Dear Amber,
I have installed Amber10 in our cluster but both "sander.MPI and pmemd"
are giving the error "Fatal error in MPI_Init: Other MPI error"
details of installation:
program: amber10 with ambertools 1.2
Processor model name : AMD Opteron(tm) Processor 6128
compilers: ifort (IFORT) 10.1 20080112
MPI: mvapich2-1.7a2 for Infiniband support
Compilation procedure:
1)extracted amber and tools from tar balls.
2)wget
http://ambermd.org/bugfixes/AmberTools/1.2/bugfix.all
3)patch -p0 <bugfix.all
4)rm -f bugfix.all
5)wget
http://ambermd.org/bugfixes/10.0/bugfix.all
6)wget
http://ambermd.org/bugfixes/10.0/apply_bugfix_all.x
7)chmod 700 apply_bugfix_all.x
8)./apply_bugfix_all.x bugfix.all
9)cd $AMBERHOME/src/
10)./configure_at gcc
11)make clean -f Makefile_at
12)make -f Makefile_at
13)./configure_amber ifort
14)make clean
15)make install
16)export MPI_HOME=/share/apps/mvapich2/
17)./configure_amber -mpich2 ifort_x86_64
18)make clean
19)make parallel
20)cd $AMBERHOME/src/pmemd
21)./configure linux_em64t ifort mpich
22)Here I have followed Prof Ross Walker's suggestion
(
http://archive.ambermd.org/200805/0631.html) and could install pmemd
successfully.
modified config.h for vectorizeation that is compatible with AMD chip.
23)make install
24)got the error when using pmemd and sander.MPI
All the steps above are successful without errors. tests are also passed
for all serial programs. But I cannot understand why sander.MPI and pmemd
are giving the fatal error "Fatal error in MPI_Init: Other MPI error".
I am also getting the standard output as:
=====================================================================================
= BAD TERMINATION OF ONE OF YOUR APPLICATION PROCESSES
= EXIT CODE: 256
= CLEANING UP REMAINING PROCESSES
= YOU CAN IGNORE THE BELOW CLEANUP MESSAGES
=====================================================================================
...
I am more puzzled when I could successfully install sander.MPI and
pmemd and they works fine in another cluster with the same system
configuration.
any suggestions?
Santosh
--
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Jul 05 2011 - 12:00:03 PDT
