MJ's Benchmark: Lid-driven cavity flow

About this project

This simple benchmark is a good test to show how depend the execution time on mesh size and processor number. This case contains six meshes, with cell sizes from a half million to 8 million. Also every mesh will be benchmarked with all combinations of cpu counts. the mesh decomposition algorithm is scotch, it tries to minimize the faces between the processors. (the best algorithm for speed up ) In the end there will be 36 results to determine the speedup and give an estimation of how much processors are needed for a desired mesh. FYI: difference between executionTime and clockTime clockTime mean how long it took to complete the task, inclusive interconnection between processors, operating system need also some cpu resources and other tasks... executionTime mean only all the time the simulation used on all processors