Dear Amber users,
In the amber manual regarding the MPI version, it says that
"The code is pure SPMD (single program multiple data) using a
master/slave, replicated data model. Basically, the master node does all
of the initial set-up and performs all the I/O". A computer guy asked me
whether the I/O is done by the master node (in a batch mode) or via
the shell node (in an interactive mode). I do not know the answer. Could
some expert in the field give me some explanation? Secondly, do MPICH
requires a shared filesystem (instead of distributed filesystem)?
Thanks a lot for your help,
Jian Hui Wu
Received on Wed Aug 07 2002 - 21:24:13 PDT
