Memory management and performance analysis using the entity‐component‐system (ECS) pattern

Abstract

This dissertation examines the performance implications of using the Entity‐Component‐System (ECS) architectural pattern compared to traditional Object‐Oriented Programming (OOP) within a real‐time simulation context. While OOP encapsulates data and behaviour in class hierarchies, ECS adopts a data‐oriented model that separates logic and state, promoting memory locality and scalability. To evaluate the practical impact of these contrasting paradigms, two deterministic simulations were implemented with identical behaviour but different architectural foundations. The study assesses execution time, memory usage, and scalability across a range of workloads. Results indicate that although OOP performs competitively at small scales, ECS consistently demonstrates superior efficiency and linear scaling as complexity increases. These findings support the conclusion that ECS is better suited to performance‐critical applications, offering tangible advantages in cache utilisation, modularity, and parallelism potential. The research contributes empirical evidence to ongoing discussions around data‐oriented design and highlights the architectural considerations necessary for optimising modern software systems.