Programming Techniques for Supercomputers - CAM project

Background: Modelling 2D steady-state heat equation

This code project for the 10 ECTS PTfS module solves a linear system of equations to find a steady-state temperature distribution on a rectangular plate using the Conjugate Gradient (CG) and Preconditioned Conjugate Gradient (PCG) with symmetric Gauss-Seidel preconditioning methods.

Please see the tutorials for more information.

Usage

1. Compile program by calling

   CXX=<compiler icpx or g++> make

   To print additional debug output during the execution, compile with -DDEBUG:

   CXX=<compiler icpx or g++> make EXTRA_FLAGS=-DDEBUG

   To compile with LIKWID use:

   LIKWID=on CXX=<compiler icpx or g++> make

2. Run tests by running

   ./test

3. Run performance tests by running

   ./perf <grid size y> <grid size x>

4. Make plots (only when compiled with --DDEBUG):

   ./test
   ./plot.sh

   This generates some .png files showing you calculations. As a reference, we provide images in the folder resources.