PC GAMESS/Firefly-related discussion club



Learn how to ask questions correctly

MPICH version full statically linked

Pasquale Morvillo
pasquale.morvillo@portici.enea.it

Hi,
I was trying to use Firefly (pcgamess_71f_linux_mpich_rsh_p4) on our system (Linux cluser, Processors: 4 Xeon Quad-Core Tigerton E7330 per node, connection: infiniband ).

I am using 4 nodes (16 cpu per node).
Sometimes the execution is fine (normally terminated).
Other times, it stops.
The following errors are obtained:

p15_32200: (1181.391463) net_recv failed for fd = 21
p15_32200:  p4_error: net_recv read, errno = : 110
p47_19582: (1181.376777) net_send: could not write to fd=4, errno = 104
p47_19582:  p4_error: net_send write: -1
   p4_error: latest msg from perror: Connection reset by peer
SIGINT caught, exiting.
SIGINT caught, exiting.
SIGINT caught, exiting.
SIGINT caught, exiting.
SIGINT caught, exiting.
SIGINT caught, exiting.
SIGINT caught, exiting.
SIGINT caught, exiting.
SIGINT caught, exiting.
SIGINT caught, exiting.
SIGINT caught, exiting.
SIGINT caught, exiting.
SIGINT caught, exiting.
SIGINT caught, exiting.
SIGINT caught, exiting.
bm_list_32201: (1184.210353) net_send: could not write to fd=8, errno = 9
bm_list_32201:  p4_error: net_send write: -1

The input keywords are:

$CONTRL RUNTYP=ENERGY SCFTYP=RHF DFTTYP=B3LYP5 COORD=CART $END
$BASIS GBASIS=N21 NGAUSS=3 $END
$SYSTEM MWORDS=20 TIMLIM=10000 $END
$SCF DIRSCF=.TRUE. $END
$p2p p2p=.t. dlb=.t. xdlb=.t. net=0a000000 mask=ff000000 $end
$smp call64=.t. $end


Alex Granovsky suggested me to add the following lines to the input file:

$mpi mxgsum=2000 $end
$system mpisnc=.t. mxbcst=2000 lensnc=2000 $end

Now, the execution is fine.
I hope that this experience can help someone.

[ This message was edited on Tue Apr 7 '09 at 6:46pm by the author ]

[ Previous ] [ Next ] [ Index ]           Tue Apr 7 '09 6:46pm
[ Reply ] [ Edit ] [ Delete ]           This message read 1301 times