The program has a function that takes one parameter, substitute it in a linear equation and return the value. This is done for 100000 numbers. This task is parallelized using ParallelContext(). The programming language used here is hoc.

The command used to run the program is mpiexec -n 4 nrniv -mpi squaredTESTER.hoc

Here, 4 represent the number of cores to be used to perform the task. When the number of cores are reduced, the program is executed faster.

What is the reason for this ?

The code that i ran is given below

whole_time=startsw() // start time

func f() { local s, y              // a function with no context that *CHANGES*

    s = $1

    // a simple linear equation

    // y = 0.0001*s + 0.0456
    y = 0.00001*s + 0.456703


    return y         //except its argument
}

objref pc
pc = new ParallelContext()
pc.runworker()          // mkaster returns immediately, workers in
                        // infinite loop running and jobs from bulletin board


first = 1
last = 100000

s = 0
if (pc.nhost == 1) {    // use the serial form
   for i=first, last {
      s += f(i)
   }
}else{                  // use the bulletin board form
   for i=first, last {        // scatter processes
      pc.submit("f", i) // any context needed by f had better be
   }                    // the same on all hosts
   while (pc.working) { // gather results
      s += pc.retval    // the return value for the executed function
   }
}

print s
pc.done                 // tell workers to quit

end_time=startsw()-whole_time // startTime - endTime

print "----------------------------"
print end_time

Command to run the code : mpiexec -n 4 nrniv -mpi squaredTESTER.hoc