We're playing a guessing game...no?

If we could only have a list of how long it takes to execute an OpenGL function so we can then use it to create optimizations where they’re needed most.

Wouldn’t that be cool to have a list like that?

Well, uh, maybe it would. But it wouldn’t be constant across different pieces of hardware, different drivers or even different parameters passed to the same function. It all depends on context, which is far too big a state space to enumerate.

Plus, profiling code isn’t too tricky to get at the vague problem areas.