Demonstrating Hundreds of AIs in One Scene

摘要

This demo shows how a server can process heavy, game-enhancing workloads without sacrificing response time by implementing a highly parallelisable game server architecture. The architecture " LEARS" allows for the utilisation of more processing resources for tasks that will enhance the game experience. Using the A~* path-finding algorithm, we demonstrate how the server distributes the resources under different levels of load, and how this impacts the response time for the game-clients. The demo allow the viewers to adjust game parameters such as the number of pathfinding entities, the number of player characters and the number of available threads and observe the impact on response time and CPU utilisation.