Áõ¹ó±ó wrote:
> I did this:
> [basic]$ lamboot -v hostfile
>
> LAM 7.1.1/MPI 2 C++/ROMIO - Indiana University
>
> n-1<3871> ssi:boot:base:linear: booting n0 (127.0.0.1)
> n-1<3871> ssi:boot:base:linear: finished
>
>
> But it still seemed that lam was booted on only one cpu.
> Becase when I ran "mpirun C cpi", the output is :
> Process 0 of 1 on n0
> pi is approximately 3.1415926544231341, Error is 0.0000000008333410
> wall clock time = 0.000347
>
> But when I ran "mpirun -np 4 cpi", the output is :
> Process 2 of 4 on n0
> Process 0 of 4 on n0
> Process 1 of 4 on n0
> Process 3 of 4 on n0
> pi is approximately 3.1415926544231239, Error is 0.0000000008333307
> wall clock time = 0.000196
>
> What's wrong in the world?
You have booted just one node. When you ask for 4 processes, they are
distributed among the nodes which are running (only 1). This is fine,
when your one node has 4 cores. If you have more than one node, heed
the advice you were given about hostfile and lamboot diagnostics.
|
