How to make the function mvncdf() work faster?

I need to compute P(X<x) for more than 10^5 bivariate normally distributed variables. The function mvncdf() seems to be a good way of doing this. It works very fast as long as all variables have the same covariance matrix. However, the variables I am working with all have the same variance (=1) but different correlations. I can't think of any other solution to this than to use a for loop, but this makes the computation much slower. Anyone who knows how to speed up the computation?

Thanks!