I have same problem. I am wondering if sander_classic is fully parallel or
not. "dhfr/Run.dhfr.classic" of test directory doesn't included
$DO_PARALLEL, and so I am thinking we can't use PME for parallel sander_classic.
http://amber.ch.ic.ac.uk/parallel.html says, "the Particle Mesh
Ewald code contributed by Tom Darden (NIEHS) is not MPI parallelized in the
AMBER 4.1 release", but I don't know for sander_classic.
Does someone know about the limitation of sander_classic in parallel?
Osamu
At 18:31 01/02/23 +0100, Khatcharin Siriwong wrote:
>Dear Sir,
>
>My machine is SGI IRIX6.5 mpich library, 10 processors.
>I have compiled Amber6 with file "Machine.sgi_mpich".
>When I tested program by "make test.sander", every thing has PASSED.
>
>Note: I set ==> setenv DO_PARALLEL "mpirun -np 4 " before testing.
>
>When I tested with "make test.sander_classic", a file "prowat/Run.plast"
>has PASSED, while a file "dhfr/Run.dhfr.classic" could not be run.
>I obtained message
>
> Bus error - core dumped
> program error
>
>and not obtained any result in mdout and restrt files.
>
>Moreover, I tried to run sander_classic with small DNA (the steps of
>simulations were carried out following tutorial in amber web site).
>The initial dynamics could be run. But when I turned on PME
>(like dhfr/Run.dhfr.classic), I obtained the same message,
>"Bus error - core dumped"
>
>Could you please give me the suggestion?
>
>Sincerely yours,
>Khatchatin Siriwong
Received on Fri Feb 23 2001 - 09:58:47 PST
