(in connection with the previous thread) how can the precision of the fixed point arithmetic be adjusted? Seems that there are more changes needed than MACRO EGL_PRECISION. Or is it not intended to do so?
That's correct. It's a constant, but too many other details depend on this number to make it actually ajustable.
Probably the best thing to do in your case would be to convert the computations up to the perspective projection to proper floating point (and using the Intel compiler, because it has a very good FP library).
In the case of maps, you are dealing with large sets of geometry anyway, so the performance decrease would not matter to you.
PS: This is for a commercial product, right? Or are you repackaging google maps in your spare time? ;-)
Log in to post a comment.