The release of a new game, which requires the high performance of hardware components is always of great interest to us for potential integration to our test protocol. This is the case of Supreme Commander, Chris Taylor’s latest RTS. Because of its results in tests we decided to write this article.
The game Supreme Commander is a real time strategy game, which functions on another scale than most current games, as Total Annihilation did a few years ago,.
The game features a zoom function called "Strategic", which gives the user the possibility to choose a different height other than that of the battlefield level. The implementation of such a function compared to standard views has of course some consequences on performances.
The other important characteristic of Supreme Commander is, of course, like many RTS the great number of calculations dependant on the CPU, whether they are AI, unit movements, management of projectiles, etc. In order to take advantage of the potential of the latest CPUs, engineers of GPG have made Supreme Commander a multithread game. This means that it is capable of using processors with 1, 2 and even 4 cores. To do so, they have chosen a "relatively" simple support of threading in that each major part of the game is a thread. There is one thread for the graphic part, another for the simulation, and less heavy ones for sound amongst other things. GPG hasn't answered our questions concerning AI, mainly if it’s included to the simulation thread or if it is on another thread of its own.
What is the impact of this implementation on performances? Is the processor really dominant? This is what we will see in this article. An interesting point is that Supcom isn't Quake-like. The view type combined with lateral scrolling mostly replaced by the zoom means that the framerate required to enjoy the game doesn't have to be much higher than 25 fps and even at 15 fps it is still possible to play comfortably. Between 10 to 15 fps, it’s a bit bothersome and below 10 fps it’s dismal.