Readme file for the Parallel code.
Only files for N=1024 are available in the git repo.

Type1 - Parallel code where the calculation of alphak, betak etc is done only on one processor and is then broad cast to all the processors.

Type2 - Parallel code where the caluclation of alphak, betak etc is done by all processors in a row and is collected by all_reduce rather than bcast.

Matrix_A: Folder with all the A matrices.
Matrix_b: Folder with all the b matrices.
Matrix_x0: Initial x for the differnet cases. vector of 0 and corresponding size.
Matrix_xinit: Final result vector that should be obtained from the CGSolver. Used to estimate the error between parallel code and MATLAB.

The number following the files is the size of the matrix A. Eg. MainParallel1024 => the size of A is 1024 X 1024.

Following is the map of the result files. The number following each batch script indicates the job number and the corresponding output file can be found in slurm-<jobnuber>.out

Type 1:
StrongScaling_Odd_10240.sh 6428480
StrongScaling_Odd.sh 6428481
StrongScaling_EvenPow_10240.sh 6428482, Only for np=9 and 15 and N=10239 with load balance. 6428781
StrongScaling_EvenPow.sh 6428483
WeakScaling.sh 6428484

Type 2:
WeakScaling 6427199
StrongScaling_EvenPow_10240.sh 6427201, 6428769
StrongScaling_EvenPow.sh 6427202
StrongScaling_Odd.sh 6427203
StrongScaling_Odd_10240.sh 6427205