AMBER Archive (2004)

Subject: RE: AMBER: sander error

From: Ross Walker (ross_at_rosswalker.co.uk)
Date: Thu Aug 05 2004 - 23:50:09 CDT

Dear Rukman
 
> I have installed amber8 in RedHat linux using ifort compiler. The
> installation looks OK, because I can passed all of the make test at
> $AMBERHOME/test. However, when I tried to run minimization to my real
> system, I found the following error:
>
> forrtl: severe (174): SIGSEGV, segmentation fault occurred
> Image PC Routine Line Source
> sander 080AF363 Unknown Unknown Unknown
> sander 080B5B2D Unknown Unknown Unknown
> sander 08079811 Unknown Unknown Unknown
> sander 080785DE Unknown Unknown Unknown
> sander 08152EB4 Unknown Unknown Unknown
> sander 42015574 Unknown Unknown Unknown
> sander 0804A161 Unknown Unknown Unknown

This is very strange considering all the tests pass.
 
> I think it is not related to my input file since I could successfully
> run the same input file to amber8 installed in our SGI
> machine.

It is unlikely there is anything wrong with your installation if the tests
all run ok. Are you running single processor here?

> Probably
> I missed something when installed amber8 in our linux machine, some
> environment setting or others. If anybody ever solved this type of
> problem please share.

> &cntrl
> imin = 1, maxcyc = 5000,
> ntpr = 10, ntf = 2, dt=0.002, tol = 0.000001,
> ntb = 0,
> cut = 30, scnb = 2.0, scee = 1.2,
> igb = 2, saltcon = 0.1, nrespa=1,
> icnstph = 1, solvph = 3.0, ntcnstph = 5,
> /

What are the specs of your linux machine? Also do you get an mdout file or
does it crash before that? Is there any more to the error message you see?

How big is the system you are trying to simulate? If your system is very
large then such a large cut off as 30 may mean that the pair list is very
very big which is leading to memory allocation problems. The SGI is 64 bit
and so less likely to have problems here. Can you run a simpler job? I.e
without the constant pH flags?

If you still have problems please post your prmtop and inpcrd files and I'll
take a look. You could also try recompiling with debug flags on - set
AMBERBUILDFLAGS=-g in config.h and then although the program will run slower
it may give you the traceback when it seg faults and we can work out where
in the code it dies.

All the best
Ross

/\
\/
|\oss Walker

| Department of Molecular Biology TPC15 |
| The Scripps Research Institute |
| Tel:- +1 858 784 8889 |
| EMail:- ross_at_rosswalker.co.uk |
| http://www.rosswalker.co.uk/ | PGP Key available on request |

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu