How to change from serial to parallel mode on cluster?

[Alauda]

Hi, I compiled a code on a Linux cluster machine with mpich and ifort. A test run shows that the program works fine. However, it seems that the program is running in serial instead of parallel mode: reducing the number of CPU's from 32 to 3 does not increase the time it takes to run the simulation. Infact, the simulation is slightly faster with 3 compared to 32 CPU's. Unfortunately I have very little experience on the subject. My questions are:
(1) How can I be 100% sure my code is running in parallel and not in serial mode?
(2) If the code is running serial, what do I need to do to make it run in parallel mode?
I am grateful for any hints. With kind regards, Chris.

Details:
For submission I use a jobscript which includes the following lines:
#!/bin/sh
#$ -S /bin/sh
#$ -cwd
#$ -pe mpich 32
mpirun ./script.sh

...and: Red Hat 4.1.2-46

[Rubberman]

Without looking at your entire mpi source code and configuration scripts no one can tell you what is going wrong here. It may be that your code isn't conducive to parallelization as well, at least as written.