mpiexec error

[root@c2403 mpich2]# mpiexec -l -n 3  examples/cpi
mpiexec_c2403 (mpiexec 392): no msg recvd from mpd when expecting ack of request

 

This error may deal with the follwing reasons:

Our environment has no high performance network.

IA64 can not surpport Nemesis, I have to change as  --with-device=ch3:ssm

I  found the explaination of nemesis in the file installation manual of mpich2-1.1..... 

 

 

ch3:nemesis  This method is our new, high performance method. It has
been made the default communication channel starting the 1.1 release
of MPICH2. It uses shared-memory to send messages between processes
on the same node and the network for processes between nodes.
Currently sockets and Myrinet-MX are supported networks. It supports
MPI THREAD MULTIPLE and other levels of thread safety.

 

 

 

I larger the value of MPIEXEC_RECV_TIMEOUT in the mpiexec.py file  and solve this issue.

 

 'MPIEXEC_RECV_TIMEOUT'        :  20,             

 

20 is the default value, I modified it as 50.  

It works when the -n 3, but not at -n 18

 

 

[root@c2403 mpich2]# mpiexec -l -n 3  examples/cpi
0: Process 0 of 3 is on c2403
2: Process 2 of 3 is on c2402
1: Process 1 of 3 is on c2401
0: pi is approximately 3.1415926544231318, Error is 0.0000000008333387
0: wall clock time = 0.003711
[root@c2403 mpich2]# mpiexec -l -n 18  examples/cpi
mpiexec_c2403 (mpiexec 392): no msg recvd from mpd when expecting ack of request

 

 

Finally, I set the value as 200, and it seems work when n is 36. As time is considered, it is better to set n as lower value than higer value.