Lattice Quantum Dynamics can be used to investigate quark structure in nuclei or electron transport in graphene, but useful calculations require the solution of matrices of sizes running into the 100's of millions. Solvers that scale linearly with matrix size and perform well over large cluster parallelization or GPU accelerators are critical. The performance of the solvers for low mass values, leading poorly conditioned matrices, is also an important consideration to support the investigation of interesting physical properties. This paper presents results for BiCGStab and CGNE with an Algebraic Multigrid preconditioner that achieves the desired linear time complexity over the range of possible mass parameters.