Hmmm, well flops are correct when we are talking about graphics (floating point operations per second.) I'd venture to guess it is a little more complicated than one number depending on the core, vertex shader, pixel pipeline, raster, or if the card has stream processors which can handle more than one task.
You'd have to also assume they are only generating graphical data and there is not any interaction with the environment or gameplay. When you add gameplay/physics then you need to simultaneously run "gametime" which controls input, physics engine refresh, database pulls, and such. Then there is the database that is constantly being updated with player information. AI needs to run, etc etc.
For ONLY the graphics portion, perhaps 2.7 teraflops is good, but when we are talking about a full game we need much more than just that.
Reply With Quote
