Search algorithm, is an efficient algorithm, which performs an important task that locates specific data among a collection of data. Often, the difference among search algorithms is the speed, and the key is to use the appropriate algorithm for the data set. Binary search is the best and fastest search algorithm that works on the principle of ‘divide and conquer’. However, it needs the data collection to be in sorted form, to work properly. In this paper, we study the efficiency of binary search, in terms of execution time and speed up, by evaluating the performance improvement of the combined search algorithms, which are sorted into three different strategies: sequential, multithread, and parallel using message passing interface. The experimental code is written in ‘C language’ and applied on an IMAN1 supercomputer system. The experimental results show that the decision variables are generated from the IMAN1 supercomputer system, which is the most efficient. It varied for the three different strategies, which applies binary search algorithm on merge sort. The improvement in performance evaluation gained by using parallel code, greatly depends on the size of data set used, and the number of processors that the speed-up of the parallel codes on 2, 4, 8, 16, 32, 64, 128, and 143 processors is best executed, using between a 50,000 and 500,000 dataset size, respectively. Moreover, on a large number of processors, parallel code achieves the best speed-up to a maximum of 2.72.