Impact of the number of texture coordinates on frame rate I wrote a benchmark which renders display lists of vertex arrays, with GL_FLOAT normals, GL_UNSIGNED_BYTE colors, and a variable number of 2D GL_FLOAT texture coordinates per vertex. GeForce 6600 GT Here is a graph showing frame rate according to the number of texture coordinates per vertex when rendering scenes of 1.0M, 1.5M, and 2.0M triangles, on a GeForce 6600 GT under Linux: Radeon X800 The same experiment with display lists on a Radeon X800 under Windows: