[gmx-users] compare 3.3.3 vs 3.99.4 (bit long)

spitaleri.andrea at hsr.it spitaleri.andrea at hsr.it
Thu Aug 28 14:18:28 CEST 2008


Dear all,
I am using the Louhi cluster on the Finnish IT center
(http://www.csc.fi/english) and I had the chance to test the latest beta
version of gromacs (3.99.4). First of all I have to say that it is very
fast and very scalable. Now, reliability. I have run a 2ns simulation of
a protein in water (ca. 200,000 atoms) using 3.3.3 and 3.99.4 and 2 fs
integration time. 
Our home cluster (ibm intel 2.66 Ghz) bearing 3.3.3 needs about 4d for
2ns on 24 nodes, whereas louhi with 3.99.4 on 64 nodes needs about 15h!
good stuff. 
I have checked the energies for both simulation and below you see the
results:

######
3.99.4
######


Statistics over 1000001 steps [ 0.0000 thru 2000.0001 ps ], 27 data sets

Energy                      Average       RMSD     Fluct.      Drift 
Tot-Drift
-------------------------------------------------------------------------------
Angle                       46984.2    307.076     298.32  -0.126109  
-252.218
Proper Dih.                 2743.89    67.4457    66.9484   -0.01416   
 -28.32
Ryckaert-Bell.              20897.7    387.982    219.399  -0.554239  
-1108.48
LJ-14                       33023.1    136.292     135.93  0.0171902   
34.3805
Coulomb-14                   148467     195.67    194.323 -0.0396933  
-79.3867
LJ (SR)                      538656    1748.75    1747.19  -0.128059  
-256.119
Coulomb (SR)             -4.0367e+06    2448.62    2428.54  -0.542011  
-1084.02
Coul. recip.                -865148    242.753    242.316 -0.0252178  
-50.4356
Potential                -4.11107e+06    1663.33    1449.76    -1.4123 
 -2824.61
Kinetic En.                  772334    983.851    983.705  0.0293455   
 58.691
Total Energy             -3.33874e+06    1346.81    1084.61   -1.38295 
 -2765.91
Temperature                 299.984    0.38214   0.382083 1.13975e-05 
0.0227951
Pressure (bar)              1.15551    37.5452    37.5448 0.000308348  
0.616696
Volume                      3153.75    1.38233     1.3728 -0.000280674 
-0.561349
Density (SI)                994.349   0.435839   0.432836 8.84706e-05  
0.176941
Coul-SR:Protein-Protein      -67091    453.617    453.564  0.0119892   
23.9785
LJ-SR:Protein-Protein      -43707.5    234.515    205.403  -0.196011  
-392.022
Coul-14:Protein-Protein      148185    195.603    194.174 -0.0408786  
-81.7573
LJ-14:Protein-Protein       32996.1     136.26    135.868  0.0178968   
35.7937
Coul-SR:Protein-Other       -116084    1057.65    1025.74  -0.446636  
-893.273
LJ-SR:Protein-Other        -6113.29    324.853    321.225  0.0838503   
167.701
Coul-14:Protein-Other       18.5116    3.66015    3.16015 -0.0031985  
-6.39701
LJ-14:Protein-Other         16.2657     3.3305    3.31851 0.000489169  
0.978339
Coul-SR:Other-Other      -3.85352e+06     2476.7    2475.92  -0.107357 
 -214.713
LJ-SR:Other-Other            588476    1731.91    1731.88 -0.0158981  
-31.7962
Coul-14:Other-Other         263.365    6.86274    6.37901 0.00438365   
 8.7673
LJ-14:Other-Other            10.752    3.07146    2.99287 -0.00119576  
-2.39153
Heat Capacity Cv:      12.4718 J/mol K (factor = 1.62274e-06)
Isothermal Compressibility: 1.46289e-05 /bar
Adiabatic bulk modulus:        68357.7  bar

