I have 3 Question :
1)
I just want to know if in this code, we can considerate that there is a lot of polygons or only one.
glBegin(GL_QUADS);
glTexCoord2f(0,0); glVertex3f(-500,-500,-500.0f);
…
I don’t think but…i try to imagine how do they succeed to have those 25 millions…
Next question, when i set a light to a scene, it improve the performance !!
I think there should be a way to improve this performance without lighting.
And last one…can we incorporate / import any object in OpenGL? A car for example.
Second. Improve performance without enabling lighting? Sure, there’s lots of ways. Use vertex arrays instead of immediate mode for example. Or what do you mean?
Third. OpenGL does now deal with high level terms like “objects”. OpenGL only knows about triangles, lines and points, and it only know about ONE triangle/line/point at a time. And IF it where to know about objects, why on earth would it prevent you from drawing certain objects? It’s up to you to load and draw objects.
per the first question: the triangles/sec measurement that vendors list for their cards are very inacurate. they are going to do anything they can to make that number as high as possible. they probably dont use texturing, or lighting, they probably draw things as triangle strips, use vertex arrays, display list, and run at very low resolutions (maybe even absurdly low resolutions, like 1x1). even if they didnt perform all the optimizations, they are certainly not going to take into account things that would be in your average game/app. stuff like physics, ai, etc. these may even be theoretical, based on clock speeds, bus width, etc, and they are just calculating how many the card could theoretically handle, without actually making a program that draws them all.
of course, you can use some of these tricks, and you probably should. whenever you can, use triangle strips, vertex arrays, display lists, etc to speed up your program. also, drawing using GL_QUADS is not going to give you an acurate triangles per second, since quads are not triangles…
there was a recent thread in the advanced forum about triangles per second that you might want to check out. also, questions about optimizations come up a lot, so you might want to look back at previous threads…
Actually, on the first question, many vendors will generate the polygon per second by benchmarking with glCullFace(FL_FRONT_AND_BACK) on, so no actual pixels are being rendered.