The survey was made possible by testing the program on 3x3 and 4x4 diagonally dominant matrixes.
Time here is not optimal and may be different from a device to another.
Execution Time Values
Input size: 3x3
Number of Threads | Execution time (MS) |
---|---|
2 | 8 MS |
3 | 6 MS |
4 | 3 MS |
5 | 11 MS |
7 | 9 MS |
10 | 9 MS |
15 | 10 MS |
Input size: 4x4
Number of Threads | Execution time (MS) |
---|---|
2 | 8 MS |
3 | 7 MS |
4 | 5 MS |
5 | 11 MS |
7 | 12 MS |
10 | 11 MS |
15 | 11 MS |
Speed Up Values (With averaging the parallel execution times)
3x3 input: 11 MS / 8 MS = 1.37
4x4 input: 12 MS / 9.28 MS = 1.29
Notes
The time calculation was recorded for running the algorithm only without calculating the time that was taken by the diagonally dominant checking method.