Solving sparse linear system

I want to solve a linear system Ax=b, where A is about 16777218x16777218 matrix, it is sparse but not symmetric positive definite. Normally I use the \ from Matlab to solve the system. However, I'm wondering if there is a more efficient way or tricks to solve this system. I don't have problems with storage, it is only the time that I care about. The iterative solves from Matlab don't give me good results since this system is part of Newton iteration, i.e, it is already an approximation of a nonlinear system.

[John D'Errico]

If there were magic tricks available, then they would be built into \ already. Big problems take time to solve. You have already stated you are not interested in iterative solvers.

[ahmad]

actually what I meant by tricks is that, if it would be efficient to call subroutines written in C++ for instance from Matlab.