PGI User Forum
 SearchSearch   MemberlistMemberlist     RegisterRegister   ProfileProfile    Log inLog in 

CUDA-x86.

MPI problems with pgi 7.0-7

 
Post new topic   Reply to topic    PGI User Forum Forum Index -> Licenses and Installation
View previous topic :: View next topic  
Author Message
pekompi



Joined: 15 Nov 2004
Posts: 23

PostPosted: Tue Aug 21, 2007 6:30 am    Post subject: MPI problems with pgi 7.0-7 Reply with quote

Hi,

we are running the old pgi 6.2 stuff on our cluster.
Now I've installed the current version (7.0-7) which is working fine.
In the ne xt step I added the MPI stuff (also downloaded from pgi)
and installed as described in the readme.
Everything seems to be fine, I can compile and also compile
parallel code. But running this code with mpirun
is not working.

If I force local execution everything seems to be normal.
ssh is also working without any problems, also the old
mpi-installation.

taiga:~ # mpirun -np 4 mpihello
hydra.bgc-jena.mpg.de: Connection refused
p0_21213: p4_error: Child process exited while making connection to remote process on hydra: 0
p0_21213: (37.031250) net_send: could not write to fd=4, errno = 32

How can I figure out the reason for this behavior ?



Kind regards,
Peer
Back to top
View user's profile
jtull



Joined: 30 Jun 2004
Posts: 438

PostPosted: Fri Aug 24, 2007 4:42 pm    Post subject: MPI problems with installation Reply with quote

Some questions.

Did you install MPI with ssh or rsh?
Did you install as root?

% more mpihello.f
program hello
include 'mpif.h'
integer ierr, myproc,hostnm
character*64 hostname
call mpi_init(ierr)
call mpi_comm_rank(MPI_COMM_WORLD, myproc, ierr)
! print *, "Hello world! I'm node", myproc
ierr=setvbuf3f(6,2,0)
ierr=hostnm(hostname)
write(6,100) myproc,hostname
100 format(1x,"hello - I am process",i3," host ",A10)
call mpi_finalize(ierr)
end

What happens when you compile and run as follows

pgf90 -o mpihello mpihello.f -Mmpi -Bstatic_pgi -v

mpirun -np 4 mpihello
Back to top
View user's profile
pekompi



Joined: 15 Nov 2004
Posts: 23

PostPosted: Mon Aug 27, 2007 5:21 am    Post subject: Reply with quote

Hi,

I installed everything as root. We are using ssh only.
I downloaded mpich v1 and v2 from the homepage and both are working
WITHOUT any problems ....

0 errors or warnings.

pkoch@tchita:~> mpirun -np 4 mpihello
pc002.bgc-jena.mpg.de: Connection refused
p0_3937: p4_error: Child process exited while making connection to remote process on pc002: 0
p0_3937: (37.121094) net_send: could not write to fd=5, errno = 32
pkoch@tchita:~> ll ./mpihello
-rwxr-xr-x 1 pkoch AG_DV 699120 2007-08-27 14:17 ./mpihello


ssh is working:
pkoch@tchita:~> ssh pc002 w
14:20:39 up 14 days, 3:40, 0 users, load average: 1.00, 1.03, 1.05
USER TTY LOGIN@ IDLE JCPU PCPU WHAT
Back to top
View user's profile
Display posts from previous:   
Post new topic   Reply to topic    PGI User Forum Forum Index -> Licenses and Installation All times are GMT - 7 Hours
Page 1 of 1

 
Jump to:  
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum


Powered by phpBB © phpBB Group