [gmx-users] g_nmeig_d error: can't allocate region
>

>
Hi,
>

>
I am trying to do normal mode analysis on a protein having 6398 atoms in vaccum.

I tried to energy minimize the structure using steepest descent, followed by "l-bfgs"
minimization. the .mdp file I used is

define = -DFLEXIBLE

constraints = none
integrator = l-bfgs
tinit = 0
nsteps = 15000
nbfgscorr = 50
emtol = .001
emstep = 0.1
gen_vel = yes
gen-temp = 300
nstcomm = 1
; NEIGHBORSEARCHING PARAMETERS
; nblist update frequency
nstlist = 0
; ns algorithm (simple or grid)
ns-type = simple
; Periodic boundary conditions: xyz (default), no (vacuum)
or full (infinite systems only)
pbc = no
rlist = 0
domain-decomposition = no
; OPTIONS FOR ELECTROSTATICS AND VDW
; Method for doing electrostatics
coulombtype = Cut-Off
rcoulomb-switch = 0
rcoulomb = 0
; Dielectric constant (DC) for cut-off or DC of reaction field
epsilon-r = 1
; Method for doing Van der Waals
vdw-type = Cut-off
; cut-off lengths
rvdw-switch = 0
rvdw = 0

>
after running the 15000 steps the Fmax was:>
Low-Memory BFGS Minimizer converged to Fmax < 0.001 in 10197 steps

Potential Energy = -8.26391832320506e+04

Maximum force = 9.37558560558845e-04 on atom 4562

Norm of force = 2.24887722104890e-04

Again the "l-bfgs" minimization was run using the same .mdp file( with emtol =
0.000001)>
the output was'
>

>
Low-Memory BFGS Minimizer converged to Fmax < 1e-06 in 4143 steps
>

>
Potential Energy = -8.26391832324998e+04
>

>
Maximum force = 9.67927896882578e-07 on atom 3271
>

>
Norm of force = 1.70637151528245e-07
>

>

>

>

>

>

>

>
After this I prepared the "nm.mdp " file for NMA, where I used exactly the same parameters
>

>
as the ones used in lbfgs energy minimization( with integrator = nm)
>

>

>

>
the commands that were used were:
>

>
grompp_d -f new_nm.mdp -t new_lbfgs_2.trr -c new_lbfgs_2.gro -o new_nm.tpr -zero -p
>

>
../topol.top
>

>

>

>
nohup mdrun_d -v -s new_nm.tpr -deffnm new_nm -mtx new_nm.mtx &
>

>

>

>

>

>
"nohup.out" had the following message:
>

>
Non-cutoff electrostatics used, forcing full Hessian format.Allocating Hessian
>

>
memory...starting normal mode calculation 'Protein'6398 steps.Maximum force: 9.67928e-07
>

>

>

>

>

>
The run ended successfully:
>

>

>

>
Then i used the command
>

>
g_nmeig_d -f new_nm.mtx -s new_nm.tpr -ol eigenvalue.xvg -v eigenvector.trr
>

>

>

>
I get the following error:
>

>
Reading file new_nm.tpr, VERSION 4.0.7 (double precision)
>

>
Reading file new_nm.tpr, VERSION 4.0.7 (double precision)
>

>
Reading double precision matrix generated by Gromacs VERSION 4.0.7
>

>
Full matrix storage format, nrow=19194, ncols=19194
>

>

>

>
Diagonalizing to find vectors 1 through 50...
>

>
g_nmeig_d(1892) malloc: *** mmap(size=18446744072353271808) failed (error code=12)
>

>
*** error: can't allocate region
>

>
*** set a breakpoint in malloc_error_break to debug
>

>

>

>
I am not being able to understand the problem. the computer has a 16gb memory
>

>

>

>

>

>

>

>
If I use different parameters in the nm.mdp file as
>

>

>

>
rlist = 1.5
>

>
domain-decomposition = no
>

>
; OPTIONS FOR ELECTROSTATICS AND VDW
>

>
; Method for doing electrostatics
>

>
coulombtype = switch
>

>
rcoulomb-switch = 1
>

>
rcoulomb = 1.2
>

>
; Dielectric constant (DC) for cut-off or DC of reaction field
>

>
epsilon-r = 1
>

>
; Method for doing Van der Waals
>

>
vdw-type = switch
>

>
; cut-off lengths
>

>
rvdw-switch = 1
>

>
rvdw = 1.2
>

>

>

>

>

>
then i get the message :Maximum force: 3.14171e+03
>

>
Maximum force probably not small enough to ensure that you are in an
>

>
energy well. Be aware that negative eigenvalues may occur when the
>

>
resulting matrix is diagonalized.
>

>

>

>

>

>
I am sorry to post such a lengthy query, but I have no clue about the root of the problem.
>

>
Any suggestion will be of great help.
>

>
Thanks in advance,
>

>
Sarbani. 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20100408/a584f68b/attachment.html>