Scalable Parallel Processing: Architectural Models, Real-Time Programming, and Performance Evaluation

• Published in: Engineering Proceedings
• Main Authors: Mirela Sino, Ervin Domazet
• Format: Article
• Language: English
• Published: MDPI AG 2025-08-01
• Subjects: parallel processing; SIMD; MIMD; parallel programming real-time; parallel sort algorithm
• Online Access: https://www.mdpi.com/2673-4591/104/1/60

This research paper analyzes and highlights the benefits of parallel processing to enhance performance and computational efficiency in modern computing systems. It explores two primary models of parallelism—single instruction, multiple data (SIMD) and multiple instruction, multiple data (MIMD)—by examining their architectures and real-world use cases such as artificial intelligence, image processing, and cloud computing. Special emphasis is placed on the role of parallel programming in real-time systems, with a focus on APIs such as OpenMP and Ada, which facilitate structured parallelism. To demonstrate the practical advantages of parallelism, a comparative study is presented between a parallel merge-sort algorithm and its serial counterpart. Experimental analysis across datasets ranging from 100,000 to 1,000,000 elements shows that execution time can be reduced by up to 60–70% when using eight-core parallelization compared to serial execution. These results illustrate the scalability and effectiveness of parallel processing in handling large-scale computations.