Detecting Source Code Hotspot in Games Software Using Call Flow Analysis

In collaborative development of games software, hotspot identification technique significantly supports maintenance and evolutions activities including modding because newly involved developers must identify hotspot. This paper focuses on game loops as hotspots because many action games, including classic action games and the latest action games for console game machines, have an architecture consisting of game loops and update functions. This study assumes that game loops and update functions have the largest number of function calls in the entire source code modules. This paper investigates whether call flow analysis helps developers identify hotspots in four games software publicly accessible in GitHub. The results of the investigation revealed that one game loop and a set of update functions could be identified precisely by call flow analysis and that other three neighborhood functions could be identified. The results showed the possibility of automated identification of game loop and update functions by call flow analysis.