Web search engines have developed into a crucial tool for effectively and quickly locating information among the vast web data. The PageRank algorithm is essential for web search as it measures the importance and relevance of web pages based on their incoming links, allowing search engines to rank results by prioritizing high-quality and authoritative content. This ensures that users receive more accurate and valuable information. The PageRank algorithm improves search engine results by assigning a numerical weight to each element in a hyperlinked set of documents, effectively measuring the importance of web pages based on quantity and quality of links pointing to them. With billions of web pages and an extensive network of hyperlinks, the traditional sequential computation of PageRank becomes impractical for timely and efficient processing. Parallelization allows the algorithm to be distributed across multiple processors or computing nodes, enabling simultaneous computation of PageRank scores for different web pages. The main objective of this work is to parallelize the PageRank algorithm using OpenMP, MPI and CUDA and to compare their execution time to find the optimal one. OpenMP simplifies shared-memory parallelism, MPI facilitates communication between distributed processes and CUDA harnesses GPU power for high-performance parallel processing in diverse parallel computing environments. The experimental results demonstrate notable performance enhancements through parallelization using different technologies: OpenMP improves the algorithm's performance by 49.7%, MPI by 62.4%, and CUDA by 84.3%. Hence, optimal results are found when the PageRank algorithm is parallelized using CUDA.