Compiled below is a list of facts you need to be aware of when trying to compile, link, and run, programs on the IBM pSeries system :
The pSeries contains IBM P4 processors. The clock speeds for most of these processors is 1.1 GHz (there are some machines with faster 1.7 GHz processors but they have more limited access).
If you have been running jobs on a personal computer with more recent
and speedier processor, you may experience a slowdown, rather than a speedup,
of your code that you might come to expect of a "supercomputer," such as
the pSeries computer. Even in that case however, you could gain in throughput if your code can
take advantage of parallel processing. How much speed gain depends
on the clock speed of your current machine, the parallel
efficiency of your code, and how many processors you use. If you are
unfamiliar with parallel programming and if you need help, please contact
Kadin Tseng (kadin@bu.edu) or Doug Sondak
(sondak@bu.edu) to discuss your
computing goals and computing resource needs.
In addition, which of the three SCV systems may be more appropriate for you is another issue to consider. As a reference, we have a comparison of the performance of the
three SCV systems.
Runtime limit is 5 hours for all multiprocessor queues. (More details ...)
You can request up to 16 nodes.
MPI and OpenMP are the primary means of multiprocessing. The languages supported for MPI and OpenMP programming are FORTRAN, C, and C++.
Parallel mathematical library is available. It is the IBM PESSL.
MPMD (Multiple Program Multiple Data) paradigm is supported.
If your program is written in C++, where necessary you need to include the header file mpi.h as usual. In addition, to turn
on the headers for C++ in the mpi.h file, the following flag
must be included during compilation:
twister % mpCC -D_MPI_CPP_BINDINGS ...
Many MPI-2 functions are supported, including MPI_Put and MPI_Get.
For 64-bit arithmetic, compile code with the -q64 flag.
