Repository displays C++ implementation of serial and multithreaded implementation of quicksort

Project supports using docker for running containerized builds and tests. After cloning the repo, use the following commands to run a containerized build and all the tests from the project root folder:

docker build -t buildimage . --target build

docker run -v ~/QuickSort:/app buildimage

docker build -t testimage . --target test

docker run -v ~/QuickSort:/app testimage

Make sure to mount the correct volume in the above commands in case of cloning project root in a folder other than the HOME directory.

In order to compile the threaded implementation, install OpenMP library first from the web. Once OpenMP is installed, to compile directly from terminal, use -fopenmp flag while compiling multithreaded source files.

For example,

/usr/bin/g++ -I/home/gaurav/QuickSort/include -fopenmp -o /home/gaurav/QuickSort/src/runScript_threaded /home/gaurav/QuickSort/src/runScript_threaded.cpp

command would be required to compile the threaded source file.

If using CMAKE for build process, type cmake .. from the top level build directory for out of source builds.

For enabling/disabling threaded implementation, set the option THREADS_ENABLED to True/False in the top level CMakeLists.txt file.

After completing generation of MAKEFILE in the build directory, simply type make and specified targets will be built in the build directory.

You can specify the number of threads used by OpenMP via the following command:

export OMP_NUM_THREADS="YOUR SPECIFIED NUMBER OF THREADS"

Implementation Limitations:

This example is just a demonstration of performing multithreading via tasking based queue implementation using OpenMP. In general, this implementation would suffer from contention (due to locking) on the queue used for storing the tasks. As a result, speedups might be limited for higher number of threads.