Exercises: Requesting threads and nodes.

Make a copy of the /shared/ucl/apps/examples/openmp_pi_dir directory. Then, build the program using make.

Exercise 1:

Modify the job script from the previous exercise to run the new program.

Exercise 2:

Run versions with 1, 2, 3, and 4 cores, and compare the timings.

Make a copy of the /shared/ucl/apps/examples/mpi_pi_dir directory. Then, build the program using make.

Exercise 3:

Modify the job script to run the new program, using 4, 8, 12, and 24 cores and the mpi parallel environment.

Exercise 4:

Try modifying the serial job script (calculate_pi) to run 4 jobs as an array.

Exercise 5:

calculate_pi can take an argument to tell it how many steps to use. Try using this with $SGE_TASK_ID to run using 300, 500, and 700 steps.