Repository for the PythonThreading research project for CS 470 at James Madison University
In our project, we attempt to compare the difference in runtime of parallel and serial computations between C and multiple Python implementations. The programs create an of size n containing random numbers, then performs many iterations of computations, and stores them in an array. Ultimately, the program will iterate n*n times over the arrays, and prints the run time of the function. There is a script included that runs all programs with inputs of varying inputs.
To begin, start of with running the provided make file
make
gcc -g -O2 --std=c99 -fopenmp -Wall -o CBenchmark CBenchmark.c
gcc -g -O2 --std=c99 -Wall -o CBenchmark_Ser CBenchmark_Ser.cIt should have compiled the CBenchmark.c and CBenchmark_Ser.c files and created an executable for each
Once you have the two executables in your directory, the files are ready to be ran
To run the serial C executable:
./CBenchmark_Ser <Number of Arugments>to run the parallel C executable:
OMP_NUM_THREADS=<NUMBER_OF_THREADS> ./CBenchmark <Number of Arugments>To run the serial Python program:
python3 PyBenchmark_Ser.py <Number of Arugments>To run the Multiprocess Python program:
python3 PyBenchmark.py <Number of Processes> <Number of Arugments>To run the Multithread Python program:
python3 Py_Thread.py <Number of Threads> <Number of Arugments>To run pypy3, we have provided the absoulte path for it to work on the cluster
Serial Pypy example:
pypy3.6-v7.3.3-linux64/bin/pypy3 PyBenchmark_Ser.py <Number of Arugments>Multiprocesses Pypy program:
pypy3.6-v7.3.3-linux64/bin/pypy3 PyBenchmark.py <Number of Processes> <Number of Arugments>To run cython with the provided provided, we have to use the absolute path. Note that we use have to use python not python3
Serial Cython example:
python Cython-0.28.1/Cy_ser_test.py <Number of Arguments>Multiprocess Cython example:
python Cython-0.28.1/Cy_multi_test.py <Number of Processes> <Number of Arguments>