Accurate prediction of the sawtooth cycle [1] is an important test for nonlinear MHD codes. The sawtooth cycle in the CDX-U tokamak [2], chosen because its small size and low temperature allow simulation using actual device parameters, has been an important benchmark for the comparison of the M3D [3] and NIMROD [5] codes for the last several years. Successive comparisons have led to improvements and refinements in both codes. The most recent comparisons show impressive agreement between the two codes both on the linear instability and on the details of nonlinear cyclical behavior. These tests are somewhat idealized and do not yet agree quantitatively with the experimentally observed sawtooth period. We expect a second generation of CDX-U sawtooth benchmarks based on an analytically specified equilibrium, with source terms that show greater fidelity to the physical device, to produce better agreement.