<html>
<body>
Hello to everybody,<br><br>
we have more or less the same problems. We are developing a FDTD code for
electromagnetic simulation in FORTRAN. The code is mainly based on a 3
loops used to compute the electric field components, and 3 identical
loops to compute the magnetic field components. <br><br>
We are using a small PC cluster made with 10 PIV 3GHz connected with a
1Gbit/s ethernet LAN built some years ago, and a Intel Vernonia 2
procesors / 4 core each (total 8 core). The processors are Intel Xeon
E5345 @ 2.33GHz. <br>
We are using the Intel 10.1 fortran compiler (compiler options as
indicated in the manual for machine optimization, with -O3), ubuntu 7.10
(kernel 2.6.22-14 generic on the cluster, kernel 2.6.22-14 server on the
multiprocessor machine).<br>
mpich2 is compiled with nemesis, and we are still with the 2.1.06p1
(still no time to upgrade to the last version) <br><br>
Testing the code for a (not too big, to keep the overall time limited)
simulation (85184 variables 44x44x44 cells, 51000 temporal iterations) we
had a good scaling on the cluster. On the total simulation time
(with parallel and sequential operations mixed) we have a speed-up of 8.5
using 10PEs ( 6.2 with 9, 8.2 with 8, 5 with 7, 5.8 with 6 etc ...).
<br><br>
The same simulation has been run on the 2PEs/quad core machine but we
didn't have good performances. <br>
The speed up is 2 if we run mpiexec -n 2 .... as the domain is divided
between the two processors which seems to work independently. But, by
increasing the number of processors (core) used, running the simulation
with .n 3, -n 4 etc ... we have a speed-up of 2.48 with 4 cores (2 on
each PE), but only 2.6 with 8 PEs.<br><br>
We also tried to use -parallel or -openmp (limiting the openmp directives
only in the loops of field computations), without obtaining significant
changes in the performances, both running with mpiexec -n 1 or mpiexec -n
2 (trying to mix mpi and openmp).<br><br>
Our idea is that we have serious problems in managing the shared
resources for memory access, but we have not expertise on that, and we
could be totally wrong. <br><br>
Regards.<br><br>
Gaetano<br>
<x-sigsep><p></x-sigsep>
<hr>
<tt>Gaetano Bellanca - Department of Engineering - University of
Ferrara <br>
Via Saragat, 1 - 44100 - Ferrara -
ITALY
<br>
Voice (VoIP): +39 0532 974809 Fax:
+39 0532 974870<br>
<a href="mailto:gaetano.bellanca@unife.it" eudora="autourl">
mailto:gaetano.bellanca@unife.it</a> <br>
<hr>
</body>
</html>