[gmx-users] gromacs 3.0.5 compilation on Compaq XP1000 TruUnix64
Dr. Y. U. Sasidhar
sasidhar at chem.iitb.ac.in
Mon Nov 26 07:12:47 CET 2001
David van der Spoel wrote:
> Your problem really is a compiler problem, but here is a workaround:
> go to src/tools and type
> make CFLAGS=-O g_rms
> then go to the top level directory and type
> make
1) It now compiles. However demo script in gmxdemo still got stuck at
editconf, genbox level.
I found that giving -nice 0 option gets it going up to mdrun (em ). So I
gave -nice 0 to all mdrun commands in the script. Now the script goes
through. Note that during all this there were only two users on the system
and other user's job seemed to be idle. I am giving accounting below: ( This
is through telnet to our compute server; using bash shell )
=============
Computing: M-Number M-Flop's % Flop's
LJ 30.858335 956.608385 2.0
Coulomb 67.454933 1821.283191 3.9
Coulomb(W) 4.961964 401.919084 0.9
LJ + Coulomb 43.507922 1653.301036 3.5
LJ + Coul(W) 13.927847 1281.361924 2.7
LJ + Coul(WW) 148.600391 36407.095795 77.5
Innerloop-Iatom 103.284824 1032.848240 2.2
NS-Pairs 54.453665 1143.526965 2.4
Reset In Box 0.805107 7.245963 0.0
Shift-X 16.073214 96.439284 0.2
CG-CoM 0.283065 8.208885 0.0
Angles 1.045209 170.369067 0.4
Propers 0.370074 84.746946 0.2
Impropers 0.350070 72.814560 0.2
Virial 8.171634 147.089412 0.3
Update 8.036607 249.134817 0.5
Stop-CM 8.035000 80.350000 0.2
Calc-Ekin 8.038214 217.031778 0.5
Lincs 0.720432 43.225920 0.1
Lincs-Mat 12.547524 50.190096 0.1
Shake-V 8.036607 120.549105 0.3
Shake-Vir 8.036607 144.658926 0.3
Settle 2.441464 788.592872 1.7
Total 46978.59225 100.0
NODE (s) Real (s) (%)
Time: 151.244 302.000 50.1
2:31
(Mnbf/s) (MFlops) (ps/NODE hour) (NODE hour/ns)
Performance: 10.155 310.615 238.026 4.201
========================================
2) The same demo script if run under dtterm the following output results:
I am always using bash.
==========================================
The genbox program reads the peptide structure file and an input file
containing the sizes of the desired water box. The output of genbox is
a gromos structure file of a peptide solvated in a box of water. The
genbox program also changes the topology file ( .top extension ) to
include water. First we will use the program editconf to define the
right boxsize for our system.
-----------------------------------------------------------------
-----------------------------------------------------------------
The output of the genbox program should appear
": Event not found.
bash-2.02$
====================================
3) Even when we were using gromacs 2.0 on this machine we observed that our
progrmas were stopping for no apparent reason without any error messages. By
trial and error we found that -nice 0 alleviates the problem.
On another similar machine ( same hardware/OS ) no such problem was
observed !!
Any comments on this behaviour ?
Sasidhar
More information about the gromacs.org_gmx-users
mailing list