#####
3.3.3
#####

Statistics over 1000001 steps [ 0.0000 thru 2000.0001 ps ], 27 data sets

Energy                      Average       RMSD     Fluct.      Drift 
Tot-Drift
-------------------------------------------------------------------------------
Angle                       46944.4    319.789    295.167   -0.21313   
-426.26
Proper Dih.                 2755.63    69.3266    66.9131 -0.0314078  
-62.8156
Ryckaert-Bell.              20953.3    374.694    221.951  -0.522876  
-1045.75
LJ-14                       33118.6    140.075    137.154  0.0492889   
98.5779
Coulomb-14                   148701    188.693    187.295 -0.0397171  
-79.4343
LJ (SR)                      538555    1749.63    1748.67  -0.100002  
-200.004
Coulomb (SR)             -4.03654e+06    2470.23    2458.61  -0.414596 
 -829.193
Coul. recip.                -865451    236.417    235.938 -0.0260338  
-52.0677
Potential                -4.11096e+06    1656.99     1477.7   -1.29848 
 -2596.96
Kinetic En.                  772333    984.094    983.984 -0.0254802  
-50.9604
Total Energy             -3.33863e+06       1338    1098.16   -1.32396 
 -2647.92
Temperature                 299.984   0.382234   0.382192 -9.89665e-06
-0.0197933
Pressure (bar)              1.01847    37.5027    37.4954  0.0012831   
 2.5662
Volume                      3154.08    1.45018    1.44809 -0.000134813 
-0.269625
Density (SI)                994.246   0.457148   0.456488 4.25118e-05 
0.0850236
Coul-SR:Protein-Protein    -67562.2     390.93    384.643    0.12095   
  241.9
LJ-SR:Protein-Protein      -43829.4    268.207     216.54   -0.27411   
-548.22
Coul-14:Protein-Protein      148701    188.693    187.295 -0.0397171  
-79.4343
LJ-14:Protein-Protein       33118.6    140.075    137.154  0.0492889   
98.5779
Coul-SR:Protein-Other       -115824    993.994    877.644  -0.808267  
-1616.54
LJ-SR:Protein-Other         -6070.4    367.068    343.991   0.221871   
443.742
Coul-14:Protein-Other             0          0          0          0   
      0
LJ-14:Protein-Other               0          0          0          0   
      0
Coul-SR:Other-Other      -3.85315e+06    2483.91    2478.92   0.272709 
  545.419
LJ-SR:Other-Other            588455    1731.91    1731.69 -0.0477627  
-95.5255
Coul-14:Other-Other               0          0          0          0   
      0
LJ-14:Other-Other                 0          0          0          0   
      0
Heat Capacity Cv:      12.4718 J/mol K (factor = 1.62354e-06)
Isothermal Compressibility: 1.60985e-05 /bar
Adiabatic bulk modulus:        62117.6  bar

A first look shows that the two simulation gives similar energy results
(except for the 14 interaction protein-other and other-other).

I am thinking to use the new gmx to produce my simulation since it seems
stable enough. Any suggestion in this stage are very welcome.

thanks in advance and sorry for the long post.

Regards,

Andrea

Andrea Spitaleri PhD
Dulbecco Telethon Institute 
c/o DIBIT Scientific Institute
Biomolecular NMR, 1B4
Via Olgettina 58
20132 Milano (Italy)
Tel: 0039-0226434348/5622/3497/4922
Fax: 0039-0226434153

********************************************************************
Sostieni la ricerca del San Raffaele con il 5permille!
E' SEMPLICE E NON COSTA NULLA.
Basta indicare nell'apposito riquadro della dichiarazione dei
redditi ("Finanziamento agli enti della ricerca sanitaria") 
il codice fiscale della 
Fondazione Centro S. Raffaele del Monte Tabor:
03 06 42 80 153 e ricordarsi di firmare.
Se vuoi saperne di piu' scrivi a 5permille at hsr.it o vai sul sito
www.5xmille.org
-------------- next part --------------
A non-text attachment was scrubbed...
Name: spitaleri.andrea.vcf
Type: text/x-vcard
Size: 241 bytes
Desc: Card for <spitaleri.andrea at hsr.it>
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20080828/293efa40/attachment.vcf>


More information about the gromacs.org_gmx-users mailing list