Why cgs does not converge

I have a matrix equation. The dimension of my matrix is 16,000. I called `cgs(A,b)` to solve it, but it said the it did not converge. I myself have written a conjugate gradient iterative method for any general matrices (symmetric or anti symmetric ). The algorithm is quite simple. I used it to solve my matrix. It converges well. How can this happen?

This question may seem strange. But for me, I think the build in cgs must have made some modifications to the original conjugate gradient method.