[gmx-users] LINCS warnings and number of cpus

[Marani Alessandro]

Dear users,
I would like to ask your help about understanding a problem i'm not able 
to recognize by myself.
Basically, a user of our sistem (IBM SP6, power6 architecture) is trying 
to run a simulation of a very simple sistem, a polymer chain in a lot of 
water molecules. While the simulation works perfectly in serial on her 
local pc, when she tries to run it in SP6 using 2 cpus in parallel, the 
simulation doesn't even start due of these errors:

/starting mdrun 'PVA head29tail in water'
2500000 steps,   5000.0 ps.

Step 0, time 0 (ps)  LINCS WARNING
relative constraint deviation after LINCS:
rms 83427404711319.468750, max 431899260485632.000000 (between atoms 59 
and 60)
bonds that rotated more than 30 degrees:
  atom 1 atom 2  angle  previous, current, constraint length
      37     38   89.6    0.1530 124222742528.0000      0.1530
      38     41   89.1    0.1530 104382357504.0000      0.1530

Step 0, time 0 (ps)  LINCS WARNING
relative constraint deviation after LINCS:
rms 56829711484907.867188, max 212898492186624.000000 (between atoms 3 
and 4)
bonds that rotated more than 30 degrees:
  atom 1 atom 2  angle  previous, current, constraint length
      61     62   89.0    0.1530 1380386603008.0000      0.1530
      58     61   87.6    0.1530 669758914560.0000      0.1530
      58     59   89.2    0.1430 1994809540608.0000      0.1430
      59     60   90.0    0.1000 43189926887424.0000      0.1000
      57     58   88.3    0.1530 1126801342464.0000      0.1530
      54     57   85.9    0.1530 198269452288.0000      0.1530
      54     55   89.6    0.1430 1261346488320.0000      0.1430
      38     39   89.6    0.1430 156364750848.0000      0.1430
      39     40   90.0    0.1000 3183089549312.0000      0.1000
//     [...]
      18     21   89.9    0.1530 215265542144.0000      0.1530
      18     19   90.0    0.1430 872386920448.0000      0.1430
      19     20   90.0    0.1000 6085561286656.0000      0.1000
      95     96   90.0    0.1000 5006930477056.0000      0.1000
      97     98   89.7    0.1530 215830478848.0000      0.1530
      98    101   90.5    0.1530 232671739904.0000      0.1530
      98     99   90.0    0.1430 746076962816.0000      0.1430
      99    100   90.0    0.1000 6068662435840.0000      0.1000
step 0: Water molecule starting at atom 6014 can not be settled.
Check for bad contacts and/or reduce the timestep if appropriate.

step 0: Water molecule starting at atom 7355 can not be settled.
Check for bad contacts and/or reduce the timestep if appropriate.
Wrote pdb files with previous and current coordinates
Wrote pdb files with previous and current coordinates
step 0
Warning: 1-4 interaction between 37 and 40 at distance 3039839053625.364 
which is larger than the 1-4 table size 2.200 nm
These are ignored for the rest of the simulation
This usually means your system is exploding,
if not, you should increase table-extension in your mdp file
or with user tables increase the table size
Warning: 1-4 interaction between 61 and 64 at distance 
15924520737123.646 which is larger than the 1-4 table size 2.200 nm
These are ignored for the rest of the simulation
This usually means your system is exploding,
if not, you should increase table-extension in your mdp file
or with user tables increase the table size
ERROR: 0031-250  task 0: Segmentation fault
ERROR: 0031-250  task 1: Segmentation fault

/The same errors occur when trying the simulation up to 4 cpus, but (and 
that's the strange thing), everything works fine with 6+ cpus (actually, 
there are some numbers giving an incompatibility error, like /"There is 
no domain decomposition for 7 nodes that is compatible with the given 
box and a minimum cell size of 0.95625 nm"/, but for example 6 or 8 cpus 
give a successful run).

Can anyone understand what is the reason of this strange behaviour?
Thanks,
Marani Alessandro (HPC User support, CINECA - Italy)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20120116/f32fecac/attachment.html>