2016 Participants

Andrew Warburton

  • MComp (Hons) Computing Science (with Industrial Placement)
  • Game AI- Done differently

Video games can often run slowly when many agents (enemies for example) are present, or when an agent with complex AI is present. My research was to program and test a method to make AI processing less taxing on the rest of the game. Processing, consisting of things such as AI and physics, is normally done on the Central Processing Unit (CPU) - a device within the computer. There is another device which is usually completely dormant in PCs with graphics cards – the CPU’s Graphics Coprocessor.

I successfully made the graphics coprocessor process all of the AI for a game, which freed up the CPU to do more without the game slowing. This essentially added extra processing power to the computer and can be done on nearly every desktop PC with a graphics card. This also has implications to improve AI performance with large numbers of agents.

Funding source: Newcastle University

Supervisor: Dr William Blewitt