Developing and running Parallel application with MPICH

1. Go get Ubuntu at http://www.ubuntu.com/getubuntu/download

I am using Desktop version version 8.10, Intrepid Ibex

2. Linux environment setup

a. setup NFS server on master node (ub0)

a.1 install NFS server

sudo apt-get install nfs-kernel-server

a.2 sharing Master folder

add one line to /etc/exports

sudo gedit /etc/exports

/mirror *(rw,sync)

a.3 create folder

sudo mkdir /mirror

a.4 start NFS service on master node

sudo /etc/init.d/nfs-kernel-server start

b. mounting master shared folder in other non-master nodes

b.1 add the ip address of ub0 in /etc/hosts

sudo gedit /etc/hosts

you will find

127.0.0.1 localhost

192.168.0.71 muhammad-desktop

192.168.0.70 ub0

Note: if you found that your IP is 127.0.1.1 please change it to your IP

b.2 create mount point and mount it

sudo mkdir /mirror

sudo mount ub0:/mirror /mirror

c. create a universal user mpiu on all nodes

sudo chmod 777 /mirror

sudo useradd -gusers -s/bin/bash -d/mirror -m mpiu

d. Setting up SSH with no pass phrase for communication between nodes

d.1 login as mpiu

sudo su - mpiu

d.2 generate DSA key for mpiu:

ssh -keygen -t dsa

Note: Leave passphrase empty.

d.3 add content in id_dsa.pub to authorized keys

cd .ssh

cat id_dsa.pub >> authorized_keys

d.4 test

ssh ub1

3. Download MPICH-2 and build it

resources: MPICH-2 document
http://www.mcs.anl.gov/research/projects/mpich2/documentation/index.php?s=docs

downloads:

http://www.mcs.anl.gov/research/projects/mpich2/downloads/index.php?s=downloads

3.1 Installing GCC

sudo apt-get install build-essential

3.2 Installing MPICH2

cd /mirror

mkidr mpich2

tar xvf mpich2- 1.0.7rc1.tar.gz

cd mpich2 -1.0.7rc1

./configure -- prefix=/mirror/mpich2

make

sudo make install

3.3 setup linux environment variables in .bash_profile

add two lines into .bash_profile

export PATH=/mirror/mpich2/bin:$PATH

export LD_LIBRARY_PATH=/mirror/mpich2/lib:$LD_LIBRARY_PATH

update PATH in /etc/environment (variables defined there will

automatically apply at login for all sessions)

sudo gedit /etc/environment

you will find

PATH="/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games"

add :/mirror/mpich2/bin into the end PATH

to be something like that

PATH="/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/mirror/mpich2/bin"

4. Setup MPICH-2 MPD

4.1 Create mpd.hosts in mpiu's home directory with nodes names, for

example

ub0

ub1

4.2 Create some credential

echo secretword=something >> ~/.mpd.conf

chmod 600 .mpd.conf

4.3. Start MPI MPD on the cluster for all nodes

mpdboot -n <number of hosts defined in mpd.hosts>

mpdtrace

The output should be name of all nodes.

if this does not work

we can start the mpds manually

go to the master node (ub0)

mpd &

it will output something like that

  [1] 9230

  mpdtrace -l  it will output something like that
  muhammad-desktop_43814 (192.168.0.71)

  note: the Master serves at the port 43814 we get it from    muhammad-desktop_43814(192.168.0.71)

  go to the slave nodes and connect to the master (ub0) mpd daemon
  mpd -h ub0 -p 43814 &
  then list all the cluster mode with this command
  mpdtrace -l

  it will output like this

  muhammad-desktop_43814 (192.168.0.71)
mpiuser_43932 (192.168.0.70)
4.4. compile an example and run it

in mpich2-1.0.7rc1/examples",there is a parallel program to calculate PI (cpi)

copy cpi.c to /mirror

mpicc cpi.c -o cpi

mpiexec -n 4 ./cpi

it should display something like:

0: pi is approximately 3.1415926...., Error is .....

0: wall clock time = ....

troubleshoot in mpich2-doc-install.pdf Appendix A: Troubleshooting MPDs

HelloWorld.c (from wiki)
/*
"Hello World" Type MPI Test Program
*/
#include <mpi.h>
#include <stdio.h>
#include <string.h>

#define BUFSIZE 128
#define TAG 0

int main(int argc, char *argv[])
{

char idstr[32];

char buff[BUFSIZE];

int numprocs;

int myid;

int i;

MPI_Status stat;

MPI_Init(&argc,&argv); /* all MPI programs start with MPI_Init; all 'N'

processes exist thereafter */

MPI_Comm_size(MPI_COMM_WORLD,&numprocs); /* find out how big the SPMD

world is */

MPI_Comm_rank(MPI_COMM_WORLD,&myid); /* and this processes' rank is */

/* At this point, all the programs are running equivalently, the rank

is used to

distinguish the roles of the programs in the SPMD model, with rank 0

often used

specially... */

if(myid == 0)

{

printf("%d: We have %d processors\n", myid, numprocs);

for(i=1;i<numprocs;i++)

{

sprintf(buff, "Hello %d! ", i);

MPI_Send(buff, BUFSIZE, MPI_CHAR, i, TAG, MPI_COMM_WORLD);

}

for(i=1;i<numprocs;i++){

MPI_Recv(buff, BUFSIZE, MPI_CHAR, i, TAG, MPI_COMM_WORLD, &stat);

printf("%d: %s\n", myid, buff);

}

}

else

{

/* receive from rank 0: */

MPI_Recv(buff, BUFSIZE, MPI_CHAR, 0, TAG, MPI_COMM_WORLD, &stat);

sprintf(idstr, "Processor %d ", myid);

strcat(buff, idstr);

strcat(buff, "reporting for duty\n");

/* send to rank 0: */

MPI_Send(buff, BUFSIZE, MPI_CHAR, 0, TAG, MPI_COMM_WORLD);

}

MPI_Finalize(); /* MPI Programs end with MPI Finalize; this is a weak

synchronization point */

return 0;

}