This is the course project of HPC at M1 Sorbonne University. Consisting of resolve a sparse matrix multiplication system A*x = y with conjugate gradient method. The algorithm was derived from Wiki page and be parallelized with MPI and Open MP derivatives. Testing matrix are available in https://hpc.fil.cool/matrix/
-
cg_seq: containing the sequential solution
-
cg_mpi: optimized by MPI
-
cg_hybrid: optimized by MPI+OpenMP
-
ToDo: Vectorization
In each repository, using make command to compile the program. Testing matrix to be downloaded somewhere and correct the command_line variable in runner.py to fit the path to the testing matrix. Run python runner.py command to generate a receipt file and run python check_receipt.py [receipt_file] to check the solution and the preformance.