why mpiexec run uncorrectly on smp?

luyut luyut at sina.com
Mon Jun 10 09:09:28 EDT 2002


Hi,everyone, I install the mpiexec with pbs2.3.16, and mpich1.2.3 on 4 nodes of 2cpu smp of p4,when I run mpi program cpi with "#PBS -l nodes=4:ppn=1", which run one task per node,it's ok. But when I run two tasks per node, (with the "#PBS -l nodes=4:ppn=2" in qsub script),it's something wrong. the output file and error file below:
mm.o19:
node3
node3
node2
node2
node1
node1
node0
node0
bm_slave_1_1879: (0.001482) process not in process table; my_unix_id = 1879 my_host=node3
bm_slave_1_1879: (0.001616) Probable cause:  local slave on uniprocessor without shared memory
bm_slave_1_1879: (0.001638) Probable fix:  ensure only one process on node3
bm_slave_1_1879: (0.001659) (on master process this means 'local 0' in the procgroup file)
bm_slave_1_1879: (0.001681) You can also remake p4 with SYSV_IPC set in the OPTIONS file
bm_slave_1_1879: (0.001702) Alternate cause:  Using localhost as a machine name in the progroup
bm_slave_1_1879: (0.001723) file.  The names used should match the external network names.
bm_slave_1_1879:  p4_error: p4_get_my_id_from_proc: 0
p0_1872: (4.022104) net_send: could not write to fd=4, errno = 32
rm_1772: (-) net_recv failed for fd = 3
rm_1772:  p4_error: net_recv read, errno = : 104

mm.e19:
mpiexec: Warning: main: task 0 exited with status 1 (raw 0x1).
mpiexec: Warning: main: task 1 exited with status 1 (raw 0x1).
 
Can anyone tell me how to do with it? thank you.
 
sara
______________________________________

===================================================================
ÐÂÀËÃâ·Ñµç×ÓÓÊÏä (http://mail.sina.com.cn)
ÐÂÀË·ÖÀàÐÅÏ¢£º¶þÊÖÊг¡×ßÒ»×ߣ¬¸Ã³öÊÖʱ¾Í³öÊÖ£¡ (http://classad.sina.com.cn/2shou/)



More information about the mpiexec mailing list