Gradient Descent apprach

When you run simulations i notice that you use a brute force approach to find which is the best. Would there be a way to use a gradient approach that would allow the simulations to be 100% faster?

I’m not sure I follow exactly. What do you mean by “find which is best”? The simulations run multiple iterations to get an average result. The optimizer uses an algorithm that we have developed over the years to search through gear setups and find one that scores highest or nearly highest - we certainly do not use brute force as that would take prohibitively long.