Examining Sorting Algorithm Performance Under System Load

Abstract: In this thesis, we conducted tests in order to determine how system load affected the speed of commonly used algorithms. The difficulty of obtaining these type of results theoretically due to a large number of factors that can affect performance motivated the use of simulation. An implementation was constructed that ran the sorting algorithms Mergesort, Quicksort and Radixsort while the system was put under different levels of stress. This stress was created by generating cache misses using the software stress-ng. The resulting completion times, as well as the cache activity was logged. The result of the simulations were mostly within expectations, faster algorithms still performed better even under heavy loads and even handled the load better than the competition