Project Description
Time analysis and comparision of concurrent mergesort with normal mergesort.
Compile and Run
- Compile code using gcc command
- After compilation, run './a.out'
Time comparison of Concurrent and Normal mergesort
For n = 10^5
- Normal mergesort - 0.001 second
- Concurrent mergesort - 17.83 second
For n = 10^7
- Normal mergesort - 0.044 second
- Concurrent mergesort - It will not be able to run on normal computer because it will fork too much for this big input.
Analysis
- Expectation - Concurrent mergesort would run faster because of shared memory.
- Result - Simple mergesort is running faster than concurrent mergesort.
- Reason about above result -
- Concurrent mergesort is accessing left array and right array simultaneously.
- When concurrent mergesort is accessing left array, left array is loaded into cache memory. Simultaneously, concurrent mergesort tries to access right array then cache miss would happen . Because of cache miss happens multiple times, execution of code slows down.
- Whereas in simple mergesort, 2 or 3 elements of left and right array is accessing at a time. therefore that elements can stay in cache and retrieval time of those elements is much lesser than retrieval time of elements in concurrent mergesort.