Hi everyone,
here's the patch, since I cant check in stuff anymore. This removes the
last three lines erroneously introduced to runmd.f in my rev 10.35. This
was experimental code that slipped into my commit without being fully
tested. Applying this to runmd.f makes the 4 cpu failures in the softcore
test suite go away.
*** 2181,2198 ****
if( icfe /= 0 .and. mod(nstep+1,20) == 0 ) then
! In dual-topology this is done within softcore.f
- ! also, velocities are treated as well as coords.
if (ifsc /= 1) then
if( master ) call mpi_bcast(x,nr3,MPI_DOUBLE_PRECISION, &
0,commmaster,ierr)
else
if( master ) call sc_sync_x(x,nr3)
- if( master ) call sc_sync_x(v,nr3)
end if
if( numtasks>1 ) call mpi_bcast(x,nr3,MPI_DOUBLE_PRECISION, &
0,commsander,ierr)
- if( numtasks>1 ) call mpi_bcast(v,nr3,MPI_DOUBLE_PRECISION, &
- 0,commsander,ierr)
end if
call timer_stop(TIME_DISTCRD)
--- 2160,2173 ----
Kind Regards,
Thomas
Dr. Thomas Steinbrecher
BioMaps Institute
Rutgers University
610 Taylor Rd.
Piscataway, NJ 08854
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Thu Mar 11 2010 - 00:30:02 PST
