I have several libraries that I am linking against and using them in a project. I can compile the libraries as well. If I have, in the final project, Code Generation > Floating Point Model = Fast, and Precise in the libraries that I am linking against, does that mean I am not benefiting from the maximum speedup that I will get if the libraries were compiled with the Floating Point Model = Fast as well? I am getting confused because I am thinking that it is up to the final binary project is using all the libraries and decide which model to use.
Info on the project: I am performing heavy calculations on large grids (marching cubes) with spatial graphs for fast calculations. On larger grids, it can still take quite a few seconds. There is (roughly - I did very few tests) a 20-30% speed increase in calculations and the end result is the same so far. The project is linking against release libraries with Floating Point Model set to Precise.