Progress towards separating physics and dynamics onto separate computational cores will be presented.  With an eye towards the future in high performance computing, the climate modeling community has been focusing on how to improve performance on a growing number of computational cores.  The two most costly components of the atmosphere model, dynamics and physics, are both perfectly scalable up to the total number of elements and physics columns, respectively.  However, the current coupling framework of sequential tendency splitting, by which the tendencies from physics are passed to the SE-dycore to advance the model state lead to a bottleneck in overall performance at the limit of the number of elements.  With 9 times more physics columns than dynamics elements, a vast potential for performance enhancement is left untapped.  Our hypothesis is that by switching to a parallel split architecture between physics and dynamics and solving each component on separate cores it is possible to extend the scalability of the model past the current limit.  Early progress in applying a parallel-split architecture on separate cores will be presented as well as some promising results.