Submission #4: Load Balancing Real-Time Physics for Cloud Based Video Game Delivery =================================================================================== Authors ------- 1. Alexander Brown (Newcastle University) 2. Graham Morgan (Newcastle University) Abstract -------- In this paper we present a solution to scaling real-time physics simulations using cloud computing. Although high performance computing simulations of physics related problems do exist, these tend not to be real-time and do not model the real-time intricate interactions of rigid bodies for visual effect common in video games (favouring accuracy over real-time). As such, our work presents the first real-time delivery of scalable, video game quality physics to players. We do this by taking the physics engine out of the player's machine and deploying it across multiple cloud instances. We built a prototype solution to allow the physics demands of players to be satisfied remotely without hindering the player's real-time interaction. Gaming environments are commonly presented to multiple geographically distant players using server side scalable technologies. This can allow scalable gaming platforms that present players with large online worlds to explore while maintaining the visually highly realistic environments generated by the player's local machine. Research into delivering scalability in such online worlds is always focussed on balancing real-time and consistency issues for enabling player-player interaction. Player interaction can be measured in milliseconds, which current networking technology cannot model for all gaming genres in a scalable manner without giving rise to inconsistency. Therefore, the notion of trying to model real-time physics, which requires solvers working in iterations measured in microseconds, would be considered a near impossible task. However, achieving this would allow the physics engine element of the gaming console to be transplanted to the server side (cloud), freeing up resources for gameplay and rendering. In the context of current research we can place ourselves at the juncture of balancing physics calculations across nodes in the cloud for game streaming services. This should provide improved economic use of resources as current game streaming services simply mimic the machine requirements of the player without distributing any of the component parts of the gaming engine. In addition, streamed games may benefit from increased numbers of artefacts (way beyond currently possible in a player's machine) as other nodes may be utilised to solve the physics problems. We demonstrate the validity of our results through experimentation and benchmarking.