OpenMPI benchmark collection
Currently there are:
- Distributed insertion sort (Not very effective)
- Distributed matrix multiplication (Also not very effective)
- Distributed addition (kind-a effective)
To-be-implemented:
- Distributed Strassen
- Distributed Quick/mergesort
- Distributed DFS over large trees
- Distributed PI
- Distributed solution to SAT/TSP problems (very inefficient)
Let's hope we don't run out of time yet...
To use:
mpirun -np [num of workers per host] --hostfile [path-to-hostfile] ./MPI_Test_Programs [Test|Matrix|Addition|TestConcept|Sort] <Optional args per selection>
Arguments for Matrix:
--size <SIZE>specifies the size of the rectangular matrix A, B="SIZE*SIZE"-G <G>specifies the number of rows distributed to each host. The partition algorithm we use guarantees a minimal fairness of one row per host but may encounter errors when granularity is very, very large.
Arguments for Addition:
<Target>The number we're going to reach.
Arguments for Sort:
...More to be added.