MSc in High Performance Computing Coursework for Threaded Programming Part 1
The object of this assessment is to parallelise a simple piece of code using OpenMP. This assessment in worth 10% of the total marks for the course.
You are provided with a piece of code which you should parallelise with OpenMP directives.
The code can be found on the course pages on Learn. You may choose to work with either the C (C/adi.c) or the Fortran (Fortran90/adi.f90) version.
You should use the Intel 19 compilers to build the code. On Cirrus, type module load intel-compilers-19
before building the code. There is a Makefile supplied that shows how the marker will build your code for correctness testing.
Parallelisation
Add OpenMP directives to parallelise all the computation between the calls to the timer routine omp get wtime(). Your are not expected to otherwise optimise the code, or modify it except for the addition of OpenMP. Your solution will be marked for correctness, by running it on up to 36 threads. The diff value printed should be identical to that of the sequential code for all thread numbers. You will be given credit for keeping the changes made to the original C or Fortran code to a minimum. Your code will not be marked directly for performance, but you will be given additional credit for minimising the number of implicit barriers that are executed, and you will be penalised if the performance is very poor (i.e. worse than the sequential code).
