parallelSorts

Comparison of parallel sort algorithms (bucket sort, merge sort, odd-even sort) implemented with MPI.

#######################################################

Check that it works (for 5 processors and 17 data)

####################################################### NB: the line "CFLAGS+=-DSAVE_DATA" must be uncommented in the Makefile. (This is the case by default)

Compile: if you have already compiled once without -DSAVE_DATA: make clean make

Run: (17 data on 5 processors) mpirun -np 5 ./merge_sort 17 mpirun -np 5 ./odd_even_sort 17 mpirun -np 5 ./bucket_sort 17

This will produce files with the unsorted data and the sorted data. To check that these results are correct, run:

./check_result.py merge_sort_data_P5_N17.txt merge_sort_result_P5_N17.txt ./check_result.py odd_even_sort_data_P5_N17.txt odd_even_sort_result_P5_N17.txt ./check_result.py bucket_sort_data_P5_N17.txt bucket_sort_result_P5_N17.txt

This should output "OK"

###################################

Plot speedups (for 10000000 data)

################################### WARNING: If you want to run the program with a lot of data (like 10000000), comment the line "CFLAGS+=-DSAVE_DATA" in the Makefile and do a make clean --> Otherwise the program will try to create really big data/result files.

Compile: make Run: ./run.bash 10000000

Plot the results: On a machine with python and matplotlib installed: In the Project directory, launch ./plot_speedup.py 10000000 This should create 3 png files containing graph of the speedup