Parallelized-Bucket-Sort

Parallelizing Bucket Sort with MPI

Done as an assignment for 491: Parallel Processing at MSU. The code in bucket_sort_skeleton.c was provided to us, the v1, v2, & v3 files are my solutions. These were run on MSU's HPCC to compare run-times.

v1:

1. generate: create the array at the root process
2. bin: root process determines buckets -- which data belongs on which processor
3. distribute: root sends buckets to appropriate processes
4. local sort: each process sorts the data locally using qsort()
5. gather: results are gathered at the root process

v2:

1. generate: create N/P elements randomly on each process
2. bin: each process determines buckets - which of their data belongs on which processor
3. distribute: each process sends buckets to appropriate processes
4. local sort: each process sorts the data locally using qsort()
5. gather: results are gathered at the root process

v2.1: Different data distribution: random distribution to create unbalanced load on each processor

v3: Pivots with sampling to deal with imbalanced loads:

1. Each process selects S/P elements randomly within its part of array A
2. S samples are gathered at the root process
3. root process sorts samples locally using qsort()
4. root selects and broadcasts pivots = [S_sorted[S/P], S_sorted[2S/P], S_sorted[3S/P] ...